Human Dignity in the Age of AI | Yuval Noah Harari
What this covers
Watch Yuval Noah Harari in conversation with Kohei Itoh, President of @keiouniversity on the the future of artificial intelligence, and why it's crucial that human beings trust each other before they give more power to this alien technology.
Filmed in Tokyo on 16 March 2025, at Keio University's X Dignity Center.
#AI #Future #Humanism #History #YuvalNoahHarari
Source description (no synthesized summary yet).
Harari argues that AI is humanity's most urgent challenge because it moves exponentially faster than other threats, operates as an autonomous agent rather than a tool, and poses unique risks precisely because we lack trust between humans to govern it safely during this critical window.
- AI evolution occurs in days while biotechnology takes 20-30 year generations, making speed the decisive factor in threat ranking
- AI is fundamentally an agent that makes independent decisions, not a tool humans control—the first technology with this property
- The paradox of trust: leaders won't slow AI development because they distrust competitors, yet simultaneously claim they'll trust the superintelligent AIs they're creating
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
Humans have zero experience with trusting superintelligent AIs and no idea what goals and tricks they might develop or what happens when millions of superintelligent AIs interact with each other and with millions of human beings, whereas humans do have experience building trust across billions of people through global trade networks and other mechanisms over thousands of years.
“because you know with humans I understand why it's difficult to trust the other humans uh but we have thousands of years of experience with other human beings we understand their psychology their motivation we know a lot about the human craving for power but we also understand the mechanisms to check the pursuit of power and over thousands of years humans have developed veloped ways to learn to trust each other. And you now have, you know, 10,000 years ago or 100,000 years ago, humans lived in tiny bands of a few dozen individuals couldn't trust anybody outside their band. Now you have nations like Japan with hundreds of millions of citizens that can trust each other. The global trade network connects all 8 billion people on the planet. The food we eat, the clothes we wear, the medicine that protect us is often produced by strangers on the other side of the planet. So yes, there there's still a huge problem how to trust other people, but we have experience in doing that. We have zero experience with super intelligent AIs. We have no idea what kind of goals and tricks they might develop. We have no idea what happens when you have millions and millions of super intelligent AIs interacting with each other and with millions of human beings.”
AI is not evil, but alien—it is an intelligence that could be superhuman in many ways but is fundamentally not human and not organic, making it impossible to predict how it will develop in the same way we cannot predict when airplanes will reach bird-level flight because they operate through completely different mechanisms.
“The key problem with AI is not that it is evil. It's not evil. The key problem is that it is alien. Um, it is an intelligence which could be superhuman in many ways, but it's not human. It's not organic. A lot of people ask when will AI reach human level intelligence. The answer is never because it's not like us at all. It's like asking when will airplanes reach bird level flight. Never. Airplanes are just not like birds. They fly in a different way.”
Truth is costly because it requires research, fact-checking, and time, whereas fiction is cheap because you can write anything that comes to mind; truth is complicated because reality is complicated, whereas fiction can be made as simple as desired; and truth is often painful, whereas fiction can be made flattering, so in competition between costly/complicated/painful truth and cheap/simple/attractive fiction, fiction wins and floods the world.
“And uh the main thing to understand is that information isn't truth. The truth is a very rare subset of information. If you think about all the information in the world, so the truth is a very very small percentage of that. Most information is junk. Most information is fiction and fantasy and and lies and propaganda and so forth. And there is a very it's easy to understand why the truth is costly. Whereas fiction is cheap. If you want to write a truthful account of anything, you need to do research. You need to uh spend time on analyzing it, on factchecking it. It takes time and effort and money. Fiction is very cheap. You just write anything that comes up in your mind very cheap. Secondly, the truth tends to be complicated because reality is complicated. Like if you want to write an account about quantum physics, it's complicated. Whereas f and people usually don't like complicated narratives. They want simple stories. And again, fiction can be made as simple as you would like it to be because you just invent it. And finally, the truth is often painful. Not always but uh some parts of of reality are uncomfortable you know from the level of individual. This is why people go to therapy because if it was everything about me was so pleasant and fun I wouldn't need a therapist to help me acknowledge the painful memories or painful patterns that I that I have inside myself. And this is true all the way to entire nations that often don't want to recognize some painful parts of their history or of their society. So the truth is sometimes painful. Fiction can be made as flattering as you would like it to be. So in this competition between costly and complicated and painful truth and cheap simple attractive fiction, fiction wins and the world is flooded by enormous amounts of fiction.”
The task of institutions like universities and newspapers is not to provide people with more information (they have too much of it) but to provide them with a method to find rare gems of truth inside the ocean of information and to know how to tell the difference between reliable and unreliable sources.
