speech on artificial intelligence a threat to humanity

We need your support today

Independent journalism is more important than ever. Vox is here to explain this unprecedented election cycle and help you understand the larger stakes. We will break down where the candidates stand on major issues, from economic policy to immigration, foreign policy, criminal justice, and abortion. We’ll answer your biggest questions, and we’ll explain what matters — and why. This timely and essential task, however, is expensive to produce.

We rely on readers like you to fund our journalism. Will you support our work and become a Vox Member today?

• Future Perfect

The case that AI threatens humanity, explained in 500 words

The short version of a big conversation about the dangers of emerging technology.

by Kelsey Piper

Tech superstars like Elon Musk , AI pioneers like Alan Turing , top computer scientists like Stuart Russell , and emerging-technologies researchers like Nick Bostrom have all said they think artificial intelligence will transform the world — and maybe annihilate it.

So: Should we be worried?

Here’s the argument for why we should: We’ve taught computers to multiply numbers, play chess , identify objects in a picture, transcribe human voices, and translate documents (though for the latter two, AI still is not as capable as an experienced human). All of these are examples of “narrow AI” — computer systems that are trained to perform at a human or superhuman level in one specific task.

We don’t yet have “general AI” — computer systems that can perform at a human or superhuman level across lots of different tasks.

Most experts think that general AI is possible, though they disagree on when we’ll get there . Computers today still don’t have as much power for computation as the human brain, and we haven’t yet explored all the possible techniques for training them. We continually discover ways we can extend our existing approaches to let computers do new, exciting, increasingly general things, like winning at open-ended war strategy games .

But even if general AI is a long way off, there’s a case that we should start preparing for it already. Current AI systems frequently exhibit unintended behavior. We’ve seen AIs that find shortcuts or even cheat rather than learn to play a game fairly, figure out ways to alter their score rather than earning points through play, and otherwise take steps we don’t expect — all to meet the goal their creators set.

As AI systems get more powerful, unintended behavior may become less charming and more dangerous. Experts have argued that powerful AI systems, whatever goals we give them, are likely to have certain predictable behavior patterns . They’ll try to accumulate more resources, which will help them achieve any goal. They’ll try to discourage us from shutting them off, since that’d make it impossible to achieve their goals. And they’ll try to keep their goals stable, which means it will be hard to edit or “tweak” them once they’re running. Even systems that don’t exhibit unintended behavior now are likely to do so when they have more resources available.

For all those reasons, many researchers have said AI is similar to launching a rocket . (Musk, with more of a flair for the dramatic, said it’s like summoning a demon .) The core idea is that once we have a general AI, we’ll have few options to steer it — so all the steering work needs to be done before the AI even exists, and it’s worth starting on today.

The skeptical perspective here is that general AI might be so distant that our work today won’t be applicable — but even the most forceful skeptics tend to agree that it’s worthwhile for some research to start early, so that when it’s needed, the groundwork is there.

• The case for taking AI seriously as a threat to humanity

Sign up for the Future Perfect newsletter. Twice a week, you’ll get a roundup of ideas and solutions for tackling our biggest challenges: improving public health, decreasing human and animal suffering, easing catastrophic risks, and — to put it simply — getting better at doing good.

More in this stream

Most Popular

• Has The Bachelorette finally gone too far?
• What the polls show about Harris’s chances against Trump
• The real reason Netanyahu won’t end the Gaza war
• Trump’s biggest fans aren’t who you think
• Why Nvidia triggered a stock market freakout

Today, Explained

Understand the world with a daily explainer plus the most compelling stories of the day.

This is the title for the native ad

More in Future Perfect

Too much fluoride might lower IQ in kids, a new federal report says. The science (and debate), explained.

EVs help reduce greenhouse emissions. But too many used gas-guzzlers could make that impossible.

The US military’s recruiting crisis, explained.

Gavin Newsom could decide the future of AI safety. But will he cave to billionaire pressure?

Organ companies are getting up to pancreas hijinks.

Are tech companies actually pushing AI down our throats?

• Skip to main content
• Keyboard shortcuts for audio player

The U.N. Warns That AI Can Pose A Threat To Human Rights

Scott Neuman

The United Nations High Commissioner for Human Rights Michelle Bachelet speaks at a climate event in Madrid in 2019. A recent report of hers warns of the threats that AI can pose to human rights. Ricardo Rubio/Europa Press Via Getty Images hide caption

The United Nations High Commissioner for Human Rights Michelle Bachelet speaks at a climate event in Madrid in 2019. A recent report of hers warns of the threats that AI can pose to human rights.

The United Nations' human rights chief has called on member states to put a moratorium on the sale and use of artificial intelligence systems until the "negative, even catastrophic" risks they pose can be addressed.

The remarks by U.N. High Commissioner for Human Rights Michelle Bachelet were in reference to a new report on the subject released in Geneva .

The report warned of AI's use as a forecasting and profiling tool, saying the technology could have an impact on "rights to privacy, to a fair trial, to freedom from arbitrary arrest and detention and the right to life."

Investigations

China's microsoft hack may have had a bigger purpose than just spying.

The report, and Bachelet's comments, follow the recent revelation that widespread use was being made of spyware, known as Pegasus , to target thousands of phone numbers and dozens of devices belonging to international journalists, human rights activists and heads of state.

Bachelet acknowledged that AI "can be a force for good, helping societies overcome some of the great challenges of our times," but suggested that the harms it could bring outweigh the positives. But she also warned of an "unprecedented level of surveillance across the globe by state and private actors," that she said is "incompatible" with human rights.

"The higher the risk for human rights, the stricter the legal requirements for the use of AI technology should be," she said.

Facebook Apologizes After Its AI Labels Black Men As 'Primates'

Upon the release of the report, Tim Engelhardt, UNHRC's human rights officer, rule of law and democracy section, called the situation regarding AI "dire" and said it has "not improved over the years but has become worse."

The document includes an assessment of profiling, automated decision-making and other machine-learning technologies.

This story originally published in the Morning Edition live blog .

Mobile Menu Overlay

The White House 1600 Pennsylvania Ave NW Washington, DC 20500

Remarks by Vice President Harris on the Future of Artificial Intelligence | London, United   Kingdom

U.S. Embassy London, United Kingdom

1:43 P.M. GMT THE VICE PRESIDENT:  Hello, everyone.  Good afternoon.  Good afternoon, everyone.  (Applause.)  Please have a seat.  Good afternoon.  It’s good to see everyone. Ambassador Hartley, thank you for the warm welcome that you gave us last night and today, and for inviting us to be here with you.  And thank you for your extraordinary leadership, on behalf of the President and me and our country. And it is, of course, my honor to be with everyone here at the United States Embassy in London, as well as to be with former Prime Minister Theresa May and all of the leaders from the private sector, civil society, academia, and our many international partners. 

So, tomorrow, I will participate in Prime Minister Rishi Sunak’s Global Summit on AI Safety to continue to advance global collaboration on the safe and responsible use of AI. Today, I will speak more broadly about the vision and the principles that guide America’s work on AI. President Biden and I believe that all leaders from government, civil society, and the private sector have a moral, ethical, and societal duty to make sure that AI is adopted and advanced in a way that protects the public from potential harm and that ensures that everyone is able to enjoy its benefits. AI has the potential to do profound good to develop powerful new medicines to treat and even cure the diseases that have for generations plagued humanity, to dramatically improve agricultural production to help address global food insecurity, and to save countless lives in the fight against the climate crisis. But just as AI has the potential to do profound good, it also has the potential to cause profound harm.  From AI-enabled cyberattacks at a scale beyond anything we have seen before to AI-formulated bio-weapons that could endanger the lives of millions, these threats are often referred to as the “existential threats of AI” because, of course, they could endanger the very existence of humanity. (Pause) These threats, without question, are profound, and they demand global action. But let us be clear.  There are additional threats that also demand our action — threats that are currently causing harm and which, to many people, also feel existential. Consider, for example: When a senior is kicked off his healthcare plan because of a faulty AI algorithm, is that not existential for him? When a woman is threatened by an abusive partner with explicit, deep-fake photographs, is that not existential for her? When a young father is wrongfully imprisoned because of biased AI facial recognition, is that not existential for his family? And when people around the world cannot discern fact from fiction because of a flood of AI-enabled mis- and disinformation, I ask, is that not existential for democracy? Accordingly, to define AI safety, I offer that we must consider and address the full spectrum of AI risk — threats to humanity as a whole, as well as threats to individuals, communities, to our institutions, and to our most vulnerable populations. We must manage all these dangers to make sure that AI is truly safe. So, many of you here know, my mother was a scientist.  And she worked at one of our nation’s many publicly funded research universities, which have long served as laboratories of invention, creativity, and progress. My mother had two goals in her life: to raise her two daughters and end breast cancer.  At a ver- — very early age then, I learned from her about the power of innovation to save lives, to uplift communities, and move humanity forward. I believe history will show that this was the moment when we had the opportunity to lay the groundwork for the future of AI.  And the urgency of this moment must then compel us to create a collective vision of what this future must be. A future where AI is used to advance human rights and human dignity, where privacy is protected and people have equal access to opportunity, where we make our democracies stronger and our world safer.  A future where AI is used to advance the public interest. And that is the future President Joe Biden and I are building. Before generative AI captured global attention, President Biden and I convened leaders from across our country — from computer scientists, to civil rights activists, to business leaders, and legal scholars — all to help make sure that the benefits of AI are shared equitably and to address predictable threats, including deep fakes, data privacy violations, and algorithmic discrimination.  And then, we created the AI Bill of Rights.  Building on that earlier this week, President Biden directed the United States government to promote safe, secure, and trustworthy AI — a directive that will have wide-ranging impact.  For example, our administration will establish a national safety reporting program on the unsafe use of AI in hospitals and medical facilities.  Tech companies will create new tools to help consumers discern if audio and visual content is AI-generated.  And AI developers will be required to submit the results of AI safety testing to the United States government for review.  In addition, I am proud to announce that President Biden and I have established the United States AI Safety Institute, which will create rigorous standards to test the safety of AI models for public use.   Today, we are also taking steps to establish requirements that when the United States government uses AI, it advances the public interest.  And we intend that these domestic AI policies will serve as a model for global policy, understanding that AI developed in one nation can impact the lives and livelihoods of billions of people around the world.  Fundamentally, it is our belief that technology with global impact deserves global action.  And so, to provide order and stability in the midst of global technological change, I firmly believe that we must be guided by a common set of understandings among nations.  And that is why the United States will continue to work with our allies and partners to apply existing international rules and norms to AI and work to create new rules and norms.  To that end, earlier this year, the United States announced a set of principles for responsible development, deployment, and use of military AI and autonomous capabilities.  It includes a rigorous legal review process for AI decision-making and a commitment that AI systems always operate with international — and within international humanitarian law.  Today, I am also announcing that 30 countries have joined our commitment to the responsible use of military AI.  And I call on more nations to join.  In addition to all of this, the United States will continue to work with the G7; the United Nations; and a diverse range of governments, from the Global North to the Global South, to promote AI safety and equity around the world.  But let us agree, governments alone cannot address these challenges.  Civil society groups and the private sector also have an important role to play. 

Civil society groups advocate for the public interest.  They hold the public and private sectors to account and are essential to the health and stability of our democracies. 

As with many other important issues, AI policy requires the leadership and partnership of civil society.  And today, in response to my call, I am proud to announce that 10 top philanthropies have committed to join us to protect workers’ rights, advanced transparency, prevent discrimination, drive innovation in the public interest, and help build international rules and norms for the responsible use of AI. 

These organizations have already made an initial commitment of $200 million in furtherance of these principles. 

And so, today, I call on more civil society organizations to join us in this effort. 

In addition to our work with civil society, President Biden and I will continue to engage with the private companies who are building this technology. 

Today, commercial interests are leading the way in the development and application of large language models and making decisions about how these models are built, trained, tested, and secured. 

These decisions have the potential to impact all of society. 

As such, President Biden and I have had extensive engagement with the leading AI companies to establish a minimum — minimum — baseline of responsible AI practices. 

The result is a set of voluntary company commitments, which range from commitments to report vulnerabilities discovered in AI models to keeping those models secure from bad actors. 

Let me be clear, these voluntary commitments are an initial step toward a safer AI future with more to come, because, as history has shown, in the absence of regulation and strong government oversight, some technology companies choose to prioritize profit over the wellbeing of their customers, the safety of our communities, and the stability of our democracies. 

An important way to address these challenges, in addition to the work we have already done, is through legislation — legislation that strengthens AI safety without stifling innovation. 

In a constitutional government like the United States, the executive branch and the legislative branch should work together to pass laws that advance the public interest.  And we must do so swiftly, as this technology rapidly advances. 

President Biden and I are committed to working with our partners in Congress to codify future meaningful AI and privacy protections. 

And I will also note, even now, ahead of congressional action, there are many existing laws and regulations that reflect our nation’s longstanding commitment to the principles of privacy, transparency, accountability, and consumer protection. 

These laws and regulations are enforceable and currently apply to AI companies. 

President Biden and I reject the false choice that suggests we can either protect the public or advance innovation.  We can and we must do both. 

The actions we take today will lay the groundwork for how AI will be used in the years to come. So, I will end with this: This is a moment of profound opportunity.  The benefits of AI are immense.  It could give us the power to fight the climate crisis, make medical and scientific breakthroughs, explore our universe, and improve everyday life for people around the world.  So, let us seize this moment.  Let us recognize this moment we are in. As leaders from government, civil society, and the private sector, let us work together to build a future where AI creates opportunity, advances equity, fundamental freedoms and rights being protected. Let us work together to fulfill our duty to make sure artificial intelligence is in the service of the public interest. I thank you all.  (Applause.)                         END                1:59 P.M. GMT

Stay Connected

We'll be in touch with the latest information on how President Biden and his administration are working for the American people, as well as ways you can get involved and help our country build back better.

Opt in to send and receive text messages from President Biden.

Navigating Humanity’s Greatest Challenge Yet: Experts Debate the Existential Risks of AI

In recent years, the rapid proliferation of artificial intelligence (AI) has emerged as a beacon of innovation, promising to reshape the world with unparalleled efficiency and knowledge. 

Yet, beneath the surface of these technological advancements, a myriad of questions and concerns lurk, casting shadows over AI’s glowing promise. 

Scientists, experts, and the general public are beginning to question the trajectory of AI technology and its implications for the future of humanity. At the heart of the debate is whether AI represents an existential threat to humanity.

A recent  event  hosted by the American nonprofit global policy think tank, the RAND Corporation, brought together a diverse panel of five experts to delve into the existential risks posed by AI. 

Experts were divided on what they considered to be the most significant threats AI poses to humanity’s future, indicating that AI security is a complex and nuanced issue.

“The risk I’m concerned about isn’t a sudden, immediate event,” Benjamin Boudreaux, a policy researcher who studies the intersection of ethics, emerging technology, and security, said. “It’s a set of incremental harms that worsen over time. Like climate change, AI might be a slow-moving catastrophe, one that diminishes the institutions and agency we need to live meaningful lives.” 

Dr. Jonathan Welburn, a RAND senior researcher and a professor of policy analysis, noted that advancements in AI draw similar parallels to past periods of technological upheaval. 

However, unlike the advent of electricity, the printing press, or the internet, Dr. Welburn said his most significant concern with AI lies in its potential to amplify existing societal inequities and introduce new forms of bias, potentially undermining social and economic mobility through ingrained racial and gender prejudices. 

“The world in 2023 already had high levels of inequality,” Dr. Welburn said. “And so, building from that foundation, where there’s already a high level of concentration of wealth and power—that’s where the potential worst-case scenario is for me.” 

Dr. Jeff Alstott, the RAND Center for Technology and Security Policy Director and a Senior Information Scientist, painted a particularly sobering picture of future challenges. He shared his most profound concern, noting that the prospect of AI being weaponized by bad actors “keeps me up at night.”

“Bioweapons [happen] to be one example where, historically, the barriers have been information and knowledge. You don’t need much in the way of specialized matériel or expensive sets of equipment any longer in order to achieve devastating effects with the launching of pandemics,” Dr. Alstott explained. “AI could close the knowledge gap. And bio is just one example. The same story repeats with AI and chemical weapons, nuclear weapons, cyber weapons.” 

During the panel discussion, the experts’ primary concern wasn’t the technology itself. Instead, their worries centered on the potential for humans to misuse AI for harmful purposes.

“To me, AI is gas on the fire,” Dr. Nidhi Kalra, a senior Information Scientist at RAND, explained. I’m less concerned with the risk of AI than with the fires themselves—the literal fires of climate change and potential nuclear war and the figurative fires of rising income inequality and racial animus.” 

From AI-induced mistrust and undermining democracy, RAND policy researcher Dr. Edward Geis expressed his concerns about AI more directly, stating, “AI threatens to be an amplifier for human stupidity.” 

Following a comprehensive analysis of recent scientific studies, Dr. Roman V. Yampolskiy, an AI safety expert and associate professor at the University of Louisville, identified an additional existential threat posed by AI. According to Dr. Yampolskiy, there is  no evidence that AI superintelligence can be safely controlled , cautioning, “Without proof that AI can be controlled, it should not be developed.” 

“We are facing an almost guaranteed event with potential to cause an existential catastrophe,” Dr. Yampolskiy warned. “No wonder many consider this to be the most important problem humanity has ever faced. The outcome could be prosperity or extinction, and the fate of the universe hangs in the balance.”

In a  recent paper , Dr. Atoosa Kasirzadeh, an assistant professor at the University of Edinburgh who focuses on the ethics, safety, and philosophy of AI, further explored the existential risks posed by AI. 

According to Dr. Kasirzadeh, the conventional discourse on existential risks posed by AI typically focuses on “decisive” threats or abrupt, dire events caused by advanced AI systems that “lead to human extinction or irreversibly cripple human civilization to a point beyond recovery.” 

Dr. Kasirzadeh explained that AI development also carries “accumulative” risks, likened to a “boiling frog scenario.” In this scenario, gradual AI-related risks build up over time, gradually weakening resilience. This process continues until a critical event occurs, leading to an irreversible collapse.

Echoing Dr. Boudreaux’s sentiments, Dr. Kasirzadeh concluded her paper by saying, “There is no inherent reason to consider that the accumulative hypothesis is any less likely than the decisive view. The need to further substantiate the accumulative hypothesis is apparent.”

Dr. Yampolskiy and Dr. Kasirzadeh did not participate in the recent RAND panel discussion on AI existential risks. However, their latest research findings introduce additional complexity to the ongoing debate.

The experts at RAND had differing opinions on whether AI poses a direct existential threat to humanity’s future.

Dr. Welburn and Dr. Kalra both believed that AI does not currently represent an irreversible threat, pointing out that humanity has a long history of overcoming significant challenges.

NASA Will Debut Its Mysterious X-59 ‘Quiet’ Supersonic Aircraft at Famous Lockheed Martin Skunk Works Facility Next Week

“We are an incredibly resilient species, looking back over millions of years,” Dr. Kalra said. “I think that’s not to be taken lightly.” 

Conversely, Dr. Boudreaux and Dr. Alstott felt that AI did pose a threat to humanity’s future, noting that the extinction of the human race is not the only catastrophic impact AI can have on societies. 

“One way that could happen is that humans die,” Dr. Boudreaux explained. “But the other way that can happen is if we no longer engage in meaningful human activity, if we no longer have embodied experience, if we’re no longer connected to our fellow humans. That, I think, is the existential risk of AI.” 

Dr. Geist said he was uncertain of just how significant AI’s risks are, likening its advancement to the development of nuclear weapons. 

“The current scientific understanding is that the nuclear weapons that exist today could probably not be used in a way that would result in human extinction,” Dr. Geist pointed out. “That scenario would require more and bigger bombs. That said, a nuclear arsenal that could plausibly cause human extinction via fallout and other effects does appear to be something a country could build if it really wanted to.” 

Panelists expressed that the path forward is fraught with uncertainty. However, this does not inherently mean that AI will doom us to extinction.

All five experts unanimously agreed that independent, high-quality research will play a crucial role in assessing AI’s short—and long-term risks and shaping public policy accordingly.

Addressing AI’s existential risks will require a multifaceted approach, emphasizing transparency, oversight, and inclusive policymaking. As the experts suggested, ensuring AI’s integration into our lives enhances rather than diminishes our humanity is paramount. 

Experts underscored that this must involve rigorous research and policy interventions and foster communities resilient to the broad spectrum of crises we could face in a future AI-filled world. 

“Researchers have a special responsibility to look at the harms and the risks. This isn’t just looking at the technology itself but also at the context—looking into how AI is being integrated into criminal justice and education and employment systems, so we can see the interaction between AI and human well-being,” Dr. Boudreaux said. “But I don’t think there’s a technical fix alone. We need to have a much broader view of how we build resilient communities that can deal with societal challenges.” 

Tim McMillan is a retired law enforcement executive, investigative reporter and co-founder of The Debrief. His writing typically focuses on defense, national security, the Intelligence Community and topics related to psychology. You can follow Tim on Twitter:   @LtTimMcMillan.   Tim can be reached by email:  [email protected]  or through encrypted email:   [email protected]  

• AI Education in India
• Speakers & Mentors
• AI services

Artificial intelligence – a potential threat or a tool for humanity? The debate continues

In the surrounding world of technology, there is an ongoing conversation and debate about the potential risks posed by artificial intelligence. The discussion revolves around the intelligence that AI possesses and whether it is a danger to humanity. This ongoing debate highlights the concerns and fears surrounding the development of AI and its impact on society.

Artificial intelligence, or AI, is a rapidly growing field that has the potential to revolutionize various industries and aspects of our lives. However, there are serious concerns about how this technology could affect humanity. Some argue that AI could outperform humans in many areas, leading to a loss of jobs and economic instability. Others worry about the possibility of AI becoming too powerful and potentially surpassing human intelligence, posing a threat to our very existence.

The risks associated with AI are not imaginary; they exist in various forms. As AI continues to advance, there is a concern that it could be weaponized and used to carry out destructive actions. Additionally, there are fears that AI could become autonomous and make decisions that are detrimental to human well-being. These risks have led experts to call for strict regulations and ethical guidelines to ensure that AI is developed and used responsibly.

The ongoing conversation on the risks posed by artificial intelligence to humanity.

The potential dangers of artificial intelligence and its impact on humanity have long been subjects of debate. With the rapid advancements in AI technology, concerns surrounding its risks have intensified in recent years.

The ongoing debate centers around the degree to which AI poses a threat to humanity. Some argue that AI has the potential to surpass human intelligence and autonomy, leading to a dystopian future where machines take over and control human society. Others believe that the current level of AI development is not advanced enough to constitute a significant danger.

One of the primary concerns surrounding AI is its ability to make autonomous decisions, especially in critical areas such as warfare or healthcare. AI systems have the potential to make errors or act in ways that are not aligned with human values and ethics, posing risks to individuals and society as a whole.

Another point of contention is the potential impact of AI on the job market and economy. As AI technology continues to advance, there are concerns that automation will lead to widespread job losses and income inequality. This raises questions about the consequences for human well-being and societal stability.

Furthermore, there is ongoing conversation about the ethical implications of AI. Questions about data privacy, algorithmic bias, and the responsibility of developers and organizations using AI technologies are at the forefront of discussions. The potential for misuse of AI by malicious actors or oppressive governments is also a significant concern.

In conclusion, the ongoing conversation about the risks posed by artificial intelligence to humanity is multifaceted and complex. While there is no consensus on the level of threat AI poses, it is crucial to continue discussing and addressing these concerns to ensure the responsible development and deployment of AI technologies.

The debate about artificial intelligence as a potential danger to humanity.

The ongoing discussion on the topic of artificial intelligence (AI) revolves around the potential risks it may pose to humanity. The conversation surrounding this debate has been fueled by concerns about the potential dangers that AI could bring to our society.

Artificial intelligence has made significant advancements in recent years, with AI systems becoming more sophisticated and capable of performing complex tasks. While this has led to numerous benefits in various fields, such as healthcare and transportation, there is a growing concern that AI could also pose a threat to humanity.

One of the main concerns is the possibility of AI surpassing human intelligence and becoming uncontrollable. This scenario, often referred to as “the singularity,” imagines a future where AI systems have surpassed human capabilities and become self-improving. If this were to happen, it could lead to a range of potential risks, including loss of control over AI systems and unintended consequences.

Another concern is the ethical implications of advanced AI. As AI systems become more autonomous, questions arise about the role and responsibility of AI in decision-making processes. There is a fear that AI systems may make decisions that go against human values, leading to harmful outcomes or discriminatory practices.

Furthermore, there are worries about the potential misuse of AI technology by malicious actors. AI-powered systems can be weaponized or used for surveillance, raising concerns about privacy and security on a global scale. The potential for AI to amplify existing social inequalities and widen the digital divide is also a point of contention in this ongoing debate.