“So in this competition between costly and complicated and painful truth and cheap simple attractive fiction, fiction wins and the world is flooded by enormous amounts of fiction. The job the task of institutions like universities, like newspapers, uh is not just to provide people with more information. They have too much of it. to provide them with h a method to find these rare gems of truth inside the ocean of information and to know how to tell the difference.”
Humans have a unique ability to organize very large cooperation between thousands and millions of people through shared stories and shared narratives, not just about gods and religions but also in economics and commerce, with money being the most successful story ever invented because it is the only story that everybody believes
“our special ability as homo sapiens is to organize very large cooperation between thousands and millions of millions of people through shared stories, through shared narratives, not just about gods and religions, but also in the economy. Like the most successful story ever invented is probably money. That it's the only story that everybody believes.”
Homo sapiens' special ability is to organize very large cooperation between thousands and millions of people through shared stories and narratives, not just religious ones but also economic narratives like money, which is the most successful story ever invented because it's the only story that everybody believes regardless of other ideological disagreements.
“So yes, um, our special ability as homo sapiens is to organize very large cooperation between thousands and millions of millions of people through shared stories, through shared narratives, not just about gods and religions, but also in the economy. Like the most successful story ever invented is probably money. That it's the only story that everybody believes. Not everybody believes in in God, in this God or that God, but even people who completely disagree about other things. You know, you look at the United States, Republicans and Democrats cannot agree about anything except the dollar. If a Republican comes with a dollar, the Democrat will accept it and and sell whatever it is, bread or gasoline. and and vice versa.”
Nuclear war has no upside and universal understanding of its danger, whereas AI poses a more complicated challenge because it has enormous positive potential that people find difficult to balance against its dangers, making AI fundamentally harder to regulate.
“nuclear war has no upside so it has no supporters everybody it's a huge huge threat especially In today's world when we see that countries like Russia are just completely disregarding the biggest taboos of the international system. Um since 1945 the biggest taboo of the international system was that a strong country cannot just invade and try to conquer and destroy a weaker country. And now this is broken. So the danger is is is huge. But at least everybody understands the danger. There are no winners in a nuclear war. With AI, it's a much more complicated challenge because there is a huge positive potential”
The verification of AI target selections is becoming difficult or impossible because the algorithms analyze so many data points that human analysts cannot realistically review all the information, reducing human oversight to a rubber-stamp function.
“the big question is can we believe the AI? Maybe it made a mistake...Some people say, 'Oh, after the AI points to a target, we get human analysts to go over all the information and make sure there is no mistake, right?' Other people say, 'Nah, the human analysts spend something like half a minute.' Uh, nobody can really check the AI. It goes over. so so many data points. So basically they are just rubber stump”
The physical world and information layer are separate—information technology creates an additional layer built on top of physical reality, and this layer has changed how humans obtain resources (a 100,000-year-old person hunting for food now depends on global trade networks managed by information systems) without abolishing the underlying physical reality.
“Yeah. I mean the physical world is still there. I mean none of the development of information technology uh abolishes the physical world. It's another layer built on top of it. Um this has been you know if if you go back thousands of years. So all the information created by humans, whether it's laws and constitutions, whether it's religion, whether it's money, it did not replace or abolish the physical and biological reality. People still have bodies, they need to eat, they need clean water. But all this additional layer of stories, it enabled people to create things like trade networks that changed the way we we get food. Like a 100,000 years ago, you're hungry. Uh you go to the forest, to the savannah, to the to the sea, you fish, you hunt, you gather fruits, you eat. Now, very few people do that. Almost everybody depends on a global trade network which is managed by information in order to get their food.”
Humans have developed mechanisms over thousands of years to learn to trust each other, as evidenced by the scale of modern cooperation—from nations with hundreds of millions of citizens to a global trade network connecting 8 billion people who depend on strangers across the planet for food, clothes, and medicine—but humans have zero experience with superintelligent AIs and no idea what goals, tricks, or strategies they might develop
“you know, 10,000 years ago or 100,000 years ago, humans lived in tiny bands of a few dozen individuals couldn't trust anybody outside their band. Now you have nations like Japan with hundreds of millions of citizens that can trust each other. The global trade network connects all 8 billion people on the planet. The food we eat, the clothes we wear, the medicine that protect us is often produced by strangers on the other side of the planet.”
The truth is often painful—at the individual level people require therapy to acknowledge painful memories and patterns, and at the national level countries often refuse to recognize painful parts of their history or society—while fiction can be made as flattering or pleasant as desired, making fiction more attractive than truth in the competition for attention and belief
“the truth is often painful. Not always but uh some parts of of reality are uncomfortable you know from the level of individual. This is why people go to therapy because if it was everything about me was so pleasant and fun I wouldn't need a therapist to help me acknowledge the painful memories or painful patterns that I that I have inside myself. And this is true all the way to entire nations that often don't want to recognize some painful parts of their history or of their society.”
The main task of educational institutions like universities and newspapers is not to provide people with more information—they have too much—but to teach methods for distinguishing rare gems of truth from the ocean of information and how to tell the difference between reliable and unreliable sources
“The job the task of institutions like universities, like newspapers, uh is not just to provide people with more information. They have too much of it. to provide them with h a method to find these rare gems of truth inside the ocean of information and to know how to tell the difference.”
Previously all stories circulating in the world came from human minds and were selected by human editors who decided what would dominate the conversation through newspapers and TV stations, but now this editorial function is performed by algorithms on TikTok, Facebook, and Twitter, meaning the most important editors in the world are no longer human but algorithmic
“When it comes to stories, uh this is where social media is uh is most relevant that previously all the stories that circulated in the world, they came from the human mind and it was human editors who decided what stories will dominate the conversation. Like you have a newspaper or you have a TV station. So you have a human editor going over all the stories that were published today and say oh this is the most important story we'll publish this on the front page and everybody reads this and talk about it. Now this job too is done by an algorithm. If you ask who are the most important editors in the world right now they are the editors of Tik Tok and Facebook and Twitter and they are not human they are algorithms.”
Editors of social media platforms (TikTok, Facebook, Twitter) are no longer human editors deciding which stories circulate but algorithms, meaning stories are being produced by and spread by non-human intelligence rather than human editorial judgment.
“When it comes to stories, uh this is where social media is uh is most relevant that previously all the stories that circulated in the world, they came from the human mind and it was human editors who decided what stories will dominate the conversation. Like you have a newspaper or you have a TV station. So you have a human editor going over all the stories that were published today and say oh this is the most important story we'll publish this on the front page and everybody reads this and talk about it. Now this job too is done by an algorithm. If you ask who are the most important editors in the world right now they are the editors of Tik Tok and Facebook and Twitter and they are not human they are algorithms.”
Trust-building requires input from multiple disciplines including biology and psychology (to understand human animal inheritance), economics, and computer science (because all social systems are now based on information technology), because no single discipline can tackle the question alone.
“I think we need input and this is why a an interdisciplinary center is the ideal place to do it because no single discipline can can tackle this question by itself. You need input from biology and psychology uh because we are still animals. You need to understand our our animal inheritance. uh and you can learn a lot from how chimpanzees for instance build trust or or fail to build trust between themselves to understand how humans function. Um you need the input you know from economics of course from computer science because today all social systems are based on the new information technology.”
AI is the first technology in human history that is an agent, not a tool—it can make decisions by itself, invent new things, and create new weapons and strategies without human intervention, whereas all previous technologies (nuclear reactors, bombs) required humans to make decisions about their use.
“I'll say one more thing and that that's maybe the most important thing for everybody to understand about AI is that it is not a tool. It is an agent. It's the first technology we have ever created which is an agent and not a tool. An atom bomb or a nuclear reactor that prod peaceful nuclear reactor that produces electricity. This is a tool in our hands. We decide what to do with it. We decide, humans decide, oh, we will use nuclear technology to produce electricity and not to destroy city. We make the decision. The reactor itself cannot make any decision. And of course, it cannot invent the next generation of reactors or of bombs. Only humans until now could make decisions and could invent new ideas. AI is the first technology that can do that without our our our help, without our inter intervention.”
There is a huge discussion about what has caused the collapse of human trust over the last 10-20 years, which is paradoxical because humanity is in the best physical condition it has ever been—child mortality, disease, and famine all decrease—yet people are angry, frustrated, and unable to talk with each other because new technology (digital/social media) has come between humans and is now mediating most human communication through a nonhuman agent
“there is a huge discussion about what has caused this collapse of human trust over the last 10 or 20 years. It's such again it's it's a paradox because humanity is in the best condition we were ever in. You look at most of the kind of of of hard data things like child mortality, like disease, like famine, it all goes down. We are in the best situation ever. And yet people are so angry and so frustrated and can't talk with each other. What happened? What happened is that a new technology came between humans and now most human communication is mediated by a nonhuman agent [29:29]”
To maintain human trust and cooperation, governments should pass a law banning fake humans—bots that pretend to be humans—with the same strictness as laws banning counterfeit money, because the same way you cannot print fake yen and claim it is real, you should not be able to create fake humans and present them as real
“the stories are being produced by and spread by non-human intelligence to maintain uh in human trust and cooperation. I think that it's not the it's not a a a complete solution of course but the first and simplest thing every government should do is pass a law banning fake humans, banning counterfeit humans. The same way there is a law that bans counterfeit money, you cannot just print yens and and and pretend this is real. No, you can't. Same thing with human beings.”