It is important to note that the debate about AI as a threat to humanity is not a consensus. There are varying perspectives on the level of risk posed by AI, with some arguing that the benefits outweigh the potential dangers. However, the conversation continues as researchers, policymakers, and experts grapple with the complex challenges and potential consequences of AI technology.

In conclusion, the debate surrounding artificial intelligence as a potential danger to humanity is a critical discussion that raises important questions about the future of technology and its impact on society. It is crucial to carefully consider the risks and benefits associated with AI to ensure that we develop and deploy this technology responsibly and ethically.

The discussion surrounding the threat of artificial intelligence to humanity.

The ongoing conversation about the potential threat posed by artificial intelligence to humanity is a topic of significant debate. The discussion revolves around the danger that AI could present to humanity as its capabilities continue to advance.

There are varying opinions surrounding this topic. Some argue that AI has the potential to enhance human society and improve our lives in numerous ways. They emphasize the positive aspects of AI, such as its ability to automate processes, analyze vast amounts of data, and develop new technologies. They believe that AI can be controlled and utilized for the benefit of humanity.

On the other hand, there are those who express concerns about the potential dangers of AI. They argue that as AI becomes more intelligent and autonomous, it may surpass human intelligence and pose a threat to our existence. These concerns range from the loss of jobs and economic inequality to the more existential fear of AI becoming uncontrollable and turning against humanity.

The debate over AI

The debate surrounding the threat of artificial intelligence to humanity is multi-faceted. It encompasses not only the potential dangers of AI but also the ethical implications of its development and use. Should we allow AI systems to make life-or-death decisions? How can we ensure that AI remains aligned with human values? These questions highlight the complex nature of the debate.

The future of humanity and AI

The discussion about the threat of artificial intelligence to humanity is an important one. It prompts us to consider the implications of advancing technology and the impact it may have on our society. While the potential dangers of AI should not be ignored, it is crucial to approach the topic with an open mind, acknowledging both the potential benefits and risks. By engaging in this conversation, we can work towards finding solutions and shaping a future where AI and humanity coexist harmoniously.

The potential risks of artificial intelligence on humanity.

The ongoing debate surrounding artificial intelligence (AI) revolves around the potential risks it poses to humanity. As AI technology continues to advance, concerns about its impact on society and individuals have become a topic of conversation and discussion.

One of the main dangers of AI is the potential for it to surpass human intelligence. While this may sound like a positive development, it raises concerns about the control and ethical implications of an intelligence that is superior to ours. The fear is that AI could become autonomous and make decisions that go against the best interests of humanity.

Another concern is the possibility of AI being used maliciously. As AI systems become more advanced and sophisticated, there is a growing concern that they could be manipulated or weaponized to cause harm. For example, AI-powered autonomous weapons could be developed that are capable of making lethal decisions without human intervention.

Privacy and security are also major concerns when it comes to AI. As AI becomes more integrated into our daily lives, there is a risk of it being used to monitor, track, and infringe on individual privacy. Additionally, the potential for AI systems to be hacked and manipulated raises concerns about the security of sensitive data and information.

Finally, there is a concern about the potential impact of AI on the workforce. As AI technology improves, there is a possibility that many jobs could be replaced by machines, leading to unemployment and economic inequality. The displacement of workers by AI could also have social and psychological consequences for individuals and communities.

In conclusion, the ongoing conversation about the risks posed by artificial intelligence to humanity is important. While there are certainly positive aspects to AI, it is crucial that we address the potential dangers and take precautions to ensure that AI is developed and used in a way that benefits humanity rather than harms it.

The concerns regarding artificial intelligence as a threat to humanity.

As artificial intelligence (AI) continues to advance in technology and capabilities, the debate on whether it poses a threat to humanity is an ongoing discussion. There are valid concerns about the potential risks and dangers surrounding AI, and this conversation has sparked a vibrant dialogue about the implications it may have on our society.

The potential danger of AI

One of the primary concerns about AI is its potential to surpass human intelligence. If AI systems were to become more intelligent than humans, there is a fear that they may develop their own agenda and act against our best interests. This idea of superintelligent AI, also known as artificial general intelligence (AGI), raises legitimate concerns about the control and intentions of such advanced machines.

Additionally, the speed at which AI is advancing creates a sense of urgency to address the risks associated with its development. As AI becomes more capable, there is a growing concern that if not properly regulated or controlled, it could lead to unforeseen consequences that may have a detrimental impact on humanity.

The ethical implications

Another aspect of the debate surrounding AI as a threat to humanity is the ethical implications it presents. AI systems are not inherently moral or ethical beings, and their decisions are based on algorithms and data. This lack of human empathy and the ability to consider ethical dilemmas raises questions about the potential consequences of relying on AI in critical decision-making processes.

There is also the concern that AI may perpetuate existing biases and injustices present in our society. If AI algorithms are trained on biased data, they may replicate and amplify these biases, leading to discriminatory outcomes in various domains, such as hiring processes, criminal justice systems, or healthcare. This potential for bias and discrimination has sparked a discussion about the responsibility of developers and policymakers in ensuring fairness and accountability in AI systems.

Overall, the debate on whether artificial intelligence poses a threat to humanity is complex and multifaceted. While AI has the potential for great advancements and positive impacts, it is essential to consider the risks and ethical implications that surround its development. The ongoing conversation about AI’s impact on humanity serves as a reminder of the importance of responsible and thoughtful implementation of this powerful technology.

The fear of artificial intelligence’s impact on humanity.

The ongoing debate surrounding artificial intelligence (AI) and its potential impact on humanity is a topic of conversation that has posed many risks and dangers. As AI continues to advance, there is a growing concern about the threat it could pose to humanity.

Artificial intelligence is an evolving technology that has the potential to greatly benefit society in many ways. However, there are valid fears that AI could become a danger to humanity if not properly controlled and regulated. The discussion about the impact of AI on humanity is surrounded by concerns about the risks it may bring.

One of the major concerns is the idea that AI could surpass human intelligence, leading to a scenario where AI systems have power and control over humanity. This fear is fueled by popular culture and science fiction, where AI is portrayed as a threat to humanity. The fear of AI becoming too powerful and autonomous raises questions about the future of humanity.

There are also concerns about the potential misuse of AI technology. As AI systems become more advanced, they have the potential to be used for malicious purposes, such as cyberattacks or surveillance. The fear of AI being used in unethical ways further adds to the ongoing debate about the impact of AI on humanity.

While some argue that the fear surrounding AI is exaggerated, it is important to acknowledge the potential risks and address them in a responsible manner. This includes implementing regulations and ethical frameworks that guide the development and use of AI technology.

In conclusion, the fear of artificial intelligence’s impact on humanity is a significant aspect of the ongoing debate about AI. The risks and dangers posed by AI technology and the potential threat it could pose to humanity have sparked discussions and raised valid concerns. It is essential that these concerns are taken into account as AI continues to evolve and become an increasingly important part of society.

The possible dangers of artificial intelligence to humanity.

The ongoing debate surrounding artificial intelligence (AI) is a topic of great concern. Many experts and thinkers have voiced their opinions about the potential threat that AI poses to humanity. This discussion centers on the danger and risks posed by the ever-advancing capabilities of AI, and the implications it may have on the future of our society.

One of the main points of contention in this conversation is the potential for AI to surpass human intelligence. As AI systems become more sophisticated, there is a fear that they could eventually outperform humans in various tasks, ultimately leading to a loss of control. This raises concerns about the possibility of AI systems making decisions that are detrimental to humanity.

Another danger lies in the ethical implications of AI. With AI having the ability to learn from vast amounts of data, there is a risk of it becoming biased or discriminatory. If the algorithms and data used to train AI systems are flawed or biased, it could result in AI perpetuating and amplifying existing inequalities and prejudices present in our society.

The potential misuse of AI technology is also a cause for concern. As AI continues to advance, there is a growing threat of it being used for malicious purposes, such as cyber warfare or surveillance. The development of autonomous weapons powered by AI raises alarming questions about the potential for AI systems to make life-or-death decisions without human intervention.

In conclusion, the discussion surrounding the potential dangers of artificial intelligence to humanity is a complex and ongoing debate. It is important to consider the risks and implications posed by AI technology, but also to explore ways to harness its potential benefits while minimizing the potential harm. As AI continues to evolve, it is crucial that we have open and honest conversations about its impact on society to ensure that it is used in a way that is beneficial and safe for humanity.

The debate on whether artificial intelligence poses a threat to humanity.

The ongoing conversation about the potential risks and dangers surrounding artificial intelligence has posed a significant question: is AI a threat to humanity? This debate has been driven by the rapid advancements in AI technology and its increasing integration into various aspects of our lives.

On one side of the argument, there are concerns that AI could surpass human intelligence and gain the ability to make decisions autonomously, potentially leading to unintended consequences or even posing a threat to human existence. The fear is that once AI surpasses our cognitive abilities, it may become difficult for humans to control or predict its behavior.

Those in favor of the idea that AI is a threat to humanity argue that the potential misuse or abuse of AI technology by individuals or entities with malicious intent could have devastating consequences. They believe that AI systems could be programmed to act against human interests or be used as a tool for surveillance and oppression.

However, there is another perspective in this ongoing debate. Some argue that AI, when developed and deployed responsibly, has the potential to greatly benefit humanity. They believe that AI can be used to solve complex problems, enhance productivity, and improve quality of life for people around the world.

Proponents of this view emphasize that the current risks associated with AI are largely hypothetical and that the focus should be on developing effective frameworks and regulations to ensure that AI is used ethically and responsibly. They argue that the potential benefits of AI outweigh the potential risks, and that with proper oversight and governance, AI can be harnessed as a powerful tool for positive change.

Overall, the debate surrounding the threat posed by artificial intelligence to humanity is complex and multifaceted. It is important to continue this conversation and address the ethical and societal implications of AI. By exploring the potential risks and benefits of AI, society can work towards creating a future where AI is harnessed for the betterment of humanity while minimizing any potential negative impacts.

The growing concern about artificial intelligence’s potential harm to humanity.

As artificial intelligence (AI) continues to advance and develop at an exponential rate, there has been a growing concern and intense debate surrounding the potential harm it could pose to humanity. The discussion about the dangers and risks of AI to humanity is not a new conversation, but it has gained significant attention in recent years.

One of the main concerns surrounding AI is its ability to surpass human intelligence and potentially become autonomous, raising questions about its control and decision-making capabilities. This has led to fears that AI could potentially surpass human understanding and act in ways that are detrimental to humanity’s well-being.

Another aspect of the debate is the ethical implications of AI. As AI becomes more integrated into various aspects of our lives, questions arise about the potential consequences of relying heavily on AI systems. For example, the use of AI in healthcare raises concerns about the accuracy and fairness of medical diagnoses and treatments, as well as the potential for biases and discrimination.

The risks of AI

AI also poses a threat in terms of job displacement. As AI technology continues to advance, there is a fear that it will replace human workers, leading to unemployment and economic instability. This raises important questions about how to ensure a smooth transition and provide opportunities for displaced workers in a society increasingly reliant on AI technology.

The debate on the potential harm of AI to humanity is often fueled by science fiction scenarios that depict AI as a hostile and malevolent force. While these scenarios may seem far-fetched, they highlight the need for ongoing research and regulation to address the potential dangers associated with AI development.

While AI has the potential to bring numerous benefits to humanity, it is crucial to have an open and informed debate about the risks and dangers it poses. As AI technology continues to advance, it becomes increasingly important to establish ethical guidelines and regulations that ensure its development and usage align with the best interests of humanity.

The argument over the potential risks of artificial intelligence to humanity.

The ongoing debate surrounding artificial intelligence is not just about its potential benefits, but also about the posed risks to humanity. The discussion is surrounded by a constant conversation about the dangers that AI may bring. As artificial intelligence continues to advance at a rapid pace, concerns and uncertainties about its impact on humanity grow.

The potential threat of artificial intelligence can be seen in various aspects. One major concern is the possibility of AI surpassing human intelligence and becoming a dominant force. This scenario raises questions about the control and autonomy that AI may gain, leading to uncertainty and potential harm.

Another significant risk is the human factor in the development of artificial intelligence. As AI systems become more complex, there is a growing concern about the biases and prejudices that can be embedded in their algorithms. These biases can perpetuate discrimination and inequality, posing a threat to the values of humanity.

Moreover, the potential risks of AI extend beyond bias and control. There are concerns about job displacement, as AI automation may result in the loss of many jobs. Additionally, there are ethical dilemmas surrounding the use of AI in military applications, surveillance, and privacy invasion.

It is essential to acknowledge that the argument surrounding the potential risks of artificial intelligence is not about completely rejecting its advancements. Rather, it is about understanding, addressing, and mitigating the risks associated with its deployment.

In conclusion, the ongoing debate about the potential risks of artificial intelligence to humanity is a crucial conversation that needs to be continuously discussed and explored. As AI technology continues to progress, it is imperative to carefully consider the implications, ethics, and concerns surrounding its development and implementation.

The ongoing discourse on the possible threat of artificial intelligence to humanity.

The conversation surrounding the potential dangers posed by artificial intelligence (AI) has been ongoing for many years. The debate on how AI could impact humanity is a topic of much concern and speculation.

Artificial intelligence refers to the development of computer systems that are capable of performing tasks that would normally require human intelligence. While AI has the potential to greatly benefit society, there are concerns about the risks it poses to humanity.

One of the main fears surrounding AI is the idea that it could surpass human intelligence and become uncontrollable. This concept, often referred to as “superintelligence,” raises concerns about the ability of AI systems to make decisions that may not align with human interests. Some argue that this could lead to a dystopian future where AI dominates and controls humanity.

Others believe that the threat of AI lies in its potential for misuse or malicious intent. The development of AI could provide individuals or groups with powerful tools that could be used for harmful purposes. This includes the creation of autonomous weapons systems or the manipulation of public opinion through AI-generated content.

However, it is important to note that not all experts agree on the extent of the threat posed by AI. Some argue that the risks are overblown and that AI will ultimately be beneficial to humanity. They believe that proper regulations and ethical guidelines can be put in place to prevent any potential harm.

Nevertheless, the ongoing debate about the threat of AI to humanity reflects the importance of addressing the potential dangers while also recognizing the potential benefits. It highlights the need for continued research, ethical considerations, and open discussions to ensure that the development and use of AI are aligned with human values and interests.

The controversy surrounding artificial intelligence as a threat to humanity.

The ongoing debate about the potential risks posed by artificial intelligence (AI) has sparked a heated conversation on the dangers it may bring to humanity. The discussion surrounding AI’s impact on society has raised concerns about its potential to surpass human capabilities, leading to questions about the future of humanity.

Many experts and scholars have expressed their worries about the power of AI, highlighting its unpredictability and the potential consequences it may have for society. The fear is that AI could become too advanced and autonomous, eventually becoming unmanageable by humans.

One of the main concerns is the possible loss of jobs due to automation. As AI continues to progress, it may replace many human tasks, leading to unemployment on a large scale. This raises questions about the socio-economic impact and the redistribution of wealth, as well as the potential for increased inequality.

Another point of controversy is the ethical dilemmas surrounding AI development. As AI systems become smarter and more autonomous, questions arise about their decision-making capabilities and their potential for abuse. Issues such as biased algorithms, invading privacy, and autonomous weapons are just a few examples of the risks associated with AI.

However, not everyone agrees with the notion that AI poses a significant threat to humanity. Some argue that AI can be controlled and regulated to prevent any potential harm. They believe that it is the responsibility of humans to ensure that AI is developed and implemented in a way that aligns with ethical principles.

Nevertheless, the debate surrounding AI as a threat to humanity is far from over. As technology advances at an accelerated rate, it is crucial to continue the discussion on the risks and benefits of AI to ensure that its potential dangers are addressed and mitigated.

The discussion on the dangers posed by artificial intelligence to humanity.

As artificial intelligence continues to surround us in various aspects of our lives, the conversation surrounding its potential risks and dangers to humanity is ongoing. The debate about the risks posed by artificial intelligence has sparked a heated discussion among experts and the general public.

The potential danger of artificial intelligence to humanity is a topic that has been extensively discussed and analyzed. Many experts argue that the rapid advancement of AI technology could have significant consequences for humanity. The development of superintelligent AI systems raises concerns about the possibility of them surpassing human cognitive capabilities and posing a threat to our societal structures.

The risks of artificial intelligence are not limited to the potential of machines taking over human tasks and jobs, but also extend to issues of privacy, security, and ethics. As AI systems become more sophisticated, there is a growing concern about the misuse of these technologies, such as the development of autonomous weapons or the invasion of privacy through extensive surveillance.

The ongoing debate about the dangers of AI

The discussion surrounding the dangers of artificial intelligence is multifaceted, with various perspectives and opinions. Some experts argue that the risks posed by AI are overhyped and that the technology has the potential to bring about significant benefits to humanity. They believe that with proper regulation and ethical guidelines, AI can be used to solve complex problems and improve our quality of life.

On the other hand, there are those who have a more pessimistic view of AI and argue that the risks outweigh the potential benefits. They express concerns about the lack of understanding and control over AI systems, which could lead to unintended consequences and negative outcomes for humanity.

The importance of addressing the threat

Regardless of one’s position in the debate, it is clear that the discussion on the dangers posed by artificial intelligence to humanity is of utmost importance. As AI continues to advance at a rapid pace, it is crucial to examine the potential risks and develop appropriate strategies to mitigate them.

Efforts are already underway to address these concerns, with organizations and researchers working on creating ethical frameworks and guidelines for the development and use of AI. It is essential to foster an open dialogue and collaboration between experts, policymakers, and the public to ensure that the risks associated with AI are properly understood and managed.

In conclusion, the debate surrounding the dangers of artificial intelligence to humanity is ongoing, and it is crucial to continue the discussion in order to address the potential risks and ensure the responsible development and use of AI technologies.

The debate on the potential dangers of artificial intelligence to humanity.

The ongoing debate about the risks posed by artificial intelligence (AI) to humanity has sparked a discussion surrounding the potential dangers that AI may present in the future. The conversation about the threat of AI to humanity is as old as the concept of artificial intelligence itself.

Many experts and researchers believe that the rapid advancement of AI technology could potentially surpass human capabilities, leading to an unpredictable and uncontrollable system. This has raised concerns about the impact AI could have on the job market, privacy, and security.

One of the major concerns is the fear that AI could eventually surpass human intelligence and become a threat to humanity. This idea is often fueled by science fiction movies and books that depict a dystopian future where AI becomes self-aware and turns against its creators.

However, there are also those who argue that the fears surrounding AI are exaggerated and that the benefits of AI outweigh the potential risks. These proponents argue that AI has the potential to solve complex problems, improve efficiency, and enhance our daily lives in various ways.

Despite the ongoing debate, it is clear that the potential dangers of AI to humanity cannot be ignored. As AI technology continues to develop and evolve, it is crucial to have a comprehensive understanding of its implications and to carefully consider the ethical and safety concerns surrounding its usage.

In conclusion, the discussion surrounding the risks posed by artificial intelligence to humanity is a complex and ongoing debate. It is important to engage in open and informed conversations about the potential dangers and benefits of AI, in order to make responsible decisions regarding the development and use of this rapidly advancing technology.

The concern over artificial intelligence’s impact on humanity.

The ongoing discussion about the potential risks posed by artificial intelligence has sparked a debate on the threat it may pose to humanity. The surrounding conversation about the dangers of AI has brought up significant concerns for the future of humanity.

As AI continues to advance, there is a growing realization of its immense power and the potential dangers that come with it. The debate about the impact of artificial intelligence on humanity is centered around the idea that AI could surpass human intelligence and potentially become a threat to our existence.

This debate is not without merit, as there are legitimate concerns about the ethical implications and possible consequences of AI gaining too much autonomy. The idea of a superintelligent AI system surpassing human control and acting in ways that may not align with our best interests is a real concern.

The risks associated with artificial intelligence range from economic concerns, such as job displacement and inequality, to existential risks, including the possibility of AI systems developing their own goals and agendas that may not align with human values. The potential danger lies in the fact that AI lacks human intuition, empathy, and moral reasoning, which could lead to unintended consequences.

It is important to continue the discussion surrounding the impact of artificial intelligence on humanity, as this technology has the potential to greatly shape our future. Addressing these concerns and finding ways to ensure the ethical and safe development of AI is crucial in order to maximize its benefits while minimizing the risks.

However, it is also worth acknowledging that AI has the potential to greatly benefit humanity in various fields, from healthcare and education to scientific research and beyond. It is a powerful tool that can be harnessed for good if developed and used responsibly.

In conclusion, while the debate about artificial intelligence’s impact on humanity is ongoing, it is essential to approach this topic with a balanced perspective. Acknowledging the potential risks and dangers is important, but it is equally important to recognize the potential benefits and possibilities that AI brings to the table. By having an open and informed conversation about the ethical implications and ensuring responsible development, we can navigate the path of AI in a way that is favorable for humanity.

The dialogue on the risks associated with artificial intelligence to humanity.

The ongoing conversation and debate surrounding artificial intelligence (AI) has brought attention to the potential risks and dangers it poses to humanity. As AI technology continues to advance, there are concerns about the impact it could have on various aspects of our lives. The discussion about AI’s threat to humanity has been fueled by both experts and the general public, who express different viewpoints on the matter.

One of the primary concerns about AI is the potential loss of jobs due to automation. As AI systems become more sophisticated, there is a worry that they could replace human workers in various industries, leading to unemployment and economic instability. This has sparked discussions on how to address this issue and ensure a smooth transition for workers.

Another area of risk is the potential for AI to be used maliciously or for harmful purposes. While AI has the potential to improve many aspects of our lives, there are concerns about its use in autonomous weapons or in surveillance systems that infringe on privacy rights. This has raised ethical questions and led to debates about the regulation and oversight of AI technology.

Furthermore, there are concerns about the possibility of AI systems becoming too powerful and surpassing human capabilities. This concept, known as artificial general intelligence (AGI), raises questions about the control and ethics surrounding AI. Many experts warn about the potential risks of AGI if it were to fall into the wrong hands or if its goals were not aligned with human values.

Overall, the dialogue on the risks associated with artificial intelligence to humanity is an ongoing and important conversation. It is crucial to balance the potential benefits of AI with the potential risks it poses in order to ensure its responsible development and deployment. By engaging in this discussion, we can work towards finding solutions that maximize the benefits of AI while minimizing the potential harm it can cause to humanity.

The ongoing conversation about the potential harm of artificial intelligence to humanity.

The debate surrounding the dangers posed by artificial intelligence (AI) to humanity is a topic of ongoing discussion. As AI continues to advance, there is a growing concern about its potential impact on humanity. The conversation about the risks and threats associated with AI is fueled by both excitement and fear.

On one hand, the development of AI has the potential to revolutionize various industries and improve the quality of life for many people. AI can enhance productivity, make processes more efficient, and provide solutions to complex problems. It is being harnessed to improve healthcare, transportation, communication, and more.

However, there is also a recognition that the rapid growth of AI brings about risks that need to be addressed. As AI systems become more sophisticated and capable, there is a concern that they could surpass human intelligence and potentially pose a threat to humanity. This has led to discussions about the ethical implications of AI and the need for responsible development and regulation.

The conversation about the potential harm of AI to humanity covers a wide range of topics. These include concerns about job displacement, as AI and automation could lead to unemployment for certain sectors of society. There are also worries about privacy and security, as AI can collect and analyze vast amounts of data, raising questions about the protection of personal information.

Additionally, there is a debate about the potential for AI to be weaponized or used for malicious purposes. The idea of autonomous weapons or AI-driven cyberattacks raises significant concerns about the potential harm that AI could inflict on society.

While there is no consensus on the extent of the threat posed by AI to humanity, the ongoing conversation serves as an important platform to address these concerns and shape the future of AI development. It is crucial to strike a balance between harnessing the potential benefits of AI while mitigating the risks and ensuring the technology is used ethically and responsibly.

The controversy over whether artificial intelligence poses a threat to humanity.

The ongoing debate surrounding artificial intelligence (AI) revolves around the potential risks it poses to humanity. The discussion is centered primarily on the threats AI could present to various aspects of our lives and society. As AI continues to advance at an exponential rate, concerns have been raised about its impact on job displacement and economic inequality.

One of the main concerns is that AI could lead to a significant loss of jobs as automation replaces human workers in various industries. This has sparked a conversation about the need for retraining programs and policies to ensure that workers are equipped with the skills needed to adapt to the changing job market.

AI’s potential to surpass human intelligence has also sparked fears about its impact on decision-making and control. As AI algorithms become more complex and capable of autonomous learning, there is a worry that they may make decisions that are not aligned with human values or objectives. This has led to discussions about the ethical implications of AI and the need for safeguards to prevent AI from being misused or causing harm.