The elite/populist division is a false dichotomy—every human group has an elite and the real question is not whether elites exist but whether they are serving elites (using power for society's benefit) or self-serving elites (using power only for personal benefit), with communism being a failed attempt to create a society without elites that always results in one elite person with all power.
“So like many divisions that we invent in our minds, this division also is is a false dichotomy like the elite and and the people. You know, you have the richest person in the world, Elon Musk and a group of other billionaires like Donald Trump who is a billionaire claiming that they are against elites. If you're not an elite, so what is an elite? Like the richest person in the world is not an elite. All these billionaires who are now running the US government, oh, they are not elites. Why? I mean, the word elite has become kind of a of a negative term that you just put on on your political rivals.”
The post-WWII international order was the most peaceful and prosperous era in human history, largely because the taboo against strong countries invading weaker ones was maintained, which can be seen in government budget allocations: for most of history, over 50% of government budgets went to military spending, but in the early 21st century this declined to 6-7% while healthcare spending rose to about 10%, the first time governments spent more on healthcare than military
“And humanity enjoyed in the early 21st century the most peaceful and prosperous era in its history largely because this taboo was maintained...You could see it most easily in the budgets of governments. For most of history, more than 50% of the budget of every king and shogun and emperor was on the military, soldiers, warships, castles. This was the main government expenditure. In the early 21st century, average government expenditure on the military went down to about 6% or 7%. This amazing achievement about 10% went to healthcare. It was the first time in human history that governments all over the world spent more on health care than on the military.”
Republicans and Democrats in the United States cannot agree about almost anything, except the dollar—when a Republican with a dollar bill meets a Democrat, the Democrat will accept it and sell them bread or gasoline, demonstrating that money transcends political divisions by being a universally accepted story
“Not everybody believes in in God, in this God or that God, but even people who completely disagree about other things. You know, you look at the United States, Republicans and Democrats cannot agree about anything except the dollar. If a Republican comes with a dollar, the Democrat will accept it and and sell whatever it is, bread or gasoline. and and vice versa.”
Conspiracy theories assuming small groups of people can secretly control world events are always wrong because the world is too complicated for any small number of people to understand and manipulate effectively, as demonstrated by historical examples like the U.S. invasion of Iraq, where even the world's strongest country with overwhelming military power failed to execute its plan and Iran unexpectedly became the war's beneficiary
“The one thing I can say as a historian is that these types of conspiracies are always wrong for the simple reason the world is too complicated for any small number of people to really understand it and manipulate it...You look at how governments even if you have a government with all the power of a government that tries to do something very often it fails. Like I don't know the Americans decide to invade Iraq. They are the strongest country in the world. They have all these military power and they fail. Yes, they they manage to conquer Iraq. Then everything goes off plan and eventually it's Iran that becomes the the the the main bene beneficiary, the victor of the war.”
The 2016 game between AlphaGo and Lee Sedol was significant not just because AlphaGo won, but because it demonstrated an alien intelligence that discovered entirely new ways to play Go that humans had never conceived despite playing for over 2,000 years—humans had been stuck on one 'mental island' unable to escape the limitations of human cognition, but AI revealed entire unexplored regions of the game space
“It came up with strategies that were completely alien [1:02:19] that, you know, humans have been playing go in East Asia for thousands of years. It was invented more than 2,000 years ago. Tens of millions of people played Go. It was considered a form of out, not just a game. Entire philosophies developed around how to play Go. And yet all those years, all those people, nobody ever thought of playing Go, like Alph Go. So you can imagine it like a like a landscape, like a planet, planet Go. All the different ways to play Go. And humans have been stuck on one island in planet Go. And they thought this was the whole planet for 2,000 years. They saw these are the only ways to play Go. They couldn't leave this this mental island because our minds are limited. Then Alph Go came and said look you can play like this you can play like this you can discovered entire new areas on the planet go.”
AI development moves millions of times faster than organic biological evolution because digital evolution is not constrained by generational reproduction cycles like biology is, whereas a generation in biotechnology applied to humans requires 20-30 years to observe genetic effects, while AI iterations occur over days or weeks
“digital evolution is moving millions of times faster than organic evolution”
AI moves millions of times faster than organic evolution, making it a more pressing threat than biotechnology despite both having world-changing potential, because a generation in biotechnology for humans is 20-30 years whereas in AI it could be just a few days.