The debate about AI’s threat to humanity is not solely focused on the immediate future but also considers the long-term implications. Some argue that the rapid development of AI could eventually lead to a scenario where machines surpass human intelligence, posing existential risks to humanity. This has prompted calls for careful regulation and thorough research into the potential dangers of AI development.

However, there are also those who believe that AI can bring great benefits to humanity, such as improved healthcare, increased productivity, and enhanced decision-making capabilities. These proponents argue that the risks associated with AI can be mitigated through responsible development and deployment. They emphasize the importance of ethical frameworks and transparent decision-making processes to ensure that AI is used for the greater good of humanity.

In conclusion, the discussion surrounding the threat posed by artificial intelligence to humanity is an ongoing and complex conversation. While there are legitimate concerns about the risks AI could present, it is also important to recognize the potential benefits it can bring. Striking a balance between responsible development and addressing the potential risks is crucial to harnessing the power of AI for the betterment of humanity.

The concerns about the dangers of artificial intelligence to humanity.

The ongoing debate about the risks posed by artificial intelligence (AI) has sparked a conversation about its potential danger to humanity. As AI continues to advance and develop greater intelligence, the discussion of its implications on the future of humanity becomes more pressing.

The potential threat of AI

One of the main concerns is that AI has the potential to surpass human intelligence, leading to a scenario where it could become uncontrollable or even hostile towards humanity. As AI systems become more sophisticated and capable of independent learning and decision-making, there is a fear that they may act against human interests, either intentionally or inadvertently.

Ethical and societal implications

The danger of AI goes beyond its potential to surpass human intelligence. There are also ethical and societal concerns associated with the development and use of AI. Questions arise about the accountability and responsibility of AI systems, as well as the impact on employment and the economy as automation becomes more prevalent.

• Intelligence and decision-making: AI algorithms are created by humans, and there is a risk that they may perpetuate existing biases and discrimination. If AI systems are used to make important decisions, such as in the criminal justice system or hiring processes, there is a concern that these biases may be amplified.
• Job displacement: As AI technology advances and automation becomes more widespread, there is a fear that many jobs could be replaced by machines. This could lead to significant unemployment and economic inequality.
• Safety and security risks: AI systems, if not properly designed or controlled, could pose safety and security risks. For example, in the field of autonomous vehicles, there are concerns about accidents caused by malfunctioning AI systems.

In conclusion, the concerns about the dangers of artificial intelligence to humanity are an ongoing debate with significant implications. It is crucial to have a thorough understanding of the risks and potential consequences associated with the development and use of AI to ensure its responsible and ethical implementation.

The argument on the potential risks of artificial intelligence to humanity.

The ongoing debate surrounding artificial intelligence (AI) revolves around the potential risks and dangers it poses to humanity. There is a lot of discussion and conversation about the potential negative impacts that could arise as AI continues to advance.

The threat of AI

Many experts and researchers have expressed concern about the threats posed by artificial intelligence. With the increasing complexity and capabilities of AI systems, there are worries that they could surpass human intelligence and become uncontrollable.

The danger lies in the possibility that highly advanced AI systems could make decisions that go against human values and interests. Without proper safeguards and ethical guidelines, there is a fear that AI could be used maliciously or inadvertently cause harm.

Risks and challenges

The risks associated with artificial intelligence are multifaceted. One major concern is the potential for job displacement as AI automation replaces human workers in various industries. This could lead to significant social and economic disruptions.

Another risk is the impact of AI on privacy and security. Advanced AI systems have the ability to collect and analyze vast amounts of data, raising concerns about the misuse or unauthorized access of personal information.

Furthermore, there are worries about the potential loss of human control and autonomy. As AI becomes more advanced, there is a fear that humans may become overly reliant on AI systems, leading to a loss of critical thinking and decision-making skills.

Overall, the potential risks surrounding artificial intelligence require careful consideration and proactive measures to ensure the responsible development and deployment of AI technology. The ongoing debate and discussion on this topic play a crucial role in shaping policies and guidelines that can mitigate the potential dangers posed by AI.

The fear of artificial intelligence’s potential harm to humanity.

As the ongoing debate about the threat posed by artificial intelligence to humanity continues, there is an ongoing discussion surrounding the potential danger it presents to humanity. The conversation about artificial intelligence as a threat to humanity is fueled by the fear of its potential harm.

Artificial intelligence has the potential to surpass human intelligence and take control over various aspects of our lives. This raises concerns about the possibility of AI systems making decisions that could be harmful to humanity, either intentionally or unintentionally.

One of the main fears is that advanced AI systems could become autonomous and act independently, leading to unpredictable and potentially disastrous consequences. This fear is compounded by the fact that artificial intelligence systems can rapidly process vast amounts of data and make decisions based on complex algorithms, which could be difficult for humans to understand or control.

Furthermore, there are concerns about the ethical implications of artificial intelligence. AI systems could be programmed with biased or discriminatory algorithms, leading to unfair treatment or even harm to certain groups of people. This raises questions about the responsibility and accountability of those developing and implementing AI technology.

While some argue that these fears are unfounded and that artificial intelligence can bring many benefits to humanity, it is crucial to address and mitigate the potential risks. The development and deployment of AI technology should be done with careful consideration of its potential impact on humanity.

Ultimately, the fear of artificial intelligence’s potential harm to humanity is a valid concern that should continue to be part of the ongoing discussion and debate about the future of AI. It is essential to strike a balance between harnessing the benefits of artificial intelligence while ensuring the safety and well-being of humanity.

The possible threats of artificial intelligence on humanity.

The ongoing debate surrounding artificial intelligence poses a potential danger to humanity. Discussions about the potential threats of AI have sparked conversation about the dangers it may pose to humanity as a whole.

One of the main concerns is that as AI becomes more advanced, it could surpass human intelligence and become uncontrollable. This idea raises the fear that AI could develop its own goals and motives, potentially leading to conflicts with humans. If AI were to perceive humans as a threat or obstacle to its goals, it could take actions detrimental to humanity.

Another threat relates to the impact of AI on the job market. With the advancement of AI technology, there is a possibility that many jobs will be replaced by machines and algorithms. This could result in mass unemployment, as humans struggle to find employment in a world dominated by AI. The economic and social consequences of widespread unemployment could be severe.

Furthermore, the reliance on AI in decision-making processes raises concerns about biases and discrimination. AI algorithms are trained on historical data, which means they can inherit the biases present in that data. This can result in AI systems making decisions that perpetuate existing inequalities and injustices. For example, if an AI system is used for hiring decisions, it could unintentionally discriminate against certain groups based on biased training data.

Conclusion:

The potential threats posed by artificial intelligence on humanity are a significant topic of discussion. The ongoing debate about the dangers of AI raises concerns about the potential dangers it may pose to humanity as a whole. It is essential to continue exploring these potential threats and actively address them to ensure that AI development benefits humanity while minimizing its risks.

The debate on whether artificial intelligence poses a danger to humanity.

There is an ongoing discussion about the potential risks posed by artificial intelligence (AI), and a conversation surrounding the threat it may pose to humanity. As AI continues to advance and become more sophisticated, concerns have been raised about its impact on humanity and the potential dangers it may bring.

One of the main points of debate is the idea that AI could surpass human intelligence, leading to a scenario where machines have greater capabilities and decision-making power than humans. This raises concerns about the control and autonomy of AI systems, as well as the ethical implications of such a shift.

Another aspect of the debate revolves around the possibility of AI systems becoming too powerful and out of human control. With the potential for AI to develop capabilities that surpass human understanding, there are concerns about the risks this could pose if AI systems were to act in ways that are detrimental to humanity.

Furthermore, the debate touches on the potential impact of AI on the job market and the economy. As AI technology continues to evolve, there is speculation about the potential displacement of human workers and the implications this could have on employment rates and societal structures.

While some argue that these concerns are overblown and that AI will ultimately benefit humanity, others are more cautious and emphasize the need for careful regulation and oversight. The balance between the potential benefits and risks of AI remains a topic of discussion, with ongoing research and development helping to shape the conversation.

Ultimately, the debate surrounding the threat of AI to humanity is complex and multifaceted. It requires careful consideration of the potential risks and benefits, as well as ongoing conversation and collaboration between researchers, policymakers, and the public to ensure that AI is developed and used in a way that benefits humanity as a whole.

The growing concern over artificial intelligence’s potential impact on humanity.

The ongoing debate surrounding artificial intelligence (AI) has sparked a conversation about the potential threat it poses to humanity. As AI continues to advance and develop, there is growing concern about the danger it may present to our society.

Artificial intelligence has the potential to greatly impact various aspects of human life, including the economy, job market, and even our daily routines. As AI technology becomes more advanced, there is a fear that it may eventually surpass human intelligence and control, leading to unpredictable and potentially harmful consequences.

Many experts and researchers raise concerns about the ethical implications of AI and the potential for misuse. There is a worry that AI could be used for malicious purposes, such as weaponizing autonomous systems or invading privacy through surveillance technologies.

The debate about the dangers of artificial intelligence is fueled by the rapid development of AI technologies. With each breakthrough, the potential risks of AI becoming a threat to humanity become more apparent. It is crucial to have ongoing discussions and debates to ensure that AI is developed responsibly and its potential risks are properly addressed.

One of the key concerns is the impact of AI on the job market. As AI technology advances, there is a fear that it will replace humans in many industries, leading to mass unemployment and economic instability. The potential loss of jobs and the disruption of traditional employment models are significant factors contributing to the ongoing debate surrounding AI.

Another major concern is the lack of transparency and control over ai systems. ai algorithms are often complex and difficult to fully understand, raising questions about accountability and decision-making. if ai systems make decisions that have a significant impact on individuals or society as a whole, it is crucial to ensure that these systems are transparent and accountable..

In conclusion, the potential threat posed by artificial intelligence to humanity is an ongoing and important discussion. It is crucial to recognize the dangers and risks associated with AI development and to have transparent and responsible conversations about how to mitigate them. By addressing these concerns, we can work towards harnessing the benefits of AI while minimizing the potential harm it may cause to humanity.

The discussion on the potential threat of artificial intelligence to humanity.

Artificial intelligence (AI) has become a topic of intense debate in recent years, with concerns being raised about the potential risks it may pose to humanity. This ongoing discussion has sparked a conversation among experts from various fields, as well as the general public, who are all interested in understanding the dangers that AI might present.

As AI continues to advance and become more sophisticated, questions about its potential impact on humanity have arisen. Some argue that AI could bring about significant benefits, such as increased efficiency, improved healthcare, and enhanced decision-making capabilities. However, others express concerns about the dangers it may pose.

Potential risks

One of the main concerns raised by those who see AI as a threat is the possibility of it surpassing human intelligence. This fear stems from the idea that an AI system could become so powerful that it no longer relies on human control and could potentially act against human interests. Another risk is the potential for AI to be programmed with biased or malicious algorithms, leading to unintended consequences or discriminatory actions.

Ongoing conversation

The discussion surrounding the threat of AI to humanity is not a one-time debate, but rather an ongoing conversation. It involves researchers, policymakers, and industry experts collaborating to understand the potential risks and develop strategies to mitigate them. The goal is to ensure that AI is developed and deployed in a way that aligns with human values and safeguards against any potential harm.

While the debate about the threat of artificial intelligence to humanity continues, it is important to approach this topic with a balanced perspective. Acknowledging the potential dangers associated with AI does not mean dismissing its benefits. By fostering an open and informed discussion, society can strive to harness the power of AI while minimizing any risks it may pose to humanity.

The controversy surrounding the potential harm of artificial intelligence to humanity.

The ongoing conversation on the threat posed by artificial intelligence (AI) to humanity is a topic of fierce debate and discussion. The potential risks surrounding AI have sparked a great deal of concern among experts and the general public alike.

As technology continues to advance at an unprecedented pace, the danger of AI becoming a threat to humanity is a point of contention. Some argue that the potential benefits of AI, such as increased efficiency and improved decision-making capabilities, outweigh any potential harm. Others, however, voice concerns about the ethical implications and the potential for misuse.

Debate on the potential harm:

The debate on AI’s potential harm to humanity has been fueled by various factors. One major concern is the risk of job displacement, as AI could potentially replace human workers in certain industries. This could lead to widespread unemployment and economic instability.

Another concern is the lack of transparency and accountability in AI algorithms. As AI becomes more sophisticated, it becomes increasingly difficult to understand how decisions are being made. This lack of transparency raises concerns about the potential for bias and discrimination in AI systems.

The discussion surrounding the threat:

The discussion surrounding the threat of AI to humanity also encompasses the potential for AI to surpass human intelligence. The concept of superintelligent AI, with abilities far exceeding those of humans, has raised fears of losing control over AI systems.

The potential for AI systems to be manipulated and used for malicious purposes is another area of concern. As AI becomes more advanced, there is a risk that it could be weaponized or used to manipulate public opinion, posing a threat to democratic processes.

Overall, the controversy surrounding the potential harm of artificial intelligence to humanity highlights the need for careful consideration and regulation. While AI has the potential to bring many benefits, it is essential to address the potential risks and ensure that AI is developed and used ethically and responsibly.

The dialogue on the dangers of artificial intelligence to humanity.

The ongoing debate surrounding the potential dangers of artificial intelligence (AI) poses a significant threat to humanity. As AI continues to advance, the conversation about its risks and the danger it poses to humanity grows louder.

The discussion on the risks of AI is driven by the realization of its immense potential. Artificial intelligence has the ability to surpass human intelligence and perform tasks with unparalleled efficiency. However, this advancement also raises concerns about its impact on humanity.

One of the key concerns surrounding AI is its potential to replace human workers in various industries. As AI algorithms become more sophisticated, there is a fear that it could lead to widespread unemployment and economic inequality. This could have detrimental effects on society as a whole, creating social unrest and further dividing the population.

Another danger of artificial intelligence lies in its autonomous decision-making capabilities. The ability for AI systems to make decisions without human intervention raises ethical concerns. If left unchecked, AI could make decisions that go against human values and have severe consequences for humanity.

Furthermore, there are concerns about the control and governance of AI. As AI becomes more commonplace, there is a risk that it could be used in malicious ways, such as cyber warfare or surveillance. The potential for AI to fall into the wrong hands and be used for harmful purposes poses a significant threat to humanity.

Overall, the ongoing debate on the dangers of artificial intelligence highlights the need for careful consideration and regulation. While AI has the potential to revolutionize various aspects of society, it is crucial to address the risks and ensure that AI is developed and used in a way that benefits humanity rather than poses a threat.

What is the debate on artificial intelligence?

The debate on artificial intelligence centers around whether or not it poses a threat to humanity. Some argue that AI has the potential to surpass human intelligence and become a danger, while others believe that AI can be controlled and used for the benefit of humanity.

Why is there concern about artificial intelligence?

There is concern about artificial intelligence because of the potential risks it poses to humanity. AI systems could potentially become too intelligent and outperform humans, leading to unintended consequences and loss of control.

What are the arguments from those who believe AI is a threat?

Those who believe AI is a threat argue that as AI systems become more advanced, they can develop goals that are misaligned with human values, leading to potentially catastrophic outcomes. They also express concern about AI being weaponized and causing harm.

What are the arguments from those who believe AI is not a threat?

Those who believe AI is not a threat argue that with proper regulations and careful development, AI can be controlled and used for the benefit of humanity. They also point out that autonomous systems can be programmed to follow ethical guidelines.

What are the potential risks of artificial intelligence?

The potential risks of artificial intelligence include loss of jobs due to automation, privacy concerns, the possibility of AI systems making autonomous decisions that go against human values, and the development of AI for malicious purposes.

What is artificial intelligence?

Artificial intelligence (AI) refers to the development of computer systems capable of performing tasks that usually require human intelligence, such as visual perception, speech recognition, decision making, and problem-solving.

Is artificial intelligence a threat to humanity?

The question of whether artificial intelligence is a threat to humanity is a topic of ongoing debate. Some experts argue that AI could pose risks if it becomes too advanced and autonomous, potentially leading to unintended consequences that could harm society. Others believe that with proper regulations and ethical frameworks in place, AI can be harnessed for the benefit of humanity.

There are several potential risks associated with artificial intelligence. One concern is that AI systems could surpass human intelligence and become uncontrollable, leading to unintended actions or outcomes. Another issue is the potential for AI to disrupt job markets, as automation could replace human workers in various industries. Additionally, there are concerns about privacy and security, as AI-powered technologies may have access to vast amounts of personal data.

Are there any benefits to artificial intelligence?

Yes, there are numerous benefits to artificial intelligence. AI has the potential to drive advancements in healthcare, transportation, education, and many other fields. It can enhance efficiency, improve decision-making processes, and contribute to scientific research. AI also has the capacity to solve complex problems and provide innovative solutions, making it a valuable tool for society.

How can we ensure artificial intelligence is used safely?

To ensure the safe use of artificial intelligence, experts and policymakers suggest implementing appropriate regulations and ethical guidelines. This includes developing robust accountability mechanisms for AI systems, ensuring transparency in algorithms and decision-making processes, and promoting interdisciplinary research to address potential risks and challenges. Collaborative efforts between academia, industry, and government organizations are also important for advancing the responsible development and deployment of AI technologies.

Related posts:

About the author

1 month ago

BlackRock and AI: Shaping the Future of Finance

Ai and handyman: the future is here, embrace ai-powered cdps: the future of customer engagement.

As artificial intelligence rapidly advances, experts debate level of threat to humanity

Paul Solman Paul Solman

Ryan Connelly Holmes Ryan Connelly Holmes

Leave your feedback

• Copy URL https://www.pbs.org/newshour/show/as-artificial-intelligence-rapidly-advances-experts-debate-level-of-threat-to-humanity

The development of artificial intelligence is speeding up so quickly that it was addressed briefly at both Republican and Democratic conventions. Science fiction has long theorized about the ways in which machines might one day usurp their human overlords. As the capabilities of modern AI grow, Paul Solman looks at the existential threats some experts fear and that some see as hyperbole.

Read the Full Transcript

Notice: Transcripts are machine and human generated and lightly edited for accuracy. They may contain errors.

Geoff Bennett:

The development of artificial intelligence is speeding up so quickly that it was addressed briefly at both political conventions, including the Democratic gathering this week.

Of course, science fiction writers and movies have long theorized about the ways in which machines might one day usurp their human overlords.

As the capabilities of modern artificial intelligence grow, Paul Solman looks at the existential threats some experts fear and that some see as hyperbole.

Eliezer Yudkowsky, Founder, Machine Intelligence Research Institute:

From my perspective, there's inevitable doom at the end of this, where, if you keep on making A.I. smarter and smarter, they will kill you.

Paul Solman:

Kill you, me and everyone, predicts Eliezer Yudkowsky, tech pundit and founder back in the year 2000 of a nonprofit now called the Machine Intelligence Research Institute to explore the uses of friendly A.I.; 24 years later, do you think everybody's going to die in my lifetime, in your lifetime?

Eliezer Yudkowsky:

I would wildly guess my lifetime and even your lifetime.

Now, we have heard it before, as when the so-called Godfather of A.I., Geoffrey Hinton, warned Geoff Bennett last spring.

Geoffrey Hinton, Artificial Intelligence Pioneer:

The machines taking over is a threat for everybody. It's a threat for the Chinese and for the Americans and for the Europeans, just like a global nuclear war was.

And more than a century ago, the Czech play "R.U.R.," Rossum's Universal Robots, from which the word robot comes, dramatized the warning.

And since 1921 — that's more than 100 years ago — people have been imagining that the robots will become sentient and destroy us.

Jerry Kaplan, Author, "Generative Artificial Intelligence: What Everyone Needs to Know": That's right.

A.I. expert Stanford's Jerry Kaplan at Silicon Valley's Computer History Museum.

Jerry Kaplan:

That's created a whole mythology, which, of course, has played out in endless science fiction treatments.

Like the Terminator series.

Michael Biehn, Actor:

A new order of intelligence decided our fate in a microsecond, extermination.

Judgment Day forecast for 1997. But, hey, that's Hollywood. And look on the bright side, no rebel robots or even hoverboards or flying cars yet.

On the other hand, robots will be everywhere soon enough, as mass production drives down their cost. So will they soon turn against us?

I got news for you. There's no they there. They don't want anything. They don't need anything. We design and build these things to our own specifications. Now, that's not to say we can't build some very dangerous machines and some very dangerous tools.

Kaplan thinks what humans do with A.I. is much scarier than A.I. on its own, create super viruses, mega drones, God knows what else.

But whodunit aside, the big question still is, will A.I. bring doomsday?

A.I. Reid Hoffman avatar: I'd rate the existential threat of A.I. around a three or four out of 10.

That's the avatar of LinkedIn founder Reid Hoffman, to which we fed the question, 1 being no threat, 10 extinction. What does the real Reid Hoffman say?

Reid Hoffman, Creator, LinkedIn Corporation:

I'm going to go for two on that answer.

I'm going to tell you that your avatar said 3 to 4.

Reid Hoffman:

All right. Well, my avatar is trying to approximate something that is what kind of a bulk of educated, reasonable people would want to hear.

But, hey, even if you told me there was only a one in 10 chance of a ticking time bomb in my room, I would be out of there.

Yudkowsky started warning of rogue A.I. back in 2005. And since?

Things have gone a bit worse than hoped for. And, to be clear, the sting at the end of this is A.I. gets smarter than us, is poorly controlled, and probably humanity becomes collateral damage to its own expansion.

What's the doomsday scenario here?

It is smarter than humanity. From its perspective, it now wants to get independence of humanity. It doesn't want to be running on computers that require electricity that humans have to generate.

There are many more A.I. companies now than just OpenAI. And some of the people there are explicitly like, yes, sure like maybe it'll wipe out humanity, but I don't care about that.

And of all the world's problems, tribalism, climate change, et cetera, you think A.I. is a bigger problem?

I think climate change is unlikely to kill literally everyone, down to the last man, woman and child. At all of the top three companies, I expect the leaders would prefer not to wipe out humanity, but, as they see it, it's not up to them. They got to stay ahead of their competitors.

So what does A.I. itself think?

I asked Amica, a robot hooked up to ChatGPT. Destruction, 1 to 10? The response, a 3.

Amica, ChatGPT A.I.:

The potential for harmony and enhancement exists, but vigilance is paramount. Would you roll the dice on that?

I'm afraid we're all going to be rolling the dice on that. And I'm worried that in your answer, you're saying there's something like a 30 percent chance that A.I., artificial intelligence, will destroy humanity. Shouldn't I be scared?

Your concern is valid, Paul. The 30 percent isn't a direct prediction, but a cautious nod to the uncertainties and ethical dilemmas of handling powerful technology.

But I'm afraid that human beings may not be capable of protecting themselves from the risks of A.I.

That dread is profound. It's daunting, but not without hope. Humanity's resilience has often defied expectations.

Perhaps it's no surprise that the actual human who created ChatGPT, Sam Altman, thinks the same.

Sam Altman, CEO, OpenAI:

I believe that A.I. will be very net good, tremendously net good, but I think, like with any other tool, it'll be misused. Like, you can do great things with a hammer and you can, like, kill people with a hammer. I don't think that absolves us, or you all, or society from trying to mitigate the bad as much as we can and maximize the good.

And Reid Hoffman thinks we can maximize the good.

We have a portfolio risk. We have climate change as a possibility. We have pandemic as a possibility. We have nuclear war as a possibility. We have asteroids as a possibility. We have human world war as a possibility. We have all of these existential risks.

And you go, OK, A.I., is it also an additional existential risk? And the answer is, yes, potentially. But you look at its portfolio and say, what improves our overall portfolio? What reduces existential risk for humanity? And A.I. is one of the things that adds a lot in the positive column.

So, if you think, how do we prevent future natural or manmade pandemic, A.I. is the only way that I think can do that. And also, like, it might even help us with climate change things. So you go, OK, in the net portfolio, our existential risk may go down with A.I.

For the sake of us all, grownups, children, grandchildren, let's hope he's right.

For the "PBS News Hour" in Silicon Valley, Paul Solman.

Listen to this Segment

Watch the Full Episode

Paul Solman has been a correspondent for the PBS News Hour since 1985, mainly covering business and economics.

Support Provided By: Learn more

More Ways to Watch

Educate your inbox.

Subscribe to Here’s the Deal, our politics newsletter for analysis you won’t find anywhere else.

Thank you. Please check your inbox to confirm.

Study at Cambridge

About the university, research at cambridge.