“digital evolution is moving millions of times faster than organic evolution”
The purpose of educational institutions should be to create a serving elite—people who have more talents in a particular field, have more influence, and use their talents and power not just for personal benefit but for the benefit of all.
“I think especially for an institution like this and for all universities the purpose is not to destroy elites or break down all barriers between this group. The purpose is to create a serving elite. Yes. that people who they have more talents in a particular field, they have more influence in that particular field and they use their talents and their power not just for their personal benefit but for the benefit of all that should be the aim.”
Information is not truth—truth is a very rare subset of information, a very small percentage of all information in the world, while most information is junk, fiction, fantasy, lies, and propaganda.
“Now it's exactly the opposite. People don't need more information. We are flooded by enormous amounts of information. And uh the main thing to understand is that information isn't truth. The truth is a very rare subset of information. If you think about all the information in the world, so the truth is a very very small percentage of that. Most information is junk. Most information is fiction and fantasy and and lies and propaganda and so forth.”
History teaches not facts about the past but how to tell the difference between a reliable historical source and an unreliable source, and this method applies to evaluating sources from the internet to documents from the Middle Ages, because people lied a thousand years ago just as they lie today.
“So, you know, as a historian, the main thing I teach my students in history class is not facts about the past. Oh, this king won that battle against that king in that year. They don't need me. they can just Google already. Uh what they still need to learn in history department is how do I tell the difference between a reliable historical source and an unreliable source? A source may be something I just read today on the internet. How do I know if I can trust this Tik Tok video or not? A source can be something from the Middle Ages. You find a document from a thousand years ago. How do you know the king wrote that he won this battle? How do you know whether to believe because people lied also a thousand years ago not just today? So that's the main question”
Conspiracy theories claim that small groups of powerful people can orchestrate world events and take over the world, but this is unrealistic because the world is too complicated for any small number of people to understand and manipulate—even governments with all their power often fail in their plans, as seen with the US invasion of Iraq which failed to achieve its objectives and instead benefited Iran.
“The one thing I can say as a historian is that these types of conspiracies are always wrong for the simple reason the world is too complicated for any small number of people to really understand it and manipulate it. The level of manipulation that conspiracy theories usually assume is completely unrealistic. You look at how governments even if you have a government with all the power of a government that tries to do something very often it fails. Like I don't know the Americans decide to invade Iraq. They are the strongest country in the world. They have all these military power and they fail. Yes, they they manage to conquer Iraq. Then everything goes off plan and eventually it's Iran that becomes the the the the main bene beneficiary, the victor of the war. So the Americans had one plan, something completely different happened. This is how the actual world functions.”
The scientific revolution was not a revolution of knowledge but a revolution of ignorance—scientists began saying 'we don't know' and 'we want to know' rather than claiming all knowledge was already in holy books or from authorities.
“For me, the scientific revolution looking back in history was a revolution of ignorance, not of knowledge. Before science, every human group, every human culture said, "We already know everything. Everything is in the holy books or our holy men and sometimes holy women know everything. No need of new research. The scientific revolution was about people coming and saying we don't know. That's a very we want to know this the answer to this question and we don't know what it is. Oh so we need to do research and this is still the case right now.”
The international order established after World War II was based on the taboo that strong countries cannot invade and conquer weaker countries, which had been the normal pattern for thousands of years, and this taboo was the foundation of the most peaceful and prosperous era in human history.
“its most important taboo its most important rule was that strong countries cannot just invade and conqu weaker countries. And this was in complete uh opposite to thousands of years of previous history when you know for thousands of years this was this was the ordinary thing that a stronger country would just invade and conquer its neighbors and establish a bigger country an empire.”
Viruses are ontologically strange—not quite alive and not quite not alive—existing as genetic information that hijacks host cells to reproduce, making them difficult for humans to understand intuitively because viral replication has no visible equivalent at human-perceptible scales.
“viruses are not exactly alive and not exactly not alive. Uh there is a huge debate in the scientific community. Are viruses a form of life or are they just are they part of biology or are they actually just some chemical information? Because a virus cannot reproduce by itself...It's a virus is really just a bit of information. And the the amazing ability of this bit of genetic code is to get into the cells of my body and tell my own body, my own cells to start producing more copies of that code...It's happening on a very very uh uh uh small level of reality.”
In 2016, AlphaGo's victory over world champion Lee Sedol in Go was significant not just because it defeated a human but because it played Go in completely alien strategies that humans had never discovered despite thousands of years of the game's history and millions of players—AlphaGo discovered entire new areas on the 'planet of Go' that humans were unable to reach due to biological brain limitations.
“In 2016 there was this very famous game between Alph Go and AI and Lisa one of the champ the go world champion and Alph Go won. It was one of the decisive moments in the AI revolution. But the amazing thing about the game wasn't just that Alph Go defeated the UN world champion. It's the way he did it. It came up with strategies that were completely alien that, you know, humans have been playing go in East Asia for thousands of years. It was invented more than 2,000 years ago. Tens of millions of people played Go. It was considered a form of out, not just a game. Entire philosophies developed around how to play Go. And yet all those years, all those people, nobody ever thought of playing Go, like Alph Go. So you can imagine it like a like a landscape, like a planet, planet Go. All the different ways to play Go. And humans have been stuck on one island in planet Go. And they thought this was the whole planet for 2,000 years. They saw these are the only ways to play Go. They couldn't leave this this mental island because our minds are limited. Then Alph Go came and said look you can play like this you can play like this you can discovered entire new areas on the planet go.”
AI is fundamentally an alien form of intelligence that will never achieve human-level intelligence because the comparison is category-confused—it is more accurate to ask when airplanes will achieve bird-level flight (never) because they operate on entirely different principles, and therefore AI superintelligence will be inherently alien, making its future development and consequences very difficult to predict
“The key problem with AI is not that it is evil. It's not evil. The key problem is that it is alien. Um, it is an intelligence which could be superhuman in many ways, but it's not human. It's not organic. A lot of people ask when will AI reach human level intelligence. The answer is never because it's not like us at all. It's like asking when will airplanes reach bird level flight. Never. Airplanes are just not like birds. They fly in a different way. So AI will surpass human intelligence and will but it will be inherently alien.”
AI is the first technology in human history that is an autonomous agent rather than a tool, because unlike nuclear reactors or bombs which humans must decide how to use and which cannot make decisions or invent new technologies themselves, AI can make independent decisions and invent new weapons, medicines, and strategies without human intervention or authorization
“I'll say one more thing and that that's maybe the most important thing for everybody to understand about AI is that it is not a tool. It is an agent. It's the first technology we have ever created which is an agent and not a tool. An atom bomb or a nuclear reactor that prod peaceful nuclear reactor that produces electricity. This is a tool in our hands. We decide what to do with it...Only humans until now could make decisions and could invent new ideas. AI is the first technology that can do that without our our our help, without our inter intervention. AI can make decisions by itself.”
The physical world still exists and is not abolished by the development of information technology—rather, information technology creates an additional layer built on top of physical reality, and all information created by humans including laws, constitutions, religions, and money has always been layered on top of physical and biological reality
“the physical world is still there. I mean none of the development of information technology uh abolishes the physical world. It's another layer built on top of it. Um this has been you know if if you go back thousands of years. So all the information created by humans, whether it's laws and constitutions, whether it's religion, whether it's money, it did not replace or abolish the physical and biological reality. People still have bodies, they need to eat, they need clean water.”
Viruses are ontologically strange entities—neither alive nor not alive—that work by injecting genetic code into cells that then replicate that code, making them fundamentally different from how macroscopic life works, which is why understanding pandemics requires grasping this unfamiliar biological mechanism
“Uh there is a huge debate in the scientific community. Are viruses a form of life or are they just are they part of biology or are they actually just some chemical information? Because a virus cannot reproduce by itself. A virus doesn't eat. It's not like a bacteria that eats things and reproduces. It's a virus is really just a bit of information. And the the amazing ability of this bit of genetic code is to get into the cells of my body and tell my own body, my own cells to start producing more copies of that code.”
The uncertain question is what the role of humans will be in scientific research in 10-20 years given AI's accelerating capabilities and pace of development, which makes it difficult to predict what humans could still contribute to research.
“and it will the question then becomes so what would be the role of humans in in such a situation and again we don't know I mean given the the the the potent the pace of development and the potential it's not clear what would be still what could humans contribute ute to research uh in 10 or 20 years.”
Elon Musk and Donald Trump are simultaneously firing tens of thousands of human bureaucrats while replacing them with digital bureaucrats (algorithms), demonstrating a trend toward shifting power from humans to algorithms rather than actually reducing bureaucracy itself
“we talked about bureaucracy earlier. So you see at the same time that Elon Musk and Donald Trump they are firing tens of thousands of human bureaucrats but they are not really against bureaucracy because they simply replace them with digital bureaucrats with algorithms. Um and this is the shifting power from humans to algorithms.”
The division between 'elite' and 'people' is a false dichotomy; every human group necessarily has an elite because hierarchies of talent, power, and influence are unavoidable features of human organization, even in a football club; the real question is not whether elites exist but whether they are serving elites (using power for collective benefit) or self-serving elites (using power for personal benefit)
“Like many divisions that we invent in our minds, this division also is is a false dichotomy like the elite and and the people. You know, you have the richest person in the world, Elon Musk and a group of other billionaires like Donald Trump who is a billionaire claiming that they are against elites.”