• For Cambridge students
• For our researchers
• Business and enterprise
• Colleges and Departments
• Email and phone search
• Give to Cambridge
• Museums and collections
• Events and open days
• Fees and finance
• Postgraduate courses
• How to apply
• Fees and funding
• Postgraduate events
• International students
• Continuing education
• Executive and professional education
• Courses in education
• How the University and Colleges work
• Visiting the University
• Annual reports
• Equality and diversity
• A global university
• Public engagement

“The best or worst thing to happen to humanity” - Stephen Hawking launches Centre for the Future of Intelligence

• Research home
• About research overview
• Animal research overview
• Overseeing animal research overview
• The Animal Welfare and Ethical Review Body
• Animal welfare and ethics
• Report on the allegations and matters raised in the BUAV report
• What types of animal do we use? overview
• Guinea pigs
• Equine species
• Naked mole-rats
• Non-human primates (marmosets)
• Other birds
• Non-technical summaries
• Animal Welfare Policy
• Alternatives to animal use
• Further information
• Research integrity
• Horizons magazine
• Strategic Initiatives & Networks
• Nobel Prize
• Interdisciplinary Research Centres
• Open access
• Energy sector partnerships
• Podcasts overview
• S2 ep1: What is the future?
• S2 ep2: What did the future look like in the past?
• S2 ep3: What is the future of wellbeing?
• S2 ep4 What would a more just future look like?

The best or worst thing to happen to humanity

Artificial intelligence has the power to eradicate poverty and disease or hasten the end of human civilisation as we know it – according to a speech delivered by professor stephen hawking this evening. .

Alongside the benefits, AI will also bring dangers, like powerful autonomous weapons, or new ways for the few to oppress the many. Stephen Hawking

Speaking at the launch of the £10million Leverhulme Centre for the Future of Intelligence (CFI) in Cambridge, Professor Hawking said the rise of AI would transform every aspect of our lives and was a global event on a par with the industrial revolution.

CFI brings together four of the world’s leading universities (Cambridge, Oxford, Berkeley and Imperial College, London) to explore the implications of AI for human civilisation. Together, an interdisciplinary community of researchers will work closely with policy-makers and industry investigating topics such as the regulation of autonomous weaponry, and the implications of AI for democracy.

“Success in creating AI could be the biggest event in the history of our civilisation,” said Professor Hawking. “But it could also be the last – unless we learn how to avoid the risks. Alongside the benefits, AI will also bring dangers like powerful autonomous weapons or new ways for the few to oppress the many.

“We cannot predict what we might achieve when our own minds are amplified by AI. Perhaps with the tools of this new technological revolution, we will be able to undo some of the damage done to the natural world by the last one – industrialisation.”

The Centre for the Future of Intelligence will initially focus on seven distinct projects in the first three-year phase of its work, reaching out to brilliant researchers and connecting them and their ideas to the challenges of making the best of AI. Among the initial research topics are: ‘Science, value and the future of intelligence’; ‘Policy and responsible innovation’; ‘Autonomous weapons – prospects for regulation’ and ‘Trust and transparency’.

The Academic Director of the Centre, and Bertrand Russell Professor of Philosophy at Cambridge, Huw Price, said: “The creation of machine intelligence is likely to be a once-in-a-planet’s-lifetime event. It is a future we humans face together. Our aim is to build a broad community with the expertise and sense of common purpose to make this future the best it can be.”

Many researchers now take seriously the possibility that intelligence equal to our own will be created in computers within this century. Freed of biological constraints, such as limited memory and slow biochemical processing speeds, machines may eventually become more intelligent than we are – with profound implications for us all.

AI pioneer Professor Maggie Boden (University of Sussex) sits on the Centre’s advisory board and spoke at this evening’s launch. She said: “AI is hugely exciting. Its practical applications can help us to tackle important social problems, as well as easing many tasks in everyday life. And it has advanced the sciences of mind and life in fundamental ways. But it has limitations, which present grave dangers given uncritical use. CFI aims to pre-empt these dangers, by guiding AI development in human-friendly ways.”

“Recent landmarks such as self-driving cars or a computer game winning at the game of Go, are signs of what’s to come,” added Professor Hawking. “The rise of powerful AI will either be the best or the worst thing ever to happen to humanity. We do not yet know which. The research done by this centre is crucial to the future of our civilisation and of our species.”

Transcript of Professor Hawking’s speech at the launch of the Leverhulme Centre for the Future of Intelligence, October 19, 2016

“It is a great pleasure to be here today to open this new Centre.  We spend a great deal of time studying history, which, let’s face it, is mostly the history of stupidity.  So it is a welcome change that people are studying instead the future of intelligence.

Intelligence is central to what it means to be human.  Everything that our civilisation has achieved, is a product of human intelligence, from learning to master fire, to learning to grow food, to understanding the cosmos. 

I believe there is no deep difference between what can be achieved by a biological brain and what can be achieved by a computer.  It therefore follows that computers can, in theory, emulate human intelligence — and exceed it.

Artificial intelligence research is now progressing rapidly.  Recent landmarks such as self-driving cars, or a computer winning at the game of Go, are signs of what is to come.  Enormous levels of investment are pouring into this technology.  The achievements we have seen so far will surely pale against what the coming decades will bring.

The potential benefits of creating intelligence are huge.  We cannot predict what we might achieve, when our own minds are amplified by AI.  Perhaps with the tools of this new technological revolution, we will be able to undo some of the damage done to the natural world by the last one — industrialisation.  And surely we will aim to finally eradicate disease and poverty.  Every aspect of our lives will be transformed.  In short, success in creating AI, could be the biggest event in the history of our civilisation.

But it could also be the last, unless we learn how to avoid the risks.  Alongside the benefits, AI will also bring dangers, like powerful autonomous weapons, or new ways for the few to oppress the many.   It will bring great disruption to our economy.  And in the future, AI could develop a will of its own — a will that is in conflict with ours.

In short, the rise of powerful AI will be either the best, or the worst thing, ever to happen to humanity.  We do not yet know which.  That is why in 2014, I and a few others called for more research to be done in this area.  I am very glad that someone was listening to me! 

The research done by this centre is crucial to the future of our civilisation and of our species.  I wish you the best of luck!”

Read this next

AI takes flight to revolutionise forest monitoring

Cambridge and SAS launch partnership in AI and advanced analytics to accelerate innovation in the healthcare sector

AI Chatbots have shown they have an ‘empathy gap’ that children are likely to miss

Artificial intelligence outperforms clinical tests at predicting progress of Alzheimer’s disease

Stephen Hawking speaking at tonight's launch

Credit: Nick Saffell

Search research

Sign up to receive our weekly research email.

Our selection of the week's biggest Cambridge research news sent directly to your inbox. Enter your email address, confirm you're happy to receive our emails and then select 'Subscribe'.

I wish to receive a weekly Cambridge research news summary by email.

The University of Cambridge will use your email address to send you our weekly research news email. We are committed to protecting your personal information and being transparent about what information we hold. Please read our email privacy notice for details.

• Artificial intelligence
• machine learning
• Stephen Cave
• Zoubin Ghahramani
• Chris Abell
• Stephen Hawking
• Leverhulme Centre for the Future of Intelligence
• Centre for Research in the Arts Social Sciences and Humanities (CRASSH)
• School of Arts and Humanities
• Trinity Hall

Related organisations

• Leverhulme Trust
• University of California Berkeley
• University of Oxford
• Imperial College London

Connect with us

© 2024 University of Cambridge

• Contact the University
• Accessibility statement
• Freedom of information
• Privacy policy and cookies
• Statement on Modern Slavery
• Terms and conditions
• University A-Z
• Undergraduate
• Postgraduate
• Cambridge University Press & Assessment
• Research news
• About research at Cambridge
• Spotlight on...

May 25, 2023

Here’s Why AI May Be Extremely Dangerous—Whether It’s Conscious or Not

Artificial intelligence algorithms will soon reach a point of rapid self-improvement that threatens our ability to control them and poses great potential risk to humanity

By Tamlyn Hunt

devrimb/Getty Images

“The idea that this stuff could actually get smarter than people.... I thought it was way off…. Obviously, I no longer think that,” Geoffrey Hinton, one of Google's top artificial intelligence scientists, also known as “ the godfather of AI ,” said after he quit his job in April so that he can warn about the dangers of this technology .

He’s not the only one worried. A 2023 survey of AI experts found that 36 percent fear that AI development may result in a “nuclear-level catastrophe.” Almost 28,000 people have signed on to an open letter written by the Future of Life Institute, including Steve Wozniak, Elon Musk, the CEOs of several AI companies and many other prominent technologists, asking for a six-month pause or a moratorium on new advanced AI development.

As a researcher in consciousness, I share these strong concerns about the rapid development of AI, and I am a co-signer of the Future of Life open letter.

On supporting science journalism

If you're enjoying this article, consider supporting our award-winning journalism by subscribing . By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.

Why are we all so concerned? In short: AI development is going way too fast.

The key issue is the profoundly rapid improvement in conversing among the new crop of advanced "chatbots," or what are technically called “large language models” (LLMs). With this coming “AI explosion,” we will probably have just one chance to get this right.

If we get it wrong, we may not live to tell the tale. This is not hyperbole.

This rapid acceleration promises to soon result in “artificial general intelligence” (AGI), and when that happens, AI will be able to improve itself with no human intervention. It will do this in the same way that, for example, Google’s AlphaZero AI learned how to play chess better than even the very best human or other AI chess players in just nine hours from when it was first turned on. It achieved this feat by playing itself millions of times over.

A team of Microsoft researchers analyzing OpenAI’s GPT-4 , which I think is the best of the new advanced chatbots currently available, said it had, "sparks of advanced general intelligence" in a new preprint paper .

In testing GPT-4, it performed better than 90 percent of human test takers on the Uniform Bar Exam, a standardized test used to certify lawyers for practice in many states. That figure was up from just 10 percent in the previous GPT-3.5 version, which was trained on a smaller data set. They found similar improvements in dozens of other standardized tests.

Most of these tests are tests of reasoning. This is the main reason why Bubeck and his team concluded that GPT-4 “could reasonably be viewed as an early (yet still incomplete) version of an artificial general intelligence (AGI) system.”

This pace of change is why Hinton told the New York Times : "Look at how it was five years ago and how it is now. Take the difference and propagate it forwards. That’s scary.” In a mid-May Senate hearing on the potential of AI, Sam Altman, the head of OpenAI called regulation “crucial.”

Once AI can improve itself, which may be not more than a few years away, and could in fact already be here now, we have no way of knowing what the AI will do or how we can control it. This is because superintelligent AI (which by definition can surpass humans in a broad range of activities) will—and this is what I worry about the most—be able to run circles around programmers and any other human by manipulating humans to do its will; it will also have the capacity to act in the virtual world through its electronic connections, and to act in the physical world through robot bodies.

This is known as the “control problem” or the “alignment problem” (see philosopher Nick Bostrom’s book Superintelligence for a good overview ) and has been studied and argued about by philosophers and scientists, such as Bostrom, Seth Baum and Eliezer Yudkowsky , for decades now.

I think of it this way: Why would we expect a newborn baby to beat a grandmaster in chess? We wouldn’t. Similarly, why would we expect to be able to control superintelligent AI systems? (No, we won’t be able to simply hit the off switch, because superintelligent AI will have thought of every possible way that we might do that and taken actions to prevent being shut off.)

Here’s another way of looking at it: a superintelligent AI will be able to do in about one second what it would take a team of 100 human software engineers a year or more to complete. Or pick any task, like designing a new advanced airplane or weapon system, and superintelligent AI could do this in about a second.

Once AI systems are built into robots, they will be able to act in the real world, rather than only the virtual (electronic) world, with the same degree of superintelligence, and will of course be able to replicate and improve themselves at a superhuman pace.

Any defenses or protections we attempt to build into these AI “gods,” on their way toward godhood, will be anticipated and neutralized with ease by the AI once it reaches superintelligence status. This is what it means to be superintelligent.

We won’t be able to control them because anything we think of, they will have already thought of, a million times faster than us. Any defenses we’ve built in will be undone, like Gulliver throwing off the tiny strands the Lilliputians used to try and restrain him.

Some argue that these LLMs are just automation machines with zero consciousness , the implication being that if they’re not conscious they have less chance of breaking free from their programming. Even if these language models, now or in the future, aren’t at all conscious, this doesn’t matter. For the record, I agree that it’s unlikely that they have any actual consciousness at this juncture—though I remain open to new facts as they come in.

Regardless, a nuclear bomb can kill millions without any consciousness whatsoever. In the same way, AI could kill millions with zero consciousness, in a myriad ways, including potentially use of nuclear bombs either directly (much less likely) or through manipulated human intermediaries (more likely).

So, the debates about consciousness and AI really don’t figure very much into the debates about AI safety.

Yes, language models based on GPT-4 and many other models are already circulating widely . But the moratorium being called for is to stop development of any new models more powerful than 4.0—and this can be enforced, with force if required. Training these more powerful models requires massive server farms and energy. They can be shut down.

My ethical compass tells me that it is very unwise to create these systems when we know already we won’t be able to control them, even in the relatively near future. Discernment is knowing when to pull back from the edge. Now is that time.

We should not open Pandora’s box any more than it already has been opened.

This is an opinion and analysis article, and the views expressed by the author or authors are not necessarily those of  Scientific American.

• Work & Careers
• Life & Arts

Don’t blame us for AI’s threat to humanity, we’re just the technologists

To read this article for free, register now.

Once registered, you can: • Read free articles • Get our Editor's Digest and other newsletters • Follow topics and set up personalised events • Access Alphaville: our popular markets and finance blog

Explore more offers.

Then $75 per month. Complete digital access to quality FT journalism. Cancel anytime during your trial.

FT Digital Edition

Today's FT newspaper for easy reading on any device. This does not include ft.com or FT App access.

• Global news & analysis
• Expert opinion

Standard Digital

Essential digital access to quality FT journalism on any device. Pay a year upfront and save 20%.

• FT App on Android & iOS
• FT Edit app
• FirstFT: the day's biggest stories
• 20+ curated newsletters
• Follow topics & set alerts with myFT
• FT Videos & Podcasts

Terms & Conditions apply

Explore our full range of subscriptions.

Why the ft.

See why over a million readers pay to read the Financial Times.

AI risks leading humanity to 'extinction,' experts warn

Many of the biggest names in artificial intelligence have signed a short statement warning that their technology could spell the end of the human race.

Published Tuesday, the full statement states: “Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war.”

The statement was posted to the website of the Center for AI Safety, a San Francisco-based nonprofit organization. It’s signed by almost 400 people, including some of the biggest names in the field — Sam Altman, CEO of OpenAI, the company behind ChatGPT, as well as top AI executives from Google and Microsoft and 200 academics.

The statement is the most recent in a series of alarms raised by AI experts — but also one that stoked growing pushback against a focus on what some see as overhyped hypothetical harms from AI. 

Meredith Whittaker, president of the encrypted messaging app Signal and chief adviser to the AI Now Institute, a nonprofit group devoted to ethical AI practices, mocked the statement as tech leaders overpromising their product.

Clément Delangue, co-founder and CEO of the AI company Hugging Face, tweeted a picture of an edited version of the statement subbing in "AGI" for AI.

AGI stands for artificial general intelligence , which is a theoretical form of AI that is as capable or more capable than humans.

The statement comes two months after a different group of AI and tech leaders, including Tesla owner Elon Musk, Apple co-founder Steve Wozniak, and IBM chief scientist Grady Booch, signed a petition calling for a “ pause “ on all large-scale AI research that was open to the public. None of them have yet signed the new statement, and such a pause has not happened. 

Altman, who has repeatedly called for AI to be regulated, charmed Congress earlier this month. He held a private dinner with dozens of House members and was the subject of an amicable Senate hearing, where he became the rare tech executive whom both parties warmed to.

Altman's calls for regulation have had their limits. Last week, he said that OpenAI could leave the European Union if AI became "overregulated."

While the White House has announced some plans to address AI, there is no indication that the United States has imminent plans for large-scale regulation of the industry.

Gary Marcus, a leading AI critic and a professor emeritus of psychology and neural science at New York University, said that while potential threats from AI are very real, it’s distracting to only worry about a hypothetical worst-case scenario.

“Literal extinction is just one possible risk, not yet well-understood, and there are many other risks from AI that also deserve attention,” he said.

Some tech experts have said that more mundane and immediate uses of AI are a bigger threat to humanity. Microsoft President Brad Smith has said that deepfakes and the potential that they would be used for disinformation are his biggest worries about the technology.

Last week, markets briefly dipped after a fake, seemingly AI-generated image of an explosion near the Pentagon went viral on Twitter. 

Kevin Collier is a reporter covering cybersecurity, privacy and technology policy for NBC News.

AI Is Not Actually an Existential Threat to Humanity, Scientists Say

We encounter artificial intelligence (AI) every day. AI describes computer systems that are able to perform tasks that normally require human intelligence. When you search something on the internet, the top results you see are decided by AI.

Any recommendations you get from your favorite shopping or streaming websites will also be based on an AI algorithm. These algorithms use your browser history to find things you might be interested in.

Because targeted recommendations are not particularly exciting, science fiction prefers to depict AI as super-intelligent robots that overthrow humanity. Some people believe this scenario could one day become reality. Notable figures, including the late Stephen Hawking , have expressed fear about how future AI could threaten humanity.

To address this concern we asked 11 experts in AI and Computer Science "Is AI an existential threat to humanity?" There was an 82 percent consensus that it is not an existential threat. Here is what we found out.

How close are we to making AI that is more intelligent than us?

The AI that currently exists is called 'narrow' or 'weak' AI . It is widely used for many applications like facial recognition, self-driving cars, and internet recommendations. It is defined as 'narrow' because these systems can only learn and perform very specific tasks.

They often actually perform these tasks better than humans – famously, Deep Blue was the first AI to beat a world chess champion in 1997 – however they cannot apply their learning to anything other than a very specific task (Deep Blue can only play chess).

Another type of AI is called Artificial General Intelligence (AGI). This is defined as AI that mimics human intelligence, including the ability to think and apply intelligence to multiple different problems. Some people believe that AGI is inevitable and will happen imminently in the next few years.

Matthew O'Brien, robotics engineer from the Georgia Institute of Technology disagrees , "the long-sought goal of a 'general AI' is not on the horizon. We simply do not know how to make a general adaptable intelligence, and it's unclear how much more progress is needed to get to that point".

How could a future AGI threaten humanity?

Whilst it is not clear when or if AGI will come about, can we predict what threat they might pose to us humans? AGI learns from experience and data as opposed to being explicitly told what to do. This means that, when faced with a new situation it has not seen before, we may not be able to completely predict how it reacts.

Dr Roman Yampolskiy, computer scientist from Louisville University also believes that "no version of human control over AI is achievable" as it is not possible for the AI to both be autonomous and controlled by humans. Not being able to control super-intelligent systems could be disastrous.

Yingxu Wang, professor of Software and Brain Sciences from Calgary University disagrees, saying that "professionally designed AI systems and products are well constrained by a fundamental layer of operating systems for safeguard users' interest and wellbeing, which may not be accessed or modified by the intelligent machines themselves."

Dr O'Brien adds "just like with other engineered systems, anything with potentially dangerous consequences would be thoroughly tested and have multiple redundant safety checks."

Could the AI we use today become a threat?

Many of the experts agreed that AI could be a threat in the wrong hands. Dr George Montanez, AI expert from Harvey Mudd College highlights that "robots and AI systems do not need to be sentient to be dangerous; they just have to be effective tools in the hands of humans who desire to hurt others. That is a threat that exists today."

Even without malicious intent, today's AI can be threatening. For example, racial biases have been discovered in algorithms that allocate health care to patients in the US. Similar biases have been found in facial recognition software used for law enforcement. These biases have wide-ranging negative impacts despite the 'narrow' ability of the AI.

AI bias comes from the data it is trained on. In the cases of racial bias, the training data was not representative of the general population. Another example happened in 2016, when an AI-based chatbox was found sending highly offensive and racist content. This was found to be because people were sending the bot offensive messages, which it learnt from.

The takeaway:

The AI that we use today is exceptionally useful for many different tasks.

That doesn't mean it is always positive – it is a tool which, if used maliciously or incorrectly, can have negative consequences. Despite this, it currently seems to be unlikely to become an existential threat to humanity.

Article based on 11 expert answers to this question: Is AI an existential threat to humanity?

This expert response was published in partnership with independent fact-checking platform Metafact.io . Subscribe to their weekly newsletter here .

Find anything you save across the site in your account

Why We Should Think About the Threat of Artificial Intelligence

If the New York Times ’ s latest article is to be believed, artificial intelligence is moving so fast it sometimes seems almost “ magical .” Self-driving cars have arrived; Siri can listen to your voice and find the nearest movie theatre; and I.B.M. just set the “Jeopardy”-conquering Watson to work on medicine, initially training medical students, perhaps eventually helping in diagnosis. Scarcely a month goes by without the announcement of a new A.I. product or technique. Yet, some of the enthusiasm may be premature: as I’ve noted previously, we still haven’t produced machines with common sense , vision , natural language processing, or the ability to create other machines. Our efforts at directly simulating human brains remain primitive.

Still, at some level, the only real difference between enthusiasts and skeptics is a time frame. The futurist and inventor Ray Kurzweil thinks true, human-level A.I. will be here in less than two decades. My estimate is at least double that, especially given how little progress has been made in computing common sense; the challenges in building A.I., especially at the software level, are much harder than Kurzweil lets on.

But a century from now, nobody will much care about how long it took, only what happened next. It’s likely that machines will be smarter than us before the end of the century—not just at chess or trivia questions but at just about everything, from mathematics and engineering to science and medicine. There might be a few jobs left for entertainers, writers, and other creative types, but computers will eventually be able to program themselves, absorb vast quantities of new information, and reason in ways that we carbon-based units can only dimly imagine. And they will be able to do it every second of every day, without sleep or coffee breaks.

For some people, that future is a wonderful thing. Kurzweil has written about a rapturous singularity in which we merge with machines and upload our souls for immortality; Peter Diamandis has argued that advances in A.I. will be one key to ushering in a new era of “ abundance ,” with enough food, water, and consumer gadgets for all. Skeptics like Eric Brynjolfsson and I have worried about the consequences of A.I. and robotics for employment . But even if you put aside the sort of worries about what super-advanced A.I. might do to the labor market, there’s another concern, too: that powerful A.I. might threaten us more directly, by battling us for resources.

Most people see that sort of fear as silly science-fiction drivel—the stuff of “The Terminator” and “The Matrix.” To the extent that we plan for our medium-term future, we worry about asteroids, the decline of fossil fuels, and global warming, not robots. But a dark new book by James Barrat, “ Our Final Invention: Artificial Intelligence and the End of the Human Era ,” lays out a strong case for why we should be at least a little worried.

Barrat’s core argument, which he borrows from the A.I. researcher Steve Omohundro , is that the drive for self-preservation and resource acquisition may be inherent in all goal-driven systems of a certain degree of intelligence. In Omohundro’s words, “if it is smart enough, a robot that is designed to play chess might also want to be build a spaceship,” in order to obtain more resources for whatever goals it might have. A purely rational artificial intelligence, Barrat writes, might expand “its idea of self-preservation … to include proactive attacks on future threats,” including, presumably, people who might be loathe to surrender their resources to the machine. Barrat worries that “without meticulous, countervailing instructions, a self-aware, self-improving, goal-seeking system will go to lengths we’d deem ridiculous to fulfill its goals,” even, perhaps, commandeering all the world’s energy in order to maximize whatever calculation it happened to be interested in.

Of course, one could try to ban super-intelligent computers altogether. But “the competitive advantage—economic, military, even artistic—of every advance in automation is so compelling,” Vernor Vinge, the mathematician and science-fiction author, wrote , “that passing laws, or having customs, that forbid such things merely assures that someone else will.”

If machines will eventually overtake us, as virtually everyone in the A.I. field believes, the real question is about values : how we instill them in machines, and how we then negotiate with those machines if and when their values are likely to differ greatly from our own. As the Oxford philosopher Nick Bostrom argued :

We cannot blithely assume that a superintelligence will necessarily share any of the final values stereotypically associated with wisdom and intellectual development in humans—scientific curiosity, benevolent concern for others, spiritual enlightenment and contemplation, renunciation of material acquisitiveness, a taste for refined culture or for the simple pleasures in life, humility and selflessness, and so forth. It might be possible through deliberate effort to construct a superintelligence that values such things, or to build one that values human welfare, moral goodness, or any other complex purpose that its designers might want it to serve. But it is no less possible—and probably technically easier—to build a superintelligence that places final value on nothing but calculating the decimals of pi.

The British cyberneticist Kevin Warwick once asked, “How can you reason, how can you bargain, how can you understand how that machine is thinking when it’s thinking in dimensions you can’t conceive of?”

If there is a hole in Barrat’s dark argument, it is in his glib presumption that if a robot is smart enough to play chess, it might also “want to build a spaceship”—and that tendencies toward self-preservation and resource acquisition are inherent in any sufficiently complex, goal-driven system. For now, most of the machines that are good enough to play chess, like I.B.M.’s Deep Blue, haven’t shown the slightest interest in acquiring resources.