In recent wars including the Russian invasion of Ukraine and the Israeli-Palestinian war, AI has primarily changed not the execution of killing (which remains human) but the selection and decision-making of targets—identifying targets now takes seconds via AI analysis of massive data rather than days of human intelligence analysis, and increasingly AI is making these targeting decisions with minimal human oversight.
“War ultimately is still about killing people but the way it is done is changing. Um, what we saw in recent wars, not just in the Russian invasion of Ukraine, also I know it more from the war in my country between Israelis and Palestinians, a lot of people expected AI to change war by doing the actual shooting. Like people watched all these Hollywood movies about killer robots. So they now look on television and they don't see killer robots. The K the killing is still done by human beings. What really changed already with AI is who is choosing the targets, who is making the decisions, who decided to bomb this building, who decided to bomb this person and increasingly the answer is AI.”
As AI becomes sophisticated enough to produce entertainment content (poetry, theater, television, movies), more and more of both the form and content of entertainment will come from AI rather than human imagination, since AI can produce such content more cheaply and efficiently than humans and is not limited by biological brain limitations that constrain human imagination.
“Um, and we we don't know, we can't predict, but it's very clear that more and more both the form and the content of entertainment will come from AI and not from human beings. Previously in history, all entertainment, whether it's poetry or theater, television, movies, everything came out of the human mind, of the human imagination. Now we have something on the planet that is capable of producing such things uh perhaps certainly far more cheaply than human beings and far more efficiently and also break out of the limitations of the human imagination because the human imagination is limited by the human brain by our own biological limitations. AI isn't.”
War ultimately is still about killing people, but what AI is changing about recent wars like Russia's invasion of Ukraine and the Israeli-Palestinian conflict is not that killer robots are doing the shooting (despite Hollywood depictions), but rather who is choosing the targets and making the decisions, with AI increasingly replacing human intelligence officers in target identification
“War ultimately is still about killing people but the way it is done is changing. Um, what we saw in recent wars, not just in the Russian invasion of Ukraine, also I know it more from the war in my country between Israelis and Palestinians, a lot of people expected AI to change war by doing the actual shooting. Like people watched all these Hollywood movies about killer robots. So they now look on television and they don't see killer robots. The K the killing is still done by human beings. What really changed already with AI is who is choosing the targets, who is making the decisions, who decided to bomb this building, who decided to bomb this person and increasingly the answer is AI.”
Harari's books are intentionally written to be accessible to general audiences rather than just academic experts, translating the most up-to-date theories and findings from academic and scientific communities into language that high school students or any person can understand without requiring background in computers, graphs, or equations
“Your book is so readable. Uh it's very different from academic writing so so to say. Do you do this intentionally? Absolutely. I mean I see my job as building a bridge between the uh uh uh world of academic studies where mostly experts talk with other experts in a language that is uh very difficult for lay people without academic background to access. So I try to take the most upto-date theories and models and findings of the academic and scientific community and uh relay it explain it in a way that will be easily accessible say even to high school students or to any person who wants to understand”
Many of the things discussed in Harari's talk sound frightening, but part of the problem is that AI developers and corporate leaders talk only about positive potential of AI, so philosophers, historians, and critics are necessary to balance by highlighting dangers—the question is not how to stop AI development (which is unrealistic and undesired) but how to develop it safely
“So, there is a lot to to think about. there is a lot to research. Again, the the key thing many of the things we talked about sound very frightening. Uh part of of of the problem is you have you have the people who in the companies that develop the AI that talk about all the positive potential of AI. So be to to balance it, you need philosophers and historians and critics who say, 'Wait a minute, there are also dangers.' So I find myself in occasions like this talking mainly about the dangers. Uh but of course there is also enormous positive potential otherwise nobody would develop the technology in fields ranging from medicine to uh uh preventing catastrophic climate change and it's definitely there. There is enormous positive potential. Uh the question is not how to stop the development of AI. This is unrealistic and and undesired. We want to develop it. We just want to do it in a safe way.”
The weight of addressing global problems does not fall entirely on individuals—each person bears only a share of responsibility as distributed among 8 billion humans (plus other organisms), so no one person should feel crushed by trying to carry the entire world on their shoulders; instead, each person should do what they can in their specific field of activity
“There is a lot that every person can do in your specific field. So you know sometimes people feel crushed by the weight of the world that oh all these problems how can we I don't feel like the whole world is just on your shoulders. There are eight more than eight billion other humans on the planet. So they share in this weight of holding the planet.”
Human nature has the ability to self-correct and cooperate together, which provides hope that it's not too late to address AI risks if humanity acts cooperatively before AI reaches a certain capability threshold.