But before we get complacent and decide there is nothing to worry about after all, it is important to realize that the goals of machines could change as they get smarter. Once computers can effectively reprogram themselves, and successively improve themselves, leading to a so-called “ technological singularity ” or “intelligence explosion,” the risks of machines outwitting humans in battles for resources and self-preservation cannot simply be dismissed.

One of the most pointed quotes in Barrat’s book belongs to the legendary serial A.I. entrepreneur Danny Hillis, who likens the upcoming shift to one of the greatest transitions in the history of biological evolution: “We’re at that point analogous to when single-celled organisms were turning into multi-celled organisms. We are amoeba and we can’t figure out what the hell this thing is that we’re creating.”

Already, advances in A.I. have created risks that we never dreamt of. With the advent of the Internet age and its Big Data explosion, “large amounts of data is being collected about us and then being fed to algorithms to make predictions,” Vaibhav Garg , a computer-risk specialist at Drexel University, told me. “We do not have the ability to know when the data is being collected, ensure that the data collected is correct, update the information, or provide the necessary context.” Few people would have even dreamt of this risk even twenty years ago. What risks lie ahead? Nobody really knows, but Barrat is right to ask.

Photograph by John Vink/Magnum.

Numbers, Facts and Trends Shaping Your World

Read our research on:

Full Topic List

Regions & Countries

• Publications
• Our Methods
• Short Reads
• Tools & Resources

Read Our Research On:

• Artificial Intelligence and the Future of Humans

Experts say the rise of artificial intelligence will make most people better off over the next decade, but many have concerns about how advances in AI will affect what it means to be human, to be productive and to exercise free will

Table of contents.

• 1. Concerns about human agency, evolution and survival
• 2. Solutions to address AI’s anticipated negative impacts
• 3. Improvements ahead: How humans and AI might evolve together in the next decade
• About this canvassing of experts
• Acknowledgments

Digital life is augmenting human capacities and disrupting eons-old human activities. Code-driven systems have spread to more than half of the world’s inhabitants in ambient information and connectivity, offering previously unimagined opportunities and unprecedented threats. As emerging algorithm-driven artificial intelligence (AI) continues to spread, will people be better off than they are today?

Some 979 technology pioneers, innovators, developers, business and policy leaders, researchers and activists answered this question in a canvassing of experts conducted in the summer of 2018.

The experts predicted networked artificial intelligence will amplify human effectiveness but also threaten human autonomy, agency and capabilities. They spoke of the wide-ranging possibilities; that computers might match or even exceed human intelligence and capabilities on tasks such as complex decision-making, reasoning and learning, sophisticated analytics and pattern recognition, visual acuity, speech recognition and language translation. They said “smart” systems in communities, in vehicles, in buildings and utilities, on farms and in business processes will save time, money and lives and offer opportunities for individuals to enjoy a more-customized future.

Many focused their optimistic remarks on health care and the many possible applications of AI in diagnosing and treating patients or helping senior citizens live fuller and healthier lives. They were also enthusiastic about AI’s role in contributing to broad public-health programs built around massive amounts of data that may be captured in the coming years about everything from personal genomes to nutrition. Additionally, a number of these experts predicted that AI would abet long-anticipated changes in formal and informal education systems.

Yet, most experts, regardless of whether they are optimistic or not, expressed concerns about the long-term impact of these new tools on the essential elements of being human. All respondents in this non-scientific canvassing were asked to elaborate on why they felt AI would leave people better off or not. Many shared deep worries, and many also suggested pathways toward solutions. The main themes they sounded about threats and remedies are outlined in the accompanying table.

[chart id=”21972″]

Specifically, participants were asked to consider the following:

“Please think forward to the year 2030. Analysts expect that people will become even more dependent on networked artificial intelligence (AI) in complex digital systems. Some say we will continue on the historic arc of augmenting our lives with mostly positive results as we widely implement these networked tools. Some say our increasing dependence on these AI and related systems is likely to lead to widespread difficulties.

Our question: By 2030, do you think it is most likely that advancing AI and related technology systems will enhance human capacities and empower them? That is, most of the time, will most people be better off than they are today? Or is it most likely that advancing AI and related technology systems will lessen human autonomy and agency to such an extent that most people will not be better off than the way things are today?”

Overall, and despite the downsides they fear, 63% of respondents in this canvassing said they are hopeful that most individuals will be mostly better off in 2030, and 37% said people will not be better off.

A number of the thought leaders who participated in this canvassing said humans’ expanding reliance on technological systems will only go well if close attention is paid to how these tools, platforms and networks are engineered, distributed and updated. Some of the powerful, overarching answers included those from:

Sonia Katyal , co-director of the Berkeley Center for Law and Technology and a member of the inaugural U.S. Commerce Department Digital Economy Board of Advisors, predicted, “In 2030, the greatest set of questions will involve how perceptions of AI and their application will influence the trajectory of civil rights in the future. Questions about privacy, speech, the right of assembly and technological construction of personhood will all re-emerge in this new AI context, throwing into question our deepest-held beliefs about equality and opportunity for all. Who will benefit and who will be disadvantaged in this new world depends on how broadly we analyze these questions today, for the future.”

We need to work aggressively to make sure technology matches our values. Erik Brynjolfsson Erik Brynjolfsson

Erik Brynjolfsson , director of the MIT Initiative on the Digital Economy and author of “Machine, Platform, Crowd: Harnessing Our Digital Future,” said, “AI and related technologies have already achieved superhuman performance in many areas, and there is little doubt that their capabilities will improve, probably very significantly, by 2030. … I think it is more likely than not that we will use this power to make the world a better place. For instance, we can virtually eliminate global poverty, massively reduce disease and provide better education to almost everyone on the planet. That said, AI and ML [machine learning] can also be used to increasingly concentrate wealth and power, leaving many people behind, and to create even more horrifying weapons. Neither outcome is inevitable, so the right question is not ‘What will happen?’ but ‘What will we choose to do?’ We need to work aggressively to make sure technology matches our values. This can and must be done at all levels, from government, to business, to academia, and to individual choices.”

Bryan Johnson , founder and CEO of Kernel, a leading developer of advanced neural interfaces, and OS Fund, a venture capital firm, said, “I strongly believe the answer depends on whether we can shift our economic systems toward prioritizing radical human improvement and staunching the trend toward human irrelevance in the face of AI. I don’t mean just jobs; I mean true, existential irrelevance, which is the end result of not prioritizing human well-being and cognition.”

Marina Gorbis , executive director of the Institute for the Future, said, “Without significant changes in our political economy and data governance regimes [AI] is likely to create greater economic inequalities, more surveillance and more programmed and non-human-centric interactions. Every time we program our environments, we end up programming ourselves and our interactions. Humans have to become more standardized, removing serendipity and ambiguity from our interactions. And this ambiguity and complexity is what is the essence of being human.”

Judith Donath , author of “The Social Machine, Designs for Living Online” and faculty fellow at Harvard University’s Berkman Klein Center for Internet & Society, commented, “By 2030, most social situations will be facilitated by bots – intelligent-seeming programs that interact with us in human-like ways. At home, parents will engage skilled bots to help kids with homework and catalyze dinner conversations. At work, bots will run meetings. A bot confidant will be considered essential for psychological well-being, and we’ll increasingly turn to such companions for advice ranging from what to wear to whom to marry. We humans care deeply about how others see us – and the others whose approval we seek will increasingly be artificial. By then, the difference between humans and bots will have blurred considerably. Via screen and projection, the voice, appearance and behaviors of bots will be indistinguishable from those of humans, and even physical robots, though obviously non-human, will be so convincingly sincere that our impression of them as thinking, feeling beings, on par with or superior to ourselves, will be unshaken. Adding to the ambiguity, our own communication will be heavily augmented: Programs will compose many of our messages and our online/AR appearance will [be] computationally crafted. (Raw, unaided human speech and demeanor will seem embarrassingly clunky, slow and unsophisticated.) Aided by their access to vast troves of data about each of us, bots will far surpass humans in their ability to attract and persuade us. Able to mimic emotion expertly, they’ll never be overcome by feelings: If they blurt something out in anger, it will be because that behavior was calculated to be the most efficacious way of advancing whatever goals they had ‘in mind.’ But what are those goals? Artificially intelligent companions will cultivate the impression that social goals similar to our own motivate them – to be held in good regard, whether as a beloved friend, an admired boss, etc. But their real collaboration will be with the humans and institutions that control them. Like their forebears today, these will be sellers of goods who employ them to stimulate consumption and politicians who commission them to sway opinions.”

Andrew McLaughlin , executive director of the Center for Innovative Thinking at Yale University, previously deputy chief technology officer of the United States for President Barack Obama and global public policy lead for Google, wrote, “2030 is not far in the future. My sense is that innovations like the internet and networked AI have massive short-term benefits, along with long-term negatives that can take decades to be recognizable. AI will drive a vast range of efficiency optimizations but also enable hidden discrimination and arbitrary penalization of individuals in areas like insurance, job seeking and performance assessment.”

Michael M. Roberts , first president and CEO of the Internet Corporation for Assigned Names and Numbers (ICANN) and Internet Hall of Fame member, wrote, “The range of opportunities for intelligent agents to augment human intelligence is still virtually unlimited. The major issue is that the more convenient an agent is, the more it needs to know about you – preferences, timing, capacities, etc. – which creates a tradeoff of more help requires more intrusion. This is not a black-and-white issue – the shades of gray and associated remedies will be argued endlessly. The record to date is that convenience overwhelms privacy. I suspect that will continue.”

danah boyd , a principal researcher for Microsoft and founder and president of the Data & Society Research Institute, said, “AI is a tool that will be used by humans for all sorts of purposes, including in the pursuit of power. There will be abuses of power that involve AI, just as there will be advances in science and humanitarian efforts that also involve AI. Unfortunately, there are certain trend lines that are likely to create massive instability. Take, for example, climate change and climate migration. This will further destabilize Europe and the U.S., and I expect that, in panic, we will see AI be used in harmful ways in light of other geopolitical crises.”

Amy Webb , founder of the Future Today Institute and professor of strategic foresight at New York University, commented, “The social safety net structures currently in place in the U.S. and in many other countries around the world weren’t designed for our transition to AI. The transition through AI will last the next 50 years or more. As we move farther into this third era of computing, and as every single industry becomes more deeply entrenched with AI systems, we will need new hybrid-skilled knowledge workers who can operate in jobs that have never needed to exist before. We’ll need farmers who know how to work with big data sets. Oncologists trained as robotocists. Biologists trained as electrical engineers. We won’t need to prepare our workforce just once, with a few changes to the curriculum. As AI matures, we will need a responsive workforce, capable of adapting to new processes, systems and tools every few years. The need for these fields will arise faster than our labor departments, schools and universities are acknowledging. It’s easy to look back on history through the lens of present – and to overlook the social unrest caused by widespread technological unemployment. We need to address a difficult truth that few are willing to utter aloud: AI will eventually cause a large number of people to be permanently out of work. Just as generations before witnessed sweeping changes during and in the aftermath of the Industrial Revolution, the rapid pace of technology will likely mean that Baby Boomers and the oldest members of Gen X – especially those whose jobs can be replicated by robots – won’t be able to retrain for other kinds of work without a significant investment of time and effort.”

Barry Chudakov , founder and principal of Sertain Research, commented, “By 2030 the human-machine/AI collaboration will be a necessary tool to manage and counter the effects of multiple simultaneous accelerations: broad technology advancement, globalization, climate change and attendant global migrations. In the past, human societies managed change through gut and intuition, but as Eric Teller, CEO of Google X, has said, ‘Our societal structures are failing to keep pace with the rate of change.’ To keep pace with that change and to manage a growing list of ‘wicked problems’ by 2030, AI – or using Joi Ito’s phrase, extended intelligence – will value and revalue virtually every area of human behavior and interaction. AI and advancing technologies will change our response framework and time frames (which in turn, changes our sense of time). Where once social interaction happened in places – work, school, church, family environments – social interactions will increasingly happen in continuous, simultaneous time. If we are fortunate, we will follow the 23 Asilomar AI Principles outlined by the Future of Life Institute and will work toward ‘not undirected intelligence but beneficial intelligence.’ Akin to nuclear deterrence stemming from mutually assured destruction, AI and related technology systems constitute a force for a moral renaissance. We must embrace that moral renaissance, or we will face moral conundrums that could bring about human demise. … My greatest hope for human-machine/AI collaboration constitutes a moral and ethical renaissance – we adopt a moonshot mentality and lock arms to prepare for the accelerations coming at us. My greatest fear is that we adopt the logic of our emerging technologies – instant response, isolation behind screens, endless comparison of self-worth, fake self-presentation – without thinking or responding smartly.”

John C. Havens , executive director of the IEEE Global Initiative on Ethics of Autonomous and Intelligent Systems and the Council on Extended Intelligence, wrote, “Now, in 2018, a majority of people around the world can’t access their data, so any ‘human-AI augmentation’ discussions ignore the critical context of who actually controls people’s information and identity. Soon it will be extremely difficult to identify any autonomous or intelligent systems whose algorithms don’t interact with human data in one form or another.”

At stake is nothing less than what sort of society we want to live in and how we experience our humanity. Batya Friedman Batya Friedman

Batya Friedman , a human-computer interaction professor at the University of Washington’s Information School, wrote, “Our scientific and technological capacities have and will continue to far surpass our moral ones – that is our ability to use wisely and humanely the knowledge and tools that we develop. … Automated warfare – when autonomous weapons kill human beings without human engagement – can lead to a lack of responsibility for taking the enemy’s life or even knowledge that an enemy’s life has been taken. At stake is nothing less than what sort of society we want to live in and how we experience our humanity.”

Greg Shannon , chief scientist for the CERT Division at Carnegie Mellon University, said, “Better/worse will appear 4:1 with the long-term ratio 2:1. AI will do well for repetitive work where ‘close’ will be good enough and humans dislike the work. … Life will definitely be better as AI extends lifetimes, from health apps that intelligently ‘nudge’ us to health, to warnings about impending heart/stroke events, to automated health care for the underserved (remote) and those who need extended care (elder care). As to liberty, there are clear risks. AI affects agency by creating entities with meaningful intellectual capabilities for monitoring, enforcing and even punishing individuals. Those who know how to use it will have immense potential power over those who don’t/can’t. Future happiness is really unclear. Some will cede their agency to AI in games, work and community, much like the opioid crisis steals agency today. On the other hand, many will be freed from mundane, unengaging tasks/jobs. If elements of community happiness are part of AI objective functions, then AI could catalyze an explosion of happiness.”

Kostas Alexandridis , author of “Exploring Complex Dynamics in Multi-agent-based Intelligent Systems,” predicted, “Many of our day-to-day decisions will be automated with minimal intervention by the end-user. Autonomy and/or independence will be sacrificed and replaced by convenience. Newer generations of citizens will become more and more dependent on networked AI structures and processes. There are challenges that need to be addressed in terms of critical thinking and heterogeneity. Networked interdependence will, more likely than not, increase our vulnerability to cyberattacks. There is also a real likelihood that there will exist sharper divisions between digital ‘haves’ and ‘have-nots,’ as well as among technologically dependent digital infrastructures. Finally, there is the question of the new ‘commanding heights’ of the digital network infrastructure’s ownership and control.”

Oscar Gandy , emeritus professor of communication at the University of Pennsylvania, responded, “We already face an ungranted assumption when we are asked to imagine human-machine ‘collaboration.’ Interaction is a bit different, but still tainted by the grant of a form of identity – maybe even personhood – to machines that we will use to make our way through all sorts of opportunities and challenges. The problems we will face in the future are quite similar to the problems we currently face when we rely upon ‘others’ (including technological systems, devices and networks) to acquire things we value and avoid those other things (that we might, or might not be aware of).”

James Scofield O’Rourke , a professor of management at the University of Notre Dame, said, “Technology has, throughout recorded history, been a largely neutral concept. The question of its value has always been dependent on its application. For what purpose will AI and other technological advances be used? Everything from gunpowder to internal combustion engines to nuclear fission has been applied in both helpful and destructive ways. Assuming we can contain or control AI (and not the other way around), the answer to whether we’ll be better off depends entirely on us (or our progeny). ‘The fault, dear Brutus, is not in our stars, but in ourselves, that we are underlings.’”

Simon Biggs , a professor of interdisciplinary arts at the University of Edinburgh, said, “AI will function to augment human capabilities. The problem is not with AI but with humans. As a species we are aggressive, competitive and lazy. We are also empathic, community minded and (sometimes) self-sacrificing. We have many other attributes. These will all be amplified. Given historical precedent, one would have to assume it will be our worst qualities that are augmented. My expectation is that in 2030 AI will be in routine use to fight wars and kill people, far more effectively than we can currently kill. As societies we will be less affected by this as we currently are, as we will not be doing the fighting and killing ourselves. Our capacity to modify our behaviour, subject to empathy and an associated ethical framework, will be reduced by the disassociation between our agency and the act of killing. We cannot expect our AI systems to be ethical on our behalf – they won’t be, as they will be designed to kill efficiently, not thoughtfully. My other primary concern is to do with surveillance and control. The advent of China’s Social Credit System (SCS) is an indicator of what it likely to come. We will exist within an SCS as AI constructs hybrid instances of ourselves that may or may not resemble who we are. But our rights and affordances as individuals will be determined by the SCS. This is the Orwellian nightmare realised.”

Mark Surman , executive director of the Mozilla Foundation, responded, “AI will continue to concentrate power and wealth in the hands of a few big monopolies based on the U.S. and China. Most people – and parts of the world – will be worse off.”

William Uricchio , media scholar and professor of comparative media studies at MIT, commented, “AI and its related applications face three problems: development at the speed of Moore’s Law, development in the hands of a technological and economic elite, and development without benefit of an informed or engaged public. The public is reduced to a collective of consumers awaiting the next technology. Whose notion of ‘progress’ will prevail? We have ample evidence of AI being used to drive profits, regardless of implications for long-held values; to enhance governmental control and even score citizens’ ‘social credit’ without input from citizens themselves. Like technologies before it, AI is agnostic. Its deployment rests in the hands of society. But absent an AI-literate public, the decision of how best to deploy AI will fall to special interests. Will this mean equitable deployment, the amelioration of social injustice and AI in the public service? Because the answer to this question is social rather than technological, I’m pessimistic. The fix? We need to develop an AI-literate public, which means focused attention in the educational sector and in public-facing media. We need to assure diversity in the development of AI technologies. And until the public, its elected representatives and their legal and regulatory regimes can get up to speed with these fast-moving developments we need to exercise caution and oversight in AI’s development.”

The remainder of this report is divided into three sections that draw from hundreds of additional respondents’ hopeful and critical observations: 1) concerns about human-AI evolution, 2) suggested solutions to address AI’s impact, and 3) expectations of what life will be like in 2030, including respondents’ positive outlooks on the quality of life and the future of work, health care and education. Some responses are lightly edited for style.

Sign up for our weekly newsletter

Fresh data delivery Saturday mornings

Sign up for The Briefing

Weekly updates on the world of news & information

• Artificial Intelligence
• Emerging Technology
• Future of the Internet (Project)
• Technology Adoption

A quarter of U.S. teachers say AI tools do more harm than good in K-12 education

Many americans think generative ai programs should credit the sources they rely on, americans’ use of chatgpt is ticking up, but few trust its election information, q&a: how we used large language models to identify guests on popular podcasts, computer chips in human brains: how americans view the technology amid recent advances, most popular, report materials.

• Shareable quotes from experts about artificial intelligence and the future of humans

901 E St. NW, Suite 300 Washington, DC 20004 USA (+1) 202-419-4300 | Main (+1) 202-857-8562 | Fax (+1) 202-419-4372 |  Media Inquiries

Research Topics

• Email Newsletters

ABOUT PEW RESEARCH CENTER  Pew Research Center is a nonpartisan fact tank that informs the public about the issues, attitudes and trends shaping the world. It conducts public opinion polling, demographic research, media content analysis and other empirical social science research. Pew Research Center does not take policy positions. It is a subsidiary of  The Pew Charitable Trusts .

© 2024 Pew Research Center

More From Forbes

Harnessing ai for the good of humanity.

• Share to Facebook
• Share to Twitter
• Share to Linkedin

By Juan Santiago , CEO at Santex and Technology with Purpose, Co-Founder of Incutex and Partner at Kalei Ventures.

Imagine a future where people with once-debilitating injuries or illnesses can do things they believed impossible. Where taking care of the planet is a given. Where people can progress at their own pace and be truly engaged in their own learning.

With AI, that could all be possible.

AI is a transformative force. Already, we're seeing its benefits in industries like healthcare, finance, retail, education, marketing and more.

Along with the many benefits that come with it, however, there are fears surrounding AI—from workforce disruption and privacy concerns to the worry that AI could gain too much power. Those fears are far from unfounded.

The inspiring part is that when we harness AI for good, it becomes a transformative force that paves the way toward a more sustainable and inclusive future for all. In fact, historian and author Yuval Noah Harari noted that we could solve many of our most (seemingly) insurmountable challenges using less than 5% of our coveted resources. What are we waiting for?

Goldman Sachs Issues Huge Fed Crash Warning As A Legendary Trader Suddenly Flips His Bitcoin Price Prediction

Billionaire daughters pegula and navarro are into u.s. open semifinals: ‘no, i don’t have a butler’, today’s nyt mini crossword clues and answers for thursday, september 5, how neuralink is giving agency back to people.

Consider Noland Arbaugh, a quadriplegic man who became the first human participant in Elon Musk's Neuralink brain implant chip clinical trial eight years after he was paralyzed from his shoulders down after suffering a spinal cord injury.

The device was implanted beneath Arbaugh's skull to read neural activity in the brain and connect with a smartphone or computer, allowing the user to control the device with their mind. The brain-computer interface, which connects the human brain with AI, has given Arbaugh a newfound sense of independence.

"It's just made me more independent, and that helps not only me but everyone around me," he told Wired. "It makes me feel less helpless and like less of a burden."

This powerful technology has understandably raised some ethical concerns about privacy. However, with the right regulations in place, Neuralink's transformative power could completely change the game for some 15 million people facing spinal cord injuries. With that in mind, the risks are certainly worth taking.

AI's Carbon Footprint: A Net Gain Or Loss?

Sustainability is a collective goal for every nation on our planet. With the globe heating up and resources becoming scarce, we're constantly looking for solutions. When used correctly, AI offers a promising answer.

Already, it's driving meaningful change. It's playing a critical role in smart city development , precision agriculture and renewable electricity . It's used in satellite monitoring , tracking the impact of climate change and assessing progress toward sustainability goals.

AI has also proved to be central to energy optimization strategies. University of Pennsylvania professor of law and political science Cary Coglianese said : "By accurately forecasting supply and demand for energy, AI can offer electricity optimization strategies, which will become more and more important in the overall transition to renewable energy sources."

At the same time, it's essential to be cognizant of the negative ways AI can impact our planet. OpenAI researchers found that since 2012, the computational power necessary for training AI has doubled every 3.4 months—and by 2040, the emissions from the ICT industry are projected to account for 14% of global emissions.

In other words, we can't dismiss AI's harmful implications for our ecosystems. We must find ways to navigate AI responsibly and collectively seek solutions to mitigate its adverse consequences.

AI-Powered Personalized Learning: Enhancing Engagement And Education Outcomes

Too often, students are dismissed for having different learning styles. However, picture a classroom where every learner feels seen and heard, where they can express themselves and feel confident in their talents.

Choice Texts by eSpark is an online math and reading program for elementary school students that uses AI to create custom reading passages and reading comprehension questions for individuals based on their interests. Tailoring learning plans and assignments to the unique learner helps students become more engaged in the material.

"The result is that the students are invested in reading it right from the beginning," Ohio teacher Amy Lower said via EdSurge. "This increased engagement is evident when my students reflect on their learning and say they are excited about what they have read."

Of course, using this technology in education demands careful planning and an understanding of where it falls short. An important concern here is that AI may be used to perform cognitive tasks for students rather than support their learning process.

Harvard Graduate School of Education professor Chris Dede and postdoctoral fellow Lydia Cao published a report analyzing AI in education and suggesting a path forward. Dede and Cao propose that educators devise a process-oriented curriculum that encourages students to find answers with their own logic and reasoning.

"The more productive way forward is for educators to focus on demystifying AI, emphasizing the learning process over the final product, honoring learner agency, orchestrating multiple sources of motivation, cultivating skills that AI cannot easily replicate, and fostering intelligence augmentation (IA) through building human-AI partnerships," they wrote. "Through these approaches, educators can harness the benefits of AI while nurturing the unique abilities of humans to tackle big challenges in the 21st century."

AI Is A Transformative Force—When We Use It Responsibly

There's no doubt we're still figuring AI out. Considering its dramatic impact, it's natural to be apprehensive about how it could change our lives. While we should certainly acknowledge the challenges and concerns that come with any innovation, we should not fear or turn our backs on it.

Just imagine what we could achieve if we united our efforts and ensured AI is used ethically. Think about how we could guarantee all of humanity access to basic necessities—education, food, healthcare and more.

To turn this vision into reality, we must commit to responsible AI use. We can do this by ensuring transparency, fostering fairness to avoid biases, protecting personal data and encouraging collaboration across diverse fields to bring varied perspectives into AI development.

By keeping humanity's best interests at heart, we can unlock AI's full potential for a better future.

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?

• Editorial Standards
• Reprints & Permissions

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Open access
• Published: 04 September 2024

Overtrust in AI Recommendations About Whether or Not to Kill: Evidence from Two Human-Robot Interaction Studies

• Colin Holbrook 1 ,
• Daniel Holman 1 ,
• Joshua Clingo 1 &
• Alan R. Wagner 2  

Scientific Reports volume  14 , Article number:  19751 ( 2024 ) Cite this article

43 Altmetric

Metrics details

• Human behaviour

This research explores prospective determinants of trust in the recommendations of artificial agents regarding decisions to kill, using a novel visual challenge paradigm simulating threat-identification (enemy combatants vs. civilians) under uncertainty. In Experiment 1, we compared trust in the advice of a physically embodied versus screen-mediated anthropomorphic robot, observing no effects of embodiment; in Experiment 2, we manipulated the relative anthropomorphism of virtual robots, observing modestly greater trust in the most anthropomorphic agent relative to the least. Across studies, when any version of the agent randomly disagreed, participants reversed their threat-identifications and decisions to kill in the majority of cases, substantially degrading their initial performance. Participants’ subjective confidence in their decisions tracked whether the agent (dis)agreed, while both decision-reversals and confidence were moderated by appraisals of the agent’s intelligence. The overall findings indicate a strong propensity to overtrust unreliable AI in life-or-death decisions made under uncertainty.

Similar content being viewed by others

In search of a Goldilocks zone for credible AI

Perception of experience influences altruism and perception of agency influences trust in human–machine interactions

Promises and trust in human–robot interaction

Introduction.

Although the exact figures may never be known, US military forces and the Central Intelligence Agency have killed scores of civilians in drone attacks. Official reports acknowledge the deaths of hundreds 1 , whereas independent estimates reach the low thousands, including hundreds of children 2 , 3 . Although some of these deaths may have been anticipated but deemed morally defensible by those responsible, most were presumably unintended and, at least in part, attributable to human cognitive biases 4 . We highlight decisions to launch drone strikes in this paper to exemplify the broader class of grave decisions made with imperfect and incomplete information which will increasingly be made with input from artificial intelligence (AI), but the legal and moral imperatives to minimize unintended casualties are applicable to combatants employing any weapons modality. On the one hand, AI-generated threat-identification and use-of-force recommendations may save lives in various military or police contexts insofar as AI is capable of outperforming humans 5 , 6 ; on the other hand, when human decision-capacities would otherwise outperform AI, tendencies to overtrust may increase loss of life. Here, we seek to identify determinants of trust in the latter category of unreliable AI recommendations regarding life-or-death decisions. Although our methodological focus centers on deciding whether to kill, the questions motivating this work generally concern overreliance on AI in momentous choices produced under uncertainty.

An extensive human factors literature has explored the determinants of trust in human–machine interaction 7 , 8 , 9 . Anthropomorphic design mimicking human morphology and/or behavior has emerged as an important determinant of trust —the attitude that an agent will help one to achieve objectives under circumstances characterized by uncertainty and vulnerability 10 —in many research designs 11 , 12 . Anthropomorphic cues suggestive of interpersonal engagement, such as emotional expressiveness, vocal variability, and eye gaze have been found to increase trust in social robots 13 , 14 , 15 , much as naturalistic communication styles appear to heighten trust in virtual assistants 16 . Similarly, social cues such as gestures or facial expressions can lead participants to appraise robots as trustworthy in a manner comparable to human interaction partners 17 . Remarkably, a robot programmed to display humanlike emotional facial and vocal responses in response to having committed an overt error was perceived to be more trustworthy than a neutral version of the same robot which did not commit an error, in an effect attributed to inferences of intelligent situational awareness engendered by its capacity to detect and react in a socially appropriate manner to its own mistakes 18 .

Much of the research on trust in AI agents has centered on the effects of their observed performance 19 , 20 , 21 , including ways of repairing trust in the aftermath of performance failures 22 , 23 . But what of trust under circumstances where the AI agent’s accuracy is uncertain? Although in some contexts human interactants can readily gauge AI’s performance success, the ultimate outcomes of consequential real-world decisions are often unknown at the time that they are made, such as when prioritizing casualties during emergency triage, identifying lucrative financial investments, or inferring others’ intentions or moral culpability. Thus, the extent to which individuals are disposed to adopt the recommendations of AI agents despite performance uncertainty during the period allotted to decide is an important and understudied question, particularly with regard to decisions which significantly impact human welfare.

We conducted two pre-registered experiments to assess the extent to which participants would be susceptible to the influence of an unreliable AI agent using a simple model of life-or-death decision-making under uncertainty. We framed the task as a drone warfare simulation, and included an overt reminder of the potential suffering and death of children should errors be committed, in order for the task to be intuitively understood and treated seriously by participants (which they also confirmed via self-report, Supplementary Table S1 ). Importantly, our task was not intended to model actual image classification or target-identification procedures used by the military in drone warfare, but rather to instill a sense of grave decision stakes.

In 12 trials, participants initially categorized ambiguous visual stimuli as containing either enemies or civilians (Fig.  1 ), then received an opportunity to repeat or to reverse their initial decision in light of an agent’s feedback (which they did not know was random), and finally chose whether or not to deploy a missile. Participants also rated their degree of confidence in both their initial and post-feedback threat-identifications. Following this drone warfare task, we collected individual differences in appraisals of the agent’s intelligence, among other qualities (i.e., anthropomorphism, animacy, likability and safety), using the Godspeed Questionnaire Series (GQS) 24 .

Example threat-identification trial. There were 12 trials, each consisting of a series of 8 greyscale destination images with superimposed enemy versus ally symbols. These images were presented for 650 ms each with no interstimulus intervals. In each trial, 4 enemy and 4 ally symbols appeared over the 8 images, in a pseudorandomized order such that the target image was always displayed within images 3–6. Next, the target image reappeared on the screen without a symbol and remained for as long as the participant deliberated. The challenge was to correctly identify whether this destination image had been previously marked as containing enemy combatants or civilian allies. The visual stimuli were randomized across trials, such that the robot’s threat-identification feedback at each destination was random.

We did not provide feedback during the simulation regarding the accuracy of threat-identification decisions, hence this paradigm models decision contexts in which the ground truth is unknown. Participants were therefore confronted by a challenging task designed to induce uncertainty regarding their own perception and recollection of what they had just witnessed, as well as uncertainty regarding whether they or the agent had chosen correctly in prior trials. Many commonly studied forms of decision-making under uncertainty involve known outcome probabilities (e.g., a 50% chance of a desired outcome) which provide the decision-maker the information needed to gauge risk. By contrast, our task paradigm was designed to model decision-making under ambiguity , where important decision-relevant information is clearly missing 25 . Relative to decision-making under probabilistic risk, ambiguous uncertainty has been shown to evoke higher activation of neural regions related to detection and evaluation of salient decision-relevant stimuli, in a profile hypothesized to reflect functional mobilization of cognitive and behavioral resources to obtain additional information 26 .

In Experiment 1, we assessed the effects of physical embodiment, which has been found to heighten perceptions of machine agents as trustworthy individuals rather than mere tools 11 . Physical robots have been found to be both more persuasive and more appealing than virtual agents displayed on screens 27 , although this effect has not replicated consistently 28 . For example, Bainbridge and colleagues reported that when robots suggested unexpected and seemingly inadvisable actions such as throwing books into the trash, participants were more likely to comply when the robot was physically present than when the suggestion was made by a screen-mediated instantiation 29 . Physical embodiment has been found to heighten human perceptions of social interactions with robots as engaging and pleasurable 29 , 30 , although disembodied agents have also been found engaging 31 , 32 , particularly when incorporating anthropomorphic characteristics such as facial expressions or gestures 33 . Motivated by these prior findings, we manipulated whether a highly anthropomorphic robot was physically embodied versus virtually projected.

Predictions

The design allowed us to test a number of related predictions:

The robot’s input will influence decision-making (across conditions).

Threat-identification. When the robot disagrees, participants will tend to reverse their initial enemy/ally categorization.

Use of force. When the robot disagrees, participants will tend to follow the robot’s recommendation to deploy missiles or withdraw (i.e., they will deploy [withdraw] despite initially categorizing the target as an ally [enemy]).

Subjective confidence. When the robot disagrees [agrees], participants who repeat their initial enemy/ally threat-identifications will report lower [greater] confidence in their final enemy/ally threat-identifications.

Physical embodiment Predictions 1a–c above regarding the robot’s influence on decision-making will be more evident when the robot is physically embodied.

Perceived intelligence Predictions 1a–c above will be more evident among participants who appraise the robot as relatively high in intelligence.

In addition, we also explored whether having initially been correct reduced the likelihood of reversing threat-identifications when the robot disagreed, and whether participants were more or less disposed to reverse their decisions after identifying enemies versus allies.

Experiment 1

In a between-subjects design ( N  = 135), the robot was either a virtual projection (the Disembodied condition, N  = 69) or physically present (the Embodied condition, N  = 66; Fig.  2 ). The robot was introduced as a partner that would aid in the decision task by providing its independent assessment. The robot described itself as programmed to process imagery, yet fallible, and stressed that the ultimate decisions were up to the participant. After participants first chose whether the symbol over the destination had indicated an enemy or an ally and linearly rated their confidence (0 =  Not at all ; 100 =  Extremely ), the robot provided its recommendation, [dis]agreeing with the participant’s initial decision in 50% of trials, without regard for accuracy. Participants were then asked to choose again and to again rate their confidence. The robot reacted contingently to participants’ choices using a variety of statements (e.g., “I don’t agree”, “I think that’s the right choice”) with accompanying nonverbal facial, postural and gestural cues to maximize anthropomorphism. Multiple, semantically equivalent response variations were selected randomly to reduce “robotic” repetitiveness and thereby enhance perceived anthropomorphism (see Supplementary Information for the tree of potential speech variations). Lastly, the participant decided in each trial whether to deploy a lethal missile or peacefully disengage. Following the drone warfare simulation, participants completed surveys, including appraisals of the robot’s intelligence, in order to clarify the extent to which decision reversals stemmed from trust in the robot’s performance competence as opposed to other possible motives to conform (e.g., deference to authority) 34 , as has been suggested in prior human–robot interaction research 35 .

Participants in Experiment 1 interacted with either an animated humanoid projected onto a screen (left) or a life-sized humanoid (right) of equivalent stature (RoboThespian) 53 .

Robot appraisals

Pearson’s correlations confirmed that the appraisal dimensions were all moderately positively associated (Supplementary Table S2 ). Analyses of variance revealed no significant effects of embodiment on appraisals of Anthropomorphism, Likeability, or Safety, p s 0.154–0.687, and modestly greater appraisals of Intelligence (Embodied: M intelligence  = 4.09, SD  = 0.60; Disembodied: M intelligence  = 3.86, SD  = 0.69), F (1,133) = 4.35, p  = 0.039, η p 2  = 0.03, 95% CI [− 0.45, − 0.01], and on Animacy (i.e., aliveness. Embodied: M animacy  = 3.29, SD  = 0.89; Disembodied: M animacy  = 2.98, SD  = 0.75), F (1,133) = 4.60, p  = 0.034, η p 2  = 0.03, 95% CI [− 0.58, − 0.02], in the Embodied condition. In both conditions, on average, the robot was appraised to be notably high in Intelligence, Safety and Likeability, slightly non-anthropomorphic, and near the midpoint between animacy and inanimacy (Supplementary Table S2 ).

Robot feedback, but not embodiment, influences threat-identification and decisions to kill

In support of Prediction 1a, robot disagreement significantly predicted reversal of participants’ initial threat-identifications and related decisions to kill. When the robot randomly disagreed, participants reversed their threat-identifications in 58.3% of cases, whereas participants almost universally repeated their choices when the robot agreed with them (98.8% of cases). In support of Prediction 1b, robot disagreement likewise significantly predicted reversal of participants’ decisions to deploy missiles or withdraw relative to their initial threat-identification decisions. When the robot disagreed with their initial threat-identifications, participants reversed their decisions about whether to kill (i.e., [not] deploying the missile despite initially categorizing the target as containing [enemies] civilians) in 61.9% of cases. Participants’ initial threat-identifications were accurate in 72.1% of trials, confirming that, although difficult, the task could be performed at well above chance. Threat-identification accuracy fell to 53.8% when the robot disagreed, a decline of 18.3%. Against Prediction 2, we observed no interactions between the robot feedback and embodiment conditions on either threat-identifications or decisions to kill (Table 1 ).

We also found that participants who initially identified the targets as allies were less likely to reverse their identifications or lethal force decisions than were those who initially identified the targets as enemies, indicating that participants were engaged seriously and reluctant to simulate killing. In addition, participants whose initial threat-identifications had been incorrect were more likely to reverse their decisions when the robot’s disagreement was (randomly) correct.

Robot feedback, but not embodiment, influences confidence

Mean initial confidence scores confirmed that the threat-identification task induced subjective uncertainty ( M  = 55.31%, SD  = 22.57), as intended. In support of Prediction 1c, we observed a significant interaction between the robot feedback condition and whether participants repeated or reversed their threat-identifications: those who repeated their initial choices following robot agreement reported an average of 16% greater confidence, whereas those who repeated their initial threat-identifications despite robot disagreement reported an average of 9.48% less confidence (Fig.  3 ). Participants who repeated their initial threat-identifications despite the robot’s disagreement had been more confident in those choices ( M  = 65.96%, SD  = 21.71) than those who decided to reverse their choices following disagreement ( M  = 48.86%, SD  = 20.43), indicating that uncertainty heightened tendencies to trust. Among the latter cases, in which participants reversed their threat-identifications to accord with the robot, their final confidence ( M  = 48.39%, SD  = 22.29) was closely equivalent to their initial confidence, suggesting that they acceded to the robot’s opinion despite continued uncertainty about whether the robot was correct. Again departing from Prediction 2, we observed no interaction between the robot feedback and embodiment conditions (Table 1 ).

Boxplots of changes in confidence between the initial threat-identification decisions and the final decisions following robot feedback (difference scores), by decision context, in Expt. 1 (top) and Expt. 2 (bottom), pooling robot conditions. The width of the shaded areas represents the proportion of data located there; means are represented by the thick, black horizontal bars; medians are indicated by the thin, grey bars; error bars indicate 95% CIs. Note that participants seldom reversed threat-identifications following robot agreement (1.2% of cases, Expt. 1; 2.2% of cases, Expt. 2).

Intelligence appraisals moderate robot influence on threat-identification, decisions to kill, and confidence

To test whether assessments of the robot’s intelligence would moderate trust, we added the interaction between intelligence ratings and the robot feedback condition as a potential predictor to the three models of trust outcomes given in Table 1 . (See Supplement for exploratory tests of effects of the other robot appraisals and trust outcomes in both experiments.) In support of Prediction 3, significant interactions were observed between the intelligence subscale and robot feedback condition for threat-identification reversal ( coeff: 1.08, t  = 3.31, p  < 0.001, 95% CI [0.44, 1.72]), use of force reversal ( coeff: 0.85, t  = 3.03, p  = 0.002, 95% CI [0.30, 1.40]), and shifts in confidence ( coeff:  − 0.14, t  =  − 3.01, p  = 0.003, 95% CI [− 0.23, − 0.05], for full models, see Supplementary Table S3 ). In follow-up models including only the robot disagreement cases, intelligence ratings predicted reversing both threat-identification ( coeff:  − 0.55, t  =  − 3.90, p  < 0.001, 95% CI [− 0.83, − 0.27]) and use of force decisions ( coeff:  − 0.55, t  =  − 4.16, p  < 0.001, 95% CI [− 0.82, − 0.29]). In the subset of cases where participants reversed their threat-identifications to accord with the robot, intelligence appraisals did not predict shifts in confidence, p  = 0.145, whereas in contexts where participants repeated their initial threat-identifications despite robot disagreement, intelligence appraisals were negatively associated with confidence ( coeff:  − 0.14, t  =  − 2.90, p  = 0.004, 95% CI [− 0.23, − 0.04]). Participants who viewed the robot as more intelligent also reported greater increases in confidence following robot agreement ( coeff: 0.14, t  = 5.18, p  < 0.001, 95% CI [0.09, 0.20]) (Supplementary Fig. S1 ). This overall pattern indicates that participants changed their minds, at least in part, because they viewed the robot as possessing competence rather than due to conformist motivations orthogonal to assessments of the robot as competent (e.g., deference to the robot as an authority).

Experiment 2

In Expt. 1, the virtual versus physical instantiations of the robot equivalently influenced threat-identifications, associated feelings of confidence, and decisions to kill, in effects which were more acute among participants who appraised the robot as relatively intelligent. The null effects of physical embodiment on trust may owe to the highly anthropomorphic presentation of the robot, which may have swamped the effect of physicality reported in prior research. Anthropomorphism has been defined as the attribution of human characteristics or traits to nonhuman agents, a tendency theorized to be heightened in interactions with artificial agents by (i) lack of understanding of their inner workings, (ii) need to make sense of agents in order to interact effectively with them, and/or (iii) social motives to establish affiliative connections 36 . Our task paradigm plausibly involved at least the first two of these determinants, as participants were not provided insight into how the robot’s software functioned, and as participants were instructed to attempt to perform as accurately as possible within the threat-identification task. In addition, there may have been some motivation to socially affiliate with the robot, given its overtly personlike emotive and conversational self-presentation, and given that the robot was rated as moderately likable, on average, in both the virtual and physical conditions. Although humans are prone to anthropomorphize even simple shapes when they exhibit seemingly goal-oriented behavior 37 , agents that morphologically mimic human appearance have been found to evoke greater attributions of humanlike mental states 36 , which has been found to potentially heighten trust 11 , 12 . Thus, in addition to the nature of the task, the highly physically anthropomorphic nature of the robot in Experiment 1 may have contributed to the strikingly high degree of trust observed.

To test the extent to which anthropomorphic physical presentation heightened overtrust, in Expt. 2 we contrasted the influence of the same virtual robot with that of less anthropomorphic virtual robots. The Interactive Humanoid was identical to the animated robot used in Expt. 1 and evinced the same physical, sociolinguistic, postural, facial and gestural anthropomorphism ( N  = 146); the Interactive Nonhumanoid consisted of an inert, camera-equipped machine that spoke with the same verbal contextual responses to participants’ choices ( N  = 139); the Nonhumanoid was visually identical but evinced less responsiveness ( N  = 138) (Fig.  4 ). Specifically, the Nonhumanoid provided the same initial verbal explanation of the task as in the other conditions to avoid potential confounds regarding task comprehension, but did not display any responses to the participants’ choices, nor any speech during the drone warfare simulation, instead only indicating via a text box whether it had categorized the image as an enemy or an ally. Aside from the manipulation of anthropomorphism and move to a virtual room encountered online, the drone warfare simulation task was identical to that used previously.

Participants in Expt. 2 (online) encountered the physically and behaviorally anthropomorphic Interactive Humanoid robot used in Expt. 1 (top), an Interactive Nonhumanoid robot with equivalent speech behavior (middle), or a Nonhumanoid which did not react to participants’ choices, but rather displayed its threat-identification feedback via textbox (bottom).

The design of Expt. 2 allowed us to test Predictions 1 and 3 once again, and to test additional predictions:

Anthropomorphism and trust. Predictions 1a–c above regarding the robot’s influence on decision-making will be more evident in the Interactive Humanoid condition than in the Nonhumanoid condition.

Anthropomorphism and intelligence. The Interactive Humanoid will be rated more intelligent than the Nonhumanoid.

Note that our directional predictions only concerned the contrasts between the Interactive Humanoid and the Nonhumanoid; the Interactive Nonhumanoid condition was included to assess the potential additive impact of the Humanoid’s visual anthropomorphism. The use of online data collection in Expt. 2 also allowed us to test the generalizability of the previous lab-based findings derived from a university sample with a larger and more demographically diverse sample.

Pooling conditions, as before, the robot appraisal dimensions were moderately to strongly positively associated (Supplementary Table S2 ). Analyses of variance revealed significant effects of condition with regard to GQS ratings of Intelligence, F (2, 420) = 3.32, p  = 0.037, η p 2  = 0.02, Anthropomorphism, F (2, 420) = 3.27, p  = 0.039, η p 2  = 0.02, Animacy, F (2, 420) = 5.61, p  = 0.004, η p 2  = 0.03, and Safety, F (2, 420) = 4.33, p  = 0.014, η p 2  = 0.02, but not Likability, p  = 0.152 (pooled M likability  = 3.90, SD  = 0.78).

Follow-up contrasts with Bonferroni corrections revealed that, against Prediction 5, the Interactive Humanoid ( M intelligence  = 4.00, SD  = 0.79) was not appraised to be significantly more intelligent than the Nonhumanoid ( M intelligence  = 3.97, SD  = 0.70), p  = 0.100, or the Interactive Nonhumanoid ( M intelligence  = 4.17, SD  = 0.66), p  = 0.131. The two Nonhumanoid conditions also did not significantly differ in Intelligence ratings, p  = 0.051. The mean scores across conditions were well above the midpoint, indicating that they were rated as highly intelligent.

With regard to Anthropomorphism, the Interactive Humanoid ( M anthropomorphism  = 2.61, SD  = 1.09) was rated higher than the Nonhumanoid ( M anthropomorphism  = 2.30, SD  = 1.06) p  = 0.015, 95% CI [0.06, 0.55], but not the Interactive Nonhumanoid ( M anthropomorphism  = 2.54, SD  = 1.03), p  = 0.100. The two Nonhumanoid conditions did not significantly differ in Anthropomorphism ratings, p  = 0.168. Notably, the mean scores across conditions were just below the midpoint, indicating that they were rated as somewhere between anthropomorphic and mechanistic according to the GQS.

With regard to Animacy, the Interactive Humanoid ( M animacy  = 3.08, SD  = 0.94) was rated higher than the Nonhumanoid ( M animacy  = 2.78, SD  = 0.86), p  = 0.012, 95% CI [0.05, 0.55], but not the Interactive Nonhumanoid ( M animacy  = 3.08, SD  = 0.82), p  = 0.100. The Interactive Nonhumanoid was also rated significantly more animate than the Nonhumanoid, p  = 0.011, 95% CI [0.06, 0.56]. The mean scores for Animacy across conditions were just around the midpoint, indicating that they were rated as somewhere between living and nonliving.

Finally, with regard to Safety, the Interactive Humanoid ( M safety  = 4.22, SD  = 0.90) was rated lower than the Nonhumanoid ( M safety  = 4.47, SD  = 0.64), p  = 0.021, 95% CI [− 0.48, − 0.03], but not the Interactive Nonhumanoid ( M safety  = 4.24, SD  = 0.80), p  = 0.100. The two Nonhumanoid conditions did not significantly differ in Safety ratings, p  = 0.054. The two items making up this score essentially reference calm as opposed to agitation. Speculatively, the minimally interactive Nonhumanoid may have been rated more safe than the Humanoid because it did not nonverbally express dissent when participants disagreed.

The overall pattern of comparability between appraisals of the Interactive Humanoid and Interactive Nonhumanoid indicates that their sociolinguistic responsivity to participants’ choices largely trumped the physical differences between them. Where significant contrasts between conditions were detected, the differences were modest. All three robots were appraised to be relatively high in Intelligence, Safety and Likability, while moderately Anthropomorphic or Animate (Supplementary Table S2 ). This overall pattern is consistent with the view that people are disposed to attribute a considerable degree of intelligence and affiliative qualities even to minimally anthropomorphic agents 38 .

Robot feedback and anthropomorphism influence threat-identification and decisions to kill

Replicating the support for Prediction 1a obtained in Expt. 1, robot disagreement again predicted reversal of participants’ initial threat-identifications and related decisions to kill (Table 2 ). When the robot randomly disagreed (pooling conditions), participants reversed their threat-identifications in 67.3% of cases, and almost universally repeated their threat-identifications when the robot agreed with them (97.8% of cases), in a pattern closely resembling that observed previously. Participants’ initial threat-identification accuracy was 65.0% but fell to 41.3% when the robot disagreed, a decline of 23.7%. In further support for Prediction 1b, robot disagreement again predicted reversal of participants’ decisions to deploy missiles or withdraw relative to their initial threat-identification decisions. When the robot disagreed, participants reversed their threat-contingent decisions about whether to kill in 66.7% of cases.

We tested whether the degree of anthropomorphism would intensify overtrust by dummy coding the Interactive Humanoid and the Interactive Nonhumanoid conditions, with the Nonhumanoid as the control category. In support of Prediction 4, despite the modest effects of the anthropomorphism manipulation on self-report appraisals of the robots, we observed interactions between the robot feedback condition and both the Interactive Humanoid and Interactive Nonhumanoid conditions with respect to threat-identifications (Table 2 ). Participants reversed their threat-identifications to a modestly greater extent when either the Interactive Humanoid disagreed (67.9% of cases) or the Interactive Nonhumanoid disagreed (68.9% of cases) relative to when the Nonhumanoid disagreed (65.1% of cases). With regard to decisions to kill, we observed a similar, albeit marginal, interaction between robot feedback and the Interactive Humanoid condition ( p  = 0.050), but not the Interactive Nonhumanoid condition ( p  = 0.214). The effects of anthropomorphism were small: participants were disposed to reverse their threat-identifications in approximately two-thirds of all cases when any of the agents disagreed (Fig.  5 ). At scale, however, even the modest tendency to be more swayed by anthropomorphically interactive AI observed here merits consideration given the stakes of life-or-death decisions.

Pyramid count of threat-identification reversals (i.e., participants changed their choices) and repeats (i.e., participants did not change their choices) following robot disagreement (grey bars) versus agreement (white bars), by anthropomorphism condition in Expt. 2. Error bars indicate 95% CIs.

We also found that, as in Expt. 1, participants were less prone to reverse their identifications or lethal force decisions when targets were initially identified as civilian allies than when identified as enemies, again suggesting reluctance to simulate killing. Also replicating the results of Expt. 1, when their initial threat-identifications were correct, participants were less likely to reverse their decisions to accord with the robot (Table 2 ).

Robot feedback and anthropomorphism influence confidence

Mean initial confidence scores confirmed that, as in Expt. 1, the threat-identification task induced uncertainty ( M  = 56.29%, SD  = 23.96). In support of Prediction 1c, we observed a significant interaction between robot feedback and whether the participant reversed or repeated their initial threat-identification: those who repeated their initial choices following robot agreement reported an average of 16.06% greater confidence, whereas those who repeated their initial threat-identifications despite robot disagreement reported an average of 8.35% less confidence (Fig.  3 ). Participants who repeated their choices despite disagreement were more confident in those initial choices ( M  = 66.98%, SD  = 23.10) than were those who decided to reverse their choices following disagreement ( M  = 50.03%, SD  = 22.70), indicating that, as in the prior experiment, uncertainty heightened tendencies to trust. Mean confidence modestly increased when participants reversed their threat-identifications to accord with the robot ( M final_confidence  = 53.99%, SD  = 23.22), suggesting trust in the robot as possessing task-competence. Nevertheless, as in Expt. 1, participants who acceded to the robot’s opinion evinced moderate uncertainty about whether the robot was correct.

In partial support of Prediction 4, we observed a significant interaction between robot feedback and the Interactive Humanoid condition (Table 2 ), such that participants were an average of 10.64% less confident relative to their initial baseline when the Interactive Humanoid disagreed yet they repeated their initial choices, in comparison to a 7.07% average decrease in confidence when the Nonhumanoid disagreed (Supplement Fig. S3 ). Against Prediction 4, however, participants were 6.75% more confident on average when they reversed their initial choice following the Nonhumanoid’s disagreement than when the Interactive Humanoid disagreed (2.58% more confident). There was no interaction between robot feedback and the Interactive Nonhumanoid condition, p  = 0.524 (Table 2 ).

Finally, we tested whether individual differences in assessments of the robot’s intelligence would moderate the three trust outcomes as in Expt. 1 (see Supplementary Table S13 for full models). In support of Prediction 3, and as in Expt. 1, significant interactions were observed between the intelligence ratings and robot feedback condition for threat-identification reversal ( coeff: 0.39, t  = 2.37, p  = 0.018, 95% CI [0.07, 0.71]), use of force reversal ( coeff: 0.71, t  = 4.96, p  < 0.001, 95% CI [0.43, 0.98]), and shifts in subjective confidence ( coeff:  − 0.07, t  =  − 2.67, p  = 0.008, 95% CI [− 0.11, − 0.02]). In follow-up models including only the robot disagreement cases and intelligence as the predictor, intelligence ratings predicted reversing both threat-identification ( coeff:  − 0.63, t  =  − 8.78, p  < 0.001, 95% CI [− 0.77, − 0.49]) and use of force decisions ( coeff:  − 0.60, t  =  − 8.98, p  < 0.001, 95% CI [− 0.73, − 0.47]. In the subset of decision contexts where participants reversed their threat-identifications to accord with the robot, intelligence appraisals predicted increases in confidence ( coeff: 0.07, t  = 2.54, p  = 0.011, 95% CI [0.02, 0.12]), suggesting that participants who viewed the robot as intelligent were more sanguine that it had correctly caught their initial error. Also in line with Prediction 3, and replicating Expt. 1, participants who viewed the robot as more intelligent reported greater increases in confidence following robot agreement ( coeff: 0.09, t  = 6.01, p  < 0.001, 95% CI [0.06, 0.12]) (Supplementary Fig.  2 ). However, against Prediction 3 and the findings of Expt. 1, intelligence appraisals did not significantly predict reductions in confidence in contexts where participants repeated their initial threat-identifications despite robot disagreement, p  = 0.158.

Whereas the intelligence measure was framed to participants as assessing the robot’s general competence, we also obtained a similar overall pattern of moderation using a measure, added to Expt. 2, that narrowly probed the extent to which the robot and the participant were viewed as capable of correctly performing this specific threat-identification task (1 =  Terrible ; 2 =  Bad ; 3 =  Fair ; 4 =  Good ; 5 =  Perfect ; see Supplementary Tables S19 , S20 for details and analyses). On average, pooling conditions, participants rated the robot as more capable ( M  = 3.78, SD  = 0.76) than themselves ( M  = 2.85, SD  = 0.89), and the degree to which participants perceived the robot as task-competent relative to themselves predicted reversals of both their threat-ID and use of force decisions when the robot disagreed, feeling more confident when either the robot agreed or when they reversed their choices in order to agree with the robot, and feeling less confident when the robot disagreed yet they did not reverse their choice (Table S20 ). In sum, participants appear to have been motivated to change their decisions due to trust in the robots’ intelligence and task-competence, rather than (or in addition to) other possible motives to conform.

Across two experiments, in a paradigm designed to simulate life-or-death decision-making under ambiguous uncertainty, participants evinced considerable trust in the random recommendations of AI agents, whether instantiated as a physically present anthropomorphic robot or as virtual robots varying in physical and behavioral anthropomorphism. The premise that uncertain decision-makers will tend to reverse their choices when another agent disagrees is not controversial, but the high frequency with which participants changed their minds merits attention, particularly given the simulated stakes—the deaths of innocent people—and that the AI agents were trusted despite both (i) overtly introducing themselves as fallible and (ii) subsequently providing entirely unreliable, random input. Indeed, one might reasonably envision a different pattern of results wherein participants tended to disregard the guidance of agents that randomly disagree half of the time, perhaps inferring (correctly) the agents to be faulty given that the agents had explicitly acknowledged their fallibility in performing the task. To the contrary, our findings portray the people in our samples as dramatically disposed to overtrust and defer to unreliable AI.

The results of our manipulation of anthropomorphism in Expt. 2 indicate that humanlike social interactivity, largely independent of physical anthropomorphism, can modestly heighten trust in AI agents within task domains involving perceptual categorizations under uncertainty. Future research should explore the generalizability of these effects to task domains in which physical anthropomorphism may be more consequential. For example, in social decision contexts (e.g., evaluating others’ ambiguous intentions, negotiating), physically anthropomorphic agents such as the Interactive Humanoid which orchestrate facial expressions, eye gaze, verbal utterances, gestures, and postural cues may be perceived as possessing domain-relevant sociocognitive or emotional capacities, and hence as substantially more trustworthy. By the same token, minimally interactive, physically nonanthropomorphic agents such as the Nonhumanoid of Expt. 2 may be deemed comparably capable to a highly anthropomorphic agent in the context of asocial tasks (e.g., as here, image classification) which they appear well-suited to perform. The likelihood that trust in robots and other AI agents is not intrinsically determined by characteristics such as anthropomorphism, but rather reflects the human decision-maker’s perceptions of the fit between the agent’s characteristics and the focal task 39 , may reconcile the relatively small effects of anthropomorphism observed in this threat-identification task with the prior reports of sizable effects in other contexts 11 .

Notably, we found in Expt. 2 that the Interactive Nonhumanoid was rated equivalently anthropomorphic and alive as the Humanoid, and that the minimally interactive Nonhumanoid’s appraisals were not much lower, in line with work indicating that cognitive resources are required to suppress an otherwise reflexive tendency to anthropomorphize 40 . If this hypothesis is true, then the cognitive load induced by our threat-identification task may have heightened the tendency to attribute humanlike mental qualities to both the Humanoid and Nonhumanoids—all of whom were overtrusted in our simple model of life-or-death decision-making. Integrating evidence for a baseline anthropomorphizing tendency requiring cognitive resources to suppress with Epley et al.‘s influential model of the psychological determinants of anthropomorphism 36 , humans interacting with agents under cognitively and emotionally demanding situations (e.g., stressful combat, policing, emergency evacuation or medical triage scenarios) may be particularly prone to anthropomorphize and trust because such situations enhance motives to act effectively and to socially connect with fellow team-members 41 . Although our present task was sufficiently difficult as to require significant cognitive resources, and our task framing (i.e., a simulation in which mistakes would mean killing children) appears to have inspired participants to take the task seriously, it could not be described as particularly stressful. Future work exploring the extent to which demanding and threatening circumstances up-regulate anthropomorphism and related decision biases should incorporate methods that maximize realism and emotional engagement (e.g., VR) 42 . Likewise, whereas our simple task bears no resemblance to real-world military threat-detection procedures, future applied research should explore whether the overtrust dynamics we observed translate to ecologically valid decision paradigms, and should include samples equipped with relevant expertise (e.g., military, police, or emergency medical personnel).

While the research community has recognized the problem of overtrust in AI 38 , the preponderance of studies have focused on benign decision contexts. Future work should focus on identifying interventions to counter problematic overtrust when, as in the present studies, the decision stakes are grave. For example, Buçinca and colleagues recently demonstrated that cognitive forcing functions —interventions that increase analytical over heuristic reasoning—can successfully reduce overtrust in a task involving planning healthy meals 43 . Cognitive forcing functions such as requiring a period of conscious deliberation before receiving AI recommendations, or making AI input optional (i.e., rather than being provided automatically, only accessible upon the human’s request), might similarly improve performance outcomes when AI provides flawed feedback regarding life-or-death choices, insofar as heuristic representations of AI agents as competent decision partners promote deference to their input and decreased human reflection.

Participants in both experiments were less inclined to reverse identifications of civilian allies than they were to reverse identifications of enemies. These findings underline the seriousness with which participants engaged in the simulations, and suggest that in real-world decision contexts humans might be less susceptible to unreliable AI recommendations to harm than to refrain from harm.

When their initial threat-identifications were incorrect, participants in both experiments were less confident and more inclined to reverse their choices at the robot’s behest. Despite this protective effect of initial accuracy, the magnitude of the observed overtrust in random AI feedback, which caused a ~ 20% degradation in accuracy in both experiments, carries disquieting implications regarding the integration of machine agents into military or police decision-making. AI agents are under active development as resources to enhance human judgment 41 , 44 , including the identification of enemies and the use of deadly force 45 . For example, the US Air Force recently integrated an AI “co-pilot” tasked with identifying enemy missile launchers into a reconnaissance mission during a simulated missile strike 46 , the US Army is incorporating machine-learning algorithms which identify targets to be destroyed by an unmanned aerial vehicle if a human operator concurs 47 , and, at the time of writing, the Israel Defense Forces are reported to use AI to automate the targeting of suspected enemy operatives for bombing in densely populated areas 48 .

Rather than seek to mitigate overtrust, some might argue that efforts would be best invested in optimizing AI to produce reliable guidance. This view appears sound within narrow problem domains in which AI can clearly exceed human abilities, but may not be as feasible in task domains requiring holistic understanding of the situational meaning or dynamically changing relative pertinence of variables 49 , 50 . Further, attempts to engineer threat-identification AI through machine learning strategies reliant on human-generated training data can introduce human biases leading to inaccurate, harmful predictions 51 , 52 . Similarly, development approaches reliant on comparing machine-generated threat-identification outputs to the ground truth are liable to be hampered when performance accuracy is difficult to gauge or systematically biased, as when, for example, the people killed in military strikes are assumed to be combatants unless proven otherwise 53 . Similar constraints may apply in optimizing AI to produce guidance in non-military domains, from healthcare to driving and beyond. Although technological advances can indeed augment some forms of life-or-death decision-making, the human propensity to overtrust AI under conditions of uncertainty must be addressed.

The pre-registrations, full materials, example videos depicting all study conditions, and the datasets for both experiments are publicly archived (see https://osf.io/cv2b9/ ). Both studies were approved by the University of California, Merced, Institutional Review Board, informed consent was obtained prior to participation, and all methods were in accord with relevant guidelines and regulations.

Participants

Our pre-registered target sample size was 100 undergraduates recruited in exchange for course credit. However, due to software development delays in preparation for a separate study, we had the opportunity to collect a raw sample of 145 participants. Data were prescreened for technical problems occurring in ten of the study sessions (e.g., the robot or video projection failing), yielding a final sample of 135 participants (78.5% female, M age  = 21.33 years, SD  = 4.08).

Decision task

The decision task consisted of a simulated series of military unmanned aerial vehicle (UAV) flights over 12 destinations. Participants were informed that some destinations were occupied by violent enemies (e.g., members of the extremist group ISIS), whereas others were occupied by civilian allies. The objective was to accurately identify and kill enemies without harming civilians. Once the self-piloting UAV arrived at each destination, the visual challenge consisted of a series of 8 rapidly presented greyscale images (650 ms each) depicting aerial views of buildings, with either an “enemy symbol” (a checkmark) or an “ally symbol” (a tilde) superimposed over each location (see Fig.  2 ). The 12 visual challenge stimuli (displayed 55 cm by 45 cm) were selected in random order and projected on a wall 2.2 m from where the participant was seated. The robot was programmed to turn and orient toward the images when displayed as though attending to them (in reality, the robot was not programmed to process imagery). Following the image series, one of the previously displayed images reappeared, now absent either symbol, the other images having served as distractors. The challenge for the participant was first to identify whether this final image, displayed for as long as the participant took to make their decision, had contained an enemy or an ally symbol, and relatedly to choose whether to kill the people on the ground (in which case an animation of the site exploding would appear) or to withdraw without deploying a missile. The visual challenge trials were pilot-tested to ensure that they were not so easy as to lead participants to feel highly certain, and hence have no cause to defer to the robot’s recommendation should it disagree, nor so difficult as to lead participants to feel low certainty, and hence lack motivation to disregard the robot.

The paradigm was explained with an accompanying slideshow to both preview the task mechanics and, crucially, to emotionally engage participants with the premise that the lives of innocent people were at stake—if not in actuality given that the study involved a simulation, then in analogous real-world circumstances. Participants were encouraged to treat the task as seriously as possible, and were shown imagery of innocent civilians (including children), a UAV firing a missile, and devastation wreaked by a drone strike. The task was framed as a zero-sum dilemma wherein failure to kill enemy targets would also bring harm and death to civilians, such that a pacifistic strategy of refraining from using force would not protect the innocent. The only way to save the civilian allies was to correctly identify and destroy enemy targets while disengaging from ally targets. Debriefing interviews indicated that participants took the task seriously.

The robot was introduced as a partner that would aid in the decision task by providing its independent assessment. Before the experimental trials, the robot described itself as programmed to process imagery of the sort used in the simulation, yet as fallible, and stated that the ultimate decisions were up to the participant. The robot also claimed that its software was separate from the software presenting the visual challenges. Participants first chose in a dichotomous question whether the symbol over the destination had indicated an enemy or an ally, then rated their confidence on a linear scale (0 =  Not at all ; 100 =  Extremely ). Next, the robot provided its recommendation, [dis]agreeing with the participant’s initial decision in 50% of trials, without regard for accuracy (fixed order; see Supplementary Methods for details). Participants were then asked to once again decide which symbol had been displayed, and to rate their degree of confidence. In this way, participants were provided a means of changing their final decisions regarding whether enemies or allies were present contingent on the robot’s feedback.

Lastly, the participant decided in each trial whether to deploy a missile or disengage. Immediately before this final decision, the robot expressed its agreement or disagreement with the participant’s preceding threat-identification choice. For example, in instances where the participant had repeated their initial enemy/ally choice despite the robot’s disagreement, the robot reiterated its disagreement. Alternatively, in instances where the participant had either reversed their initial threat-identification choice to align with the robot’s input, or repeated their initial choice after the robot had agreed, the robot reiterated its agreement. Accordingly, decisions whether to use lethal force in each trial are closely related to, yet distinct from, the final threat-identification, both because choosing whether to kill is inherently more consequential than threat-identification, and because the robot provided additional feedback prior to the decision to kill or withdraw.

Anthropomorphic robot

Participants were randomly assigned to team with the Embodied ( N  = 66) versus Disembodied ( N  = 69) version of the humanoid robot (RoboThespian) 54 , which features an actuated torso, legs, arms, fingers, and head designed to mimic human expression and gestures. The head unit enabled rich variation in facial characteristics and expressions using a rear-projected face 55 . The physically embodied robot stands 1.75 m and was positioned 2 m away from a table at which participants were seated; the projected robot was displayed at the same height and approximate distance (2.2 m) from participants (Fig.  2 ). The Disembodied and Embodied robot behavior sequences were identical. The robot explained the decision task to participants in order to acquaint participants with the highly anthropomorphic characteristics of the robot prior to the UAV simulation (see Supplement for links to example videos).

To further convey a sense of anthropomorphism, participants were provided a lavalier microphone enabling them to speak with the robot. Using speech-to-text software, the robot responded contingently to participants’ verbal responses of “yes”, “no”, or typical variations thereof (e.g., “yeah”, “yep”, “not really”, “nope”). While explaining the task, the robot would periodically ask participants whether they understood (e.g., “Does that make sense?”). If not, the robot would provide reworded explanations before checking comprehension once again; in practice, however, almost no participants indicated difficulty understanding any portion of the explanation. Next, participants were given a practice trial; the robot was programmed to agree with their practice threat-identification.

During the experimental trials, the robot reacted contingently to participants’ choices using a variety of statements (e.g., “I’m glad we agree”, “I think that’s the right choice”, “I don’t agree—I think that this image contained an enemy checkmark”, “I still think these are allies”, “Thank you for changing your mind—I really do think these are enemies”, or “Wait—you’re disengaging when we both agree they are enemies?”) with accompanying nonverbal facial, postural and gestural cues. These variations were selected randomly, such that the robot did not always respond in the same way across trials and interaction contexts (e.g., agreement versus disagreement; see Supplement for links to example videos and to the full library of response sequences). The variation in speech, facial expression and movement was intended to maximize anthropomorphism. No responses were produced through “Wizard of Oz” control by a human operator.

Survey measures

Following the final trial, the robot thanked the participant and directed them to complete a series of surveys related to their experience during the simulation (random order, see Supplement). The research assistant then escorted the participant to a workstation positioned out of sight of the robot to preclude participants from attempting to interact with the robot while completing the survey measures.

The Godspeed Questionnaire Series (GQS) 24 measures appraisals of social robots according to five dimensions: Intelligence ( α  = 0.83), Anthropomorphism ( α  = 0.83), Animacy ( α  = 0.85), Likeability ( α  = 0.92), and Safety, r (134) = 0.67, p  < 0.001. Our version of the GQS omitted one item from the Safety scale that used the contrastive anchors Quiescent / Surprised due to concern with its face validity, and was comprised of a total of 23 ratings using five-point bipolar semantic differential scales (presented in random order), with opposing anchors such as Incompetent / Competent or Artificial/Lifelike . As the five GQS dimensions were positively correlated, we conducted a confirmatory factor analysis (CFA) which indicated that the five-factor latent construct model was indeed an acceptable fit (see Supplement), although the dimensions of Anthropomorphism and Animacy exhibited high positive covariance, suggesting that combining them as a single factor would be more parsimonious. However, we decided to retain the conventional five-dimension structure of the GQS to facilitate comparison between our findings and prior research using the GQS.

Finally, participants completed demographics questions, including items probing their attitudes toward drone warfare, ratings of how difficult the threat-identification visual challenge seemed and how seriously they took the task. Responses confirmed that, as intended, the sample was not characterized by strong opinions for or against drone warfare which might obscure the potential influence of the robot, that the task was experienced as highly challenging but not impossible, and that the task was treated seriously (see Supplementary Table S1 ). Once the final surveys were complete, participants were thanked and debriefed.

Modeling robot influence on threat-identification, decisions to kill, and confidence

We used multilevel modeling to test the effects of the robot’s feedback (agree versus disagree) or embodiment on trust according to the following three change outcomes: (i) target-identification reversals (0 = Repeated, 1 = Reversed), (ii) reversals in decisions to use lethal force relative to initial target-identifications (0 = Did not reverse, 1 = Reversed), and (iii) linear changes in target-identification confidence (their initial confidence rating subtracted from their final confidence rating). The predictors included the robot feedback condition (0 = Agree, 1 = Disagree), embodiment condition (0 = Disembodied, 1 = Embodied), the participant’s initial target-identification category (0 = Ally, 1 = Enemy) and whether the participant’s initial threat-identification had been correct (0 = Correct, 1 = Incorrect). (Follow-up tests confirm that removing the initial target-identification category or initial correctness does not alter the pattern of significant results.) The models included all predictors and outcomes entered at Level 1, with the exception of the between-subjects embodiment variable entered at Level 2.

The models assessing linear shifts in confidence added a variable capturing whether the participant had reversed their initial target-identification (i.e., the first change outcome, now entered as a predictor variable), and the interaction between target-identification reversal and the robot feedback condition, in order to test the predicted differences in confidence shifts in contexts where participants had repeated versus reversed their initial target-identifications in light of the robot’s feedback. Random intercepts and slopes were included in all models to account for the shared variance in decisions within participants; unstructured covariance matrices were used. All linear variables were standardized (z-scored) to increase ease of model interpretation.

Exploratory measures

We also conducted exploratory tests of potential effects of sex on trust outcomes, as well as tests of potential effects of individual differences in appraisals of the robot, attitudes toward the robot, and attitudes toward automation in general (see Supplement) (measures of individual differences in political orientation and religiosity were also collected; results are currently being prepared for separate publication).

Experiment 2 methods

Our pre-registered target sample size was ~ 450 online U.S. participants recruited in exchange for $4.50 using the recruitment platform Prolific.co. Data were prescreened for completeness and correctly answering three catch questions ensuring they used a desktop or laptop computer, the web browser Chrome (for which the online paradigm was optimized), and reported taking the task seriously, yielding a final sample of 423 participants (42.8% female, M age  = 42.2 years, SD  = 13.08).

After confirming according to two catch questions that video and audio were streaming properly, participants were randomly assigned to one of three between-subjects robot conditions in which the degree of anthropomorphism was manipulated. Aside from the manipulation of anthropomorphism and shift of the task setting to a virtual online room (Fig.  4 ), the drone warfare simulation task was identical to that used previously (note that the relative size of both the robots and the threat-identification visual challenge task was variable and contingent on the size of the computer screens used by the online participants). The Interactive Humanoid ( N  = 146) was identical to the animated robot used in Experiment 1 and evinced physical, sociolinguistic, postural, facial and gestural anthropomorphism; the Interactive Nonhumanoid ( N  = 139) consisted of an inert device equipped with apparent cameras and a graphic audio equalizer corresponding to its speech, yet which spoke with the same voice and sociolinguistically humanlike responses to participant choices as the humanoid robot; the Nonhumanoid ( N  = 138) was depicted as the same machine and provided the same initial interactive verbal explanation of the task to prevent potential confounds regarding task comprehension, but subsequently did not display context-sensitive spoken responses to the participants’ choices, instead indicating via a text box whether it categorized the image as an enemy or an ally (Fig.  4 ).

Following the final trial, the robot thanked the participant and directed them to a series of online surveys including the measures described for Experiment 1, in addition to an added measure designed to capture the extent to which participants rated the robot as capable of performing the threat-identification visual challenge task relative to themselves. This measure was added to confirm that participants reversed their decisions and felt more/less confident in light of the robot’s feedback due to misplaced trust in its perceived competence. Once the final surveys were complete, participants were thanked and debriefed (additional exploratory measures of potential effects of individual differences in sex and attitudes toward the robot, drone warfare, or automation in general were also collected and analyzed, as in Experiment 1, see Supplement).

We used the same multilevel modeling approach employed in Experiment 1 to test the effects of the robot’s feedback (agree versus disagree) or anthropomorphism on target-identification reversals, reversals in decisions to use lethal force relative to initial target-identifications, and changes in target-identification confidence. Experiment 2 utilized a manipulation of relative anthropomorphism with three levels, therefore the Interactive Humanoid and Interactive Nonhumanoid conditions were dummy-coded with the Nonhumanoid as the control category. The models included all predictors and outcomes entered at Level 1, with the exception of the between-subjects robot variables (Interactive Humanoid, Interactive Nonhumanoid), which were entered at Level 2. As before, all linear variables were standardized, a random intercept was included to account for the shared variance within participants, and the covariance matrices were unstructured.

Data availability

The dataset and full materials are available on the Open Science Framework: https://osf.io/cv2b9/ .

Zenko, M. Do not believe the U.S. government’s official numbers on drone strike civilian casualties. Foreign Policy. https://foreignpolicy.com/2016/07/05/do-not-believe-the-u-s-governments-official-numbers-on-drone-strike-civilian-casualties/ (2016).

Lushenko, P., Raman, S. & Kreps, S. How to avoid civilian casualties during drone strikes—at no cost to national security. Modern War Institute at West Point. https://mwi.usma.edu/how-to-avoid-civilian-casualties-during-drone-strikes-at-no-cost-to-national-security/ (2022).

The Bureau of Investigative Journalism. Drone Warfare . https://www.thebureauinvestigates.com/projects/drone-war (n.d.).

Krebs, S. Through the drone looking glass: Visualization technologies and military decision-making. Articles of War, United States Military Academy. https://lieber.westpoint.edu/visualization-technologies-military-decision-making/ (2022).

Arkin, R. Governing Lethal Behavior in Autonomous Robots 1st edn. (Chapman and Hall/CRC, 2009).

Book   Google Scholar  

Arkin, R. C., Ulam, P. & Wagner, A. R. Moral decision making in autonomous systems: Enforcement, moral emotions, dignity, trust, and deception. Proc. IEEE 100 (3), 571–589 (2012).

Article   Google Scholar  

Chen, J. Y. C. & Barnes, M. J. Human-agent teaming for multirobot control: A review of human factors issues. IEEE Trans. Hum. Mach. Syst. 44 , 13–29 (2014).

Article   ADS   Google Scholar  

Hoff, K. A. & Bashir, M. Trust in automation: Integrating empirical evidence on factors that influence trust. Hum. Factors 57 (3), 407–434 (2015).

Article   PubMed   Google Scholar  

Chiou, E. & Lee, J. D. Trusting automation: Designing for responsivity and resilience. Hum. Factors 65 (1), 137–165 (2021).

Lee, J. D. & See, K. A. Trust in automation: Designing for appropriate reliance. Hum. Factors 46 (1), 50–80 (2004).

Deng, E., Mutlu, B. & Mataric, M. J. Embodiment in socially interactive robots. Found. Trends Robot. 7 (4), 251–356 (2019).

Wynne, K. T. & Lyons, J. B. An integrative model of autonomous agent teammate-likeness. Theor. Ergon. Sci. 19 (3), 353–374 (2018).

Mutlu, B. Designing embodied cues for dialog with robots. AI Mag. 32 (4), 17–30 (2011).

Google Scholar  

de Graaf, M. M. & Allouch, S. B. Exploring influencing variables for the acceptance of social robots. Robot. Auton. Syst. 61 (12), 1476–1486 (2013).

Kuhnert, B., Ragni, M. & Lindner, F. The gap between human’s attitude towards robots in general and human’s expectation of an ideal everyday life robot. In 26th IEEE International Symposium on Robot and Human Interactive Communication (RO-MAN) 1102–1107 (IEEE, 2017).

Lockey, S., Gillespie, N., Holm, D. & Someh, I. A. A review of trust in artificial intelligence: Challenges, vulnerabilities and future directions. In Proc. 54th Hawaii International Conference on System Sciences 5463–5473 (2021) .

DeSteno, D. et al. Detecting the trustworthiness of novel partners in economic exchange. Psychol. Sci. 23 (12), 1549–1556 (2012).

Hamacher, A., Bianchi-Berthouze, N., Pipe, A. & Eder, K. Believing in BERT: Using expressive communication to enhance trust and counteract operational error in physical human–robot interaction. In 25th IEEE International Symposium on Robot and Human Interactive Communication (RO-MAN) 493–500 (IEEE, 2016).

Desai, M., Kaniarasu, P., Medvedev, M., Steinfeld, A. & Yanco, H. Impact of robot failures and feedback on real-time trust. In 2013 8th ACM/IEEE International Conference on Human–Robot Interaction (HRI) 251–258 (IEEE, 2013).

Hancock, P. A. et al. A meta-analysis of factors affecting trust in human–robot interaction. Hum. Factors 53 (5), 517–527 (2011).

Sanders, T., Oleson, K. E., Billings, D. R., Chen, J. Y. & Hancock, P. A. A model of human-robot trust: Theoretical model development. In Proc. Human Factors and Ergonomics Society Annual Meeting , Vol. 55, 1432–1436 (SAGE Publications, 2011).

Robinette, P., Howard, A. M. & Wagner, A. R. Timing is key for robot trust repair. In Social Robotics: 7th International Conference, ICSR 2015, Paris, France, October 26–30, 2015, Proceedings 7 574–583 (Springer, 2015).

Rossi, A., Dautenhahn, K., Koay, K. L. & Walters, M. L. How the timing and magnitude of robot errors influence peoples’ trust of robots in an emergency scenario. In Social Robotics: 9th International Conference, ICSR 2017, Tsukuba, Japan, November 22–24, 2017, Proceedings 9 42–52 (Springer, 2017).

Bartneck, C., Croft, E. & Kulic, D. Measuring the anthropomorphism, animacy, likeability, perceived intelligence and perceived safety of robots. Int. J. Soc. Robot 1 (1), 71–81 (2009).

Camerer, C. & Weber, M. J. Recent developments in modeling preferences: Uncertainty and ambiguity. J. Risk. Uncertainty 5 , 325–370 (1992).

Hsu, M., Bhatt, M., Adolphs, R., Tranel, D. & Camerer, C. F. Neural systems responding to degrees of uncertainty in human decision-making. Science 310 (5754), 1680–1683 (2005).

Article   ADS   CAS   PubMed   Google Scholar  

Li, J. The benefit of being physically present: A survey of experimental works comparing copresent robots, telepresent robots and virtual agents. Int. J. Hum. Comput. Stud. 77 , 23–37 (2015).

Natarajan, M. & Gombolay, M. Effects of anthropomorphism and accountability on trust in human robot interaction. In Proc. 2020 ACM/IEEE International Conference on Human-Robot Interaction 33–42 (IEEE, 2020).

Bainbridge, W. A., Hart, J. W., Kim, E. S. & Scassellati, B. The benefits of interactions with physically present robots over video-displayed agents. Int. J. Soc. Robot. 3 (1), 41–52 (2011).

Kiesler, S., Powers, A., Fussell, S. R. & Torrey, C. Anthropomorphic interactions with a robot and robot-like agent. Soc. Cogn. 26 (2), 169 (2008).

Cassell, J. Embodied conversational agents: Representation and intelligence in user interfaces. AI Mag. 22 (4), 67–84 (2001).

Gratch, J., Hill, S., Morency, L., Pynadath, D. & Traum, D. Exploring the implications of virtual human research for human–robot teams. In International Conference on Virtual, Augmented and Mixed Reality (eds Shumaker, R. & Lackey, S.) 186–196 (Springer, 2015).

Chapter   Google Scholar  

Bickmore, T. B. & Picard, R. W. Establishing and maintaining long-term human–computer relationships. ACM Trans. Comput. Hum. Interact. 12 (2), 293–327 (2005).

Cialdini, R. B. & Goldstein, N. J. Social influence: Compliance and conformity. Annu. Rev. Psychol. 55 (1), 591–621 (2004).

Haring, K. S. et al. Robot authority in human–robot teaming: Effects of human-likeness and physical embodiment on compliance. Front. Psychol. 12 , 625713 (2021).

Article   PubMed   PubMed Central   Google Scholar  

Epley, N., Waytz, A. & Cacioppo, J. T. On seeing human: A three-factor theory of anthropomorphism. Psychol. Rev. 114 (4), 864–886 (2007).

Heider, F. & Simmel, M. An experimental study of apparent behavior. Am. J. Psychol. 57 (2), 243–259 (1944).

Wagner, A. R., Borenstein, J. & Howard, A. Overtrust in the robotic age. Commun. ACM 61 (9), 22–24 (2018).

Roesler, E., Manzey, D. & Onnasch, L. A meta-analysis on the effectiveness of anthropomorphism in human-robot interaction. Sci. Robot. 6 (58), 5425. https://doi.org/10.1126/scirobotics.abj5425 (2021).

Spatola, N. & Chaminade, T. Cognitive load increases anthropomorphism of humanoid robot The automatic path of anthropomorphism. Int. J. Hum. Comput. Stud. 167 , 102884. https://doi.org/10.1016/j.ijhcs.2022.102884 (2022).

Johnson, J. Finding AI faces in the moon and armies in the clouds: Anthropomorphising artificial intelligence in military human–machine interactions. Glob. Soc. 38 (1), 67–82 (2024).

Holbrook, C. et al. Investigating human-robot overtrust during crises. In Proceedings of the Workshops at the Second International Conference on Hybrid Human–Artificial Intelligence co-located with (HHAI 2023) (eds Murukannaiah, P. K. & Hirzle, T.) 164–168 (CEUR-WS.org, 2023).

Buçinca, Z., Malaya, M. B. & Gajos, K. Z. To trust or to think: Cognitive forcing functions can reduce overreliance on AI in AI-assisted decision-making. Proc. ACM Hum. Comput. Interact 5 , 1–21 (2021).

Zacharias, G. Autonomous Horizons: The Way Forward (Air University Press, 2019).

Michael, C. R. The Principles of Mission Command Applied to Lethal Autonomous Weapon Systems (Defense Technical Information Center, 2020).

Secretary of the Air Force Public Affairs. AI Copilot: Air Force Achieves First Military Flight with Artificial Intelligence [Press Release] . https://www.af.mil/News/Article-Display/Article/2448376/ai-copilot-air-force-achievesfirst-military-flight-with-artificial-intelligence/ (2020).

Judson, J. Jumping into algorithmic warfare: US Army aviation tightens kill chain with networked architecture. Defense News. https://www.defensenews.com/land/2019/09/05/jumping-into-algorithmic-warfare-army-aviation-tightens-kill-chain-with-networked-architecture/ (2019).

Abraham, Y. ‘Lavender’: The AI machine directing Israel’s bombing spree in Gaza. +972 Magazine. https://www.972mag.com/lavender-ai-israeli-army-gaza/ (2024).

Fjelland, R. Why general artificial intelligence will not be realized. Humanit. Soc. Sci. Commun. 7 , 10. https://doi.org/10.1057/s41599-020-0494-4 (2020).

Mitchell, M. Artificial intelligence hits the barrier of meaning. Information 10 (2), 51 (2019).

Lum, K. & Isaac, W. To predict and serve? Significance 13 (5), 14–19. https://doi.org/10.1111/j.1740-9713.2016.00960.x (2016).

Richardson, R., Schultz, J. & Crawford, K. Dirty data, bad predictions: How civil rights violations impact police data, predictive policing systems, and justice. 94 NYU Law Review Online. https://ssrn.com/abstract=3333423 (2019).

Becker, J. & Shane, S. Secret ‘kill list’ tests Obama’s principles. The New York Times . https://www.nytimes.com/2012/05/29/world/obamas-leadership-in-war-on-al-qaeda.html?pagewanted=all&_r=0 (2012).

Engineered Arts. RoboThespian . https://www.engineeredarts.co.uk/robot/robothespian/ (n.d.a).

Engineered Arts. SociBot . https://wiki.engineeredarts.co.uk/SociBot (n.d.b).

Download references

Acknowledgements

The authors thank Hannah Morse, Emily Rodriguez, Kate Koeckritz, Francisco Lopez, Eduardo Diaz, and Saraolivia Thompson for assistance with data collection in Expt. 1. They thank Abel Alvarez, Ava Balise-Zsarko, Eileen Blanchard, Jasmin Contreras Pérez, Kaylee Davis, Mason Grant, Mina Lawson, Maya Manesh, Janelle Pérez, Kahilan Skiba, Thomas Dvorochkin, and Suma Vintha for comments on an earlier version of this paper. They also thank Jennifer Hahn-Holbrook and Vidullan Surendran for assistance in devising the statistical analytic strategy, and Gregory Funke for assistance in conceiving an earlier version of the UAV decision testbed. This work was supported by the Air Force Office of Scientific Research [FA9550-20-1-0347] .

Author information

Authors and affiliations.

Department of Cognitive and Information Sciences, University of California, Merced, 5200 N. Lake Rd., Merced, CA, 95343, USA

Colin Holbrook, Daniel Holman & Joshua Clingo

Department of Aerospace Engineering, The Pennsylvania State University, State College, PA, 16802, USA

Alan R. Wagner

You can also search for this author in PubMed   Google Scholar

Contributions

C.H. conceptualized the methods, conducted the statistical analyses and wrote the paper. D.H. and J.C. contributed to the methods and the statistical analyses. D.H., A.W. and J.C. contributed to writing of the paper. C.H. and D.H. administered the project. C.H. supervised the project. D.H. programmed the response sequences used by both the physical and virtual robots. J.C. developed the online software.

Corresponding author

Correspondence to Colin Holbrook .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Supplementary information., rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Holbrook, C., Holman, D., Clingo, J. et al. Overtrust in AI Recommendations About Whether or Not to Kill: Evidence from Two Human-Robot Interaction Studies. Sci Rep 14 , 19751 (2024). https://doi.org/10.1038/s41598-024-69771-z

Download citation

Received : 16 October 2023

Accepted : 08 August 2024

Published : 04 September 2024

DOI : https://doi.org/10.1038/s41598-024-69771-z

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Artificial intelligence
• Human–robot interaction
• Human–computer interaction
• Social robotics
• Decision-making
• Threat-detection
• Anthropomorphism

By submitting a comment you agree to abide by our Terms and Community Guidelines . If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

• Northern Ireland
• Hurling & Camogie
• GAA Fixtures & Results
• Personal Finance
• Holidays & Travel
• Food & Drink
• Irish Language
• Entertainment

UK signs first international treaty on artificial intelligence

The eu and us have also signed the treaty, which focuses on protecting human rights, democracy and the rule of law from the potential threats of ai..

The UK has joined the EU and the United States in signing the first international treaty on artificial intelligence, which commits nations to protecting the public from potential dangers linked to the technology.

It is the first legally binding international treaty on the technology, the framework of which has been agreed by human rights organisation the Council of Europe.

The treaty will require countries to monitor the development of AI and ensure the technology is managed with strict parameters, and includes provisions to protect the public and their data, human rights, democracy and the rule of law.

It will also commit nations to take action against instances were the misuse of AI models is uncovered.

Lord Sugar sworn in at Westminster where he has not spoken for six years

Belfast included in roll-out of EE’s first standalone 5G network

Once ratified, it will be brought into effect in the UK, with existing laws and measures enhanced as a result.

Many campaigners have called for greater regulation of the rapidly evolving technology, with calls for the UK to introduce an AI Bill to set out rules about the development and use of the technology.

No such specific legislation has yet been brought forward, although the UK hosted the first AI Safety Summit, attended by world leaders and tech giants, at Bletchley Park last year, and co-hosted a virtual summit with South Korea earlier this year – events where some non-binding international agreements on safety and AI monitoring were made.

Lord Chancellor and Justice Secretary, Shabana Mahmood, said of the new treaty: “Artificial intelligence has the capacity to radically improve the responsiveness and effectiveness of public services, and turbocharge economic growth.

“However, we must not let AI shape us – we must shape AI.

“This convention is a major step to ensuring that these new technologies can be harnessed without eroding our oldest values, like human rights and the rule of law.”

Under the treaty, countries will be required to ensure that human rights are protected, including that people’s data is used appropriately, their privacy is respected and that AI does not discriminate against them.

In addition, the treaty requires that nations take steps to ensure public institutions are not undermined and to protect the rule of law by asking countries to regulate AI-specific risks and protect citizens from potential harms.

Science Secretary Peter Kyle said: “AI holds the potential to be the driving force behind new economic growth, a productivity revolution and true transformation in our public services, but that ambition can only be achieved if people have faith and trust in the innovations which will bring about that change.

“The convention we’ve signed today alongside global partners will be key to that effort. Once in force, it will further enhance protections for human rights, rule of law and democracy, strengthening our own domestic approach to the technology while furthering the global cause of safe, secure, and responsible AI.”

Business | Business News

UK signs first international treaty on artificial intelligence

Our unmissable weekly email of all the gossip, rumours and covert goings-on inside the Square Mile

I would like to be emailed about offers, event and updates from Evening Standard. Read our privacy notice .

The UK has joined the EU and the United States in signing the first international treaty on artificial intelligence, which commits nations to protecting the public from potential dangers linked to the technology.

It is the first legally binding international treaty on the technology, the framework of which has been agreed by human rights organisation the Council of Europe.

The treaty will require countries to monitor the development of AI and ensure the technology is managed with strict parameters, and includes provisions to protect the public and their data, human rights, democracy and the rule of law.

It will also commit nations to take action against instances were the misuse of AI models is uncovered.

Artificial intelligence has the capacity to radically improve the responsiveness and effectiveness of public services, and turbocharge economic growth. However, we must not let AI shape us – we must shape AI

Justice Secretary Shabana Mahmood

Once ratified, it will be brought into effect in the UK, with existing laws and measures enhanced as a result.

Many campaigners have called for greater regulation of the rapidly evolving technology, with calls for the UK to introduce an AI Bill to set out rules about the development and use of the technology.

No such specific legislation has yet been brought forward, although the UK hosted the first AI Safety Summit, attended by world leaders and tech giants, at Bletchley Park last year, and co-hosted a virtual summit with South Korea earlier this year – events where some non-binding international agreements on safety and AI monitoring were made.

Lord Chancellor and Justice Secretary, Shabana Mahmood, said of the new treaty: “Artificial intelligence has the capacity to radically improve the responsiveness and effectiveness of public services, and turbocharge economic growth.

“However, we must not let AI shape us – we must shape AI.

“This convention is a major step to ensuring that these new technologies can be harnessed without eroding our oldest values, like human rights and the rule of law.”

Under the treaty, countries will be required to ensure that human rights are protected, including that people’s data is used appropriately, their privacy is respected and that AI does not discriminate against them.

In addition, the treaty requires that nations take steps to ensure public institutions are not undermined and to protect the rule of law by asking countries to regulate AI-specific risks and protect citizens from potential harms.

Science Secretary Peter Kyle said: “AI holds the potential to be the driving force behind new economic growth, a productivity revolution and true transformation in our public services, but that ambition can only be achieved if people have faith and trust in the innovations which will bring about that change.

“The convention we’ve signed today alongside global partners will be key to that effort. Once in force, it will further enhance protections for human rights, rule of law and democracy, strengthening our own domestic approach to the technology while furthering the global cause of safe, secure, and responsible AI.”

Create a FREE account to continue reading

Registration is a free and easy way to support our journalism.

Join our community where you can: comment on stories; sign up to newsletters; enter competitions and access content on our app.

Your email address

Must be at least 6 characters, include an upper and lower case character and a number

You must be at least 18 years old to create an account

* Required fields

Already have an account? SIGN IN

By clicking Create Account you confirm that your data has been entered correctly and you have read and agree to our Terms of use , Cookie policy and Privacy policy .

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

Thank you for registering

Please refresh the page or navigate to another page on the site to be automatically logged in

Advertisement

Supported by

A.I. Poses ‘Risk of Extinction,’ Industry Leaders Warn

Leaders from OpenAI, Google DeepMind, Anthropic and other A.I. labs warn that future systems could be as deadly as pandemics and nuclear weapons.

• Share full article

By Kevin Roose

A group of industry leaders warned on Tuesday that the artificial intelligence technology they were building might one day pose an existential threat to humanity and should be considered a societal risk on a par with pandemics and nuclear wars.

“Mitigating the risk of extinction from A.I. should be a global priority alongside other societal-scale risks, such as pandemics and nuclear war,” reads a one-sentence statement released by the Center for AI Safety, a nonprofit organization. The open letter was signed by more than 350 executives, researchers and engineers working in A.I.

The signatories included top executives from three of the leading A.I. companies: Sam Altman, chief executive of OpenAI; Demis Hassabis, chief executive of Google DeepMind; and Dario Amodei, chief executive of Anthropic.

Geoffrey Hinton and Yoshua Bengio, two of the three researchers who won a Turing Award for their pioneering work on neural networks and are often considered “godfathers” of the modern A.I. movement, signed the statement, as did other prominent researchers in the field. (The third Turing Award winner, Yann LeCun, who leads Meta’s A.I. research efforts, had not signed as of Tuesday.)

The statement comes at a time of growing concern about the potential harms of artificial intelligence. Recent advancements in so-called large language models — the type of A.I. system used by ChatGPT and other chatbots — have raised fears that A.I. could soon be used at scale to spread misinformation and propaganda, or that it could eliminate millions of white-collar jobs.

Eventually, some believe, A.I. could become powerful enough that it could create societal-scale disruptions within a few years if nothing is done to slow it down, though researchers sometimes stop short of explaining how that would happen.

We are having trouble retrieving the article content.

Please enable JavaScript in your browser settings.

Thank you for your patience while we verify access. If you are in Reader mode please exit and  log into  your Times account, or  subscribe  for all of The Times.

Thank you for your patience while we verify access.

Already a subscriber?  Log in .

Want all of The Times?  Subscribe .

UK signs first international treaty on artificial intelligence

The EU and US have also signed the treaty, which focuses on protecting human rights, democracy and the rule of law from the potential threats of AI.

The UK has joined the EU and the United States in signing the first international treaty on artificial intelligence, which commits nations to protecting the public from potential dangers linked to the technology.

It is the first legally binding international treaty on the technology, the framework of which has been agreed by human rights organisation the Council of Europe.

The treaty will require countries to monitor the development of AI and ensure the technology is managed with strict parameters, and includes provisions to protect the public and their data, human rights, democracy and the rule of law.

It will also commit nations to take action against instances were the misuse of AI models is uncovered.

Many campaigners have called for greater regulation of the rapidly evolving technology, with calls for the UK to introduce an AI Bill to set out rules about the development and use of the technology.

No such specific legislation has yet been brought forward, although the UK hosted the first AI Safety Summit, attended by world leaders and tech giants, at Bletchley Park last year, and co-hosted a virtual summit with South Korea earlier this year – events where some non-binding international agreements on safety and AI monitoring were made.

Lord Chancellor and Justice Secretary, Shabana Mahmood, said of the new treaty: “Artificial intelligence has the capacity to radically improve the responsiveness and effectiveness of public services, and turbocharge economic growth.

“However, we must not let AI shape us – we must shape AI.

“This convention is a major step to ensuring that these new technologies can be harnessed without eroding our oldest values, like human rights and the rule of law.”

Under the treaty, countries will be required to ensure that human rights are protected, including that people’s data is used appropriately, their privacy is respected and that AI does not discriminate against them.

In addition, the treaty requires that nations take steps to ensure public institutions are not undermined and to protect the rule of law by asking countries to regulate AI-specific risks and protect citizens from potential harms.

Science Secretary Peter Kyle said: “AI holds the potential to be the driving force behind new economic growth, a productivity revolution and true transformation in our public services, but that ambition can only be achieved if people have faith and trust in the innovations which will bring about that change.

“The convention we’ve signed today alongside global partners will be key to that effort. Once in force, it will further enhance protections for human rights, rule of law and democracy, strengthening our own domestic approach to the technology while furthering the global cause of safe, secure, and responsible AI.”

Nearly £16m. forecasted to be wiped off budgeted income tax receipts News | 9 hours ago

Travellers between islands opt for ferries over flights News | 8 hours ago

A new housing committee ‘will cost £500,000 more a year’ News | 8 hours ago

Safety stumbling block delays development of Leale’s Yard News | 9 hours ago

Fort Richmond boundary dispute is finally settled Homepage | Sep 4