“at the same time you know it gave me hope that it's not too late and there's a you know human nature that is you know that can act ability to self-correct and also ability to co cooperate work together um if only we had time I I suppose right before AI gets somewhere else.”
No single person should bear the weight of the world alone; responsibility for addressing global challenges is distributed across more than eight billion humans and all organisms on the planet, with each doing their small part.
“There is a lot that every person can do in your specific field. So you know sometimes people feel crushed by the weight of the world that oh all these problems how can we I don't feel like the whole world is just on your shoulders. There are eight more than eight billion other humans on the planet. So they share in this weight of holding the planet. And of course there are not just humans, there are also animals and plants and other organisms. And each one doing their small little thing.”
There is enormous positive potential in AI development in fields ranging from medicine to preventing catastrophic climate change, and the question is not how to stop AI development (which is unrealistic and undesired) but how to develop it in a safe way.
“The question is not how to stop the development of AI. This is unrealistic and and undesired. We want to develop it. We just want to do it in a safe way. And here again the key is to build trust between human beings. There is a lot that every person can do in your specific field. So you know sometimes people feel crushed by the weight of the world that oh all these problems how can we I don't feel like the whole world is just on your shoulders. There are eight more than eight billion other humans on the planet. So they share in this weight of holding the planet. And of course there are not just humans, there are also animals and plants and other organisms. And each one doing their small little thing. When I think about trust, which was maybe the major theme of of our meeting, maybe the I I would leave you with with the the following thought that um trust is really the foundation of life”
Social media companies have the responsibility similar to newspapers and TV stations before them to ensure that what they disseminate is the truth and not hateful lies, and this responsibility cannot be evaded by claiming algorithmic neutrality or freedom of speech protections
“It is the job of the social media companies the same way the job of newspapers and TV stations before they disseminate something to make sure that it is the truth and it is not some hatefilled lies.”
Military spending decreased from over 50% of government budgets historically to about 6-7% by the early 21st century, while healthcare spending reached about 10%, the first time governments spent more on health than on military—this is now reversing due to Russia-Ukraine and other developments.
“For most of history, more than 50% of the budget of every king and shogun and emperor was on the military, soldiers, warships, castles. This was the main government expenditure. In the early 21st century, average government expenditure on the military went down to about 6% or 7%. This amazing achievement about 10% went to healthcare. It was the first time in human history that governments all over the world spent more on health care than on the military. This is now going in reverse.”
Freedom of speech (a human right protecting individuals' right to express views without fear of punishment) must be distinguished from freedom of information (a new right given to AIs, algorithms, and bots), and social media companies incorrectly conflate these two rights when using freedom of speech to justify algorithmic amplification of hate and disinformation
“When it comes to fake news and and disinformation and all that, we need to differentiate between freedom of speech [46:07] and freedom of information, freedom of speech is given to human beings. It's a human right, a very important human right which should be protected. uh whether it is here in this hall, whether it is on on Tik Tok or Facebook or anywhere else, human beings except in extreme situations should have the right to express their views without fear of being punished for it. This should be protected. Uh f of information is something very different. It's a new right uh which is given not to human beings, it's given to AIS and algorithms and bots.”
There is a paradox at the heart of the AI revolution: AI leaders say they cannot slow down development because if they do their competitors will win the race and the most ruthless people will dominate the world, but then when asked if they can trust the superintelligent AIs they are developing, they say yes—despite claiming they cannot trust other humans.
“We understand there are huge risks as as as big as human extinction. We understand this. We would like to slow down but we can't because if we slow down and our competitors in the other company in the other country they don't slow down they will win the race and the most ruthless people will then dominate the world”
Despite the many dangers and reasons for anxiety about AI, the foundation of life itself is trust—with every breath humans take in air from outside themselves, they perform a gesture of trust in what is outside; complete separation from the outside world (blocking mouth and nose) results in death within a minute or two, making trust the most fundamental aspect of existence
“When I think about trust, which was maybe the major theme of of our meeting, maybe the I I would leave you with with the the following thought that um trust is really the foundation of life in the sense that if you don't trust others, if you don't trust what is outside you, you cannot live for a single in it because even the air we breathe like this simple movement that sustains life the breath [1:24:19] what is it? It is trusting in what is outside us.”
Trust is the foundation of life itself in a literal biological sense—trust in what is outside us (breathing air from the environment, taking in food) sustains biological life; separation and barriers block this trust and lead to death within minutes.
“When I think about trust, which was maybe the major theme of of our meeting, maybe the I I would leave you with with the the following thought that um trust is really the foundation of life in the sense that if you don't trust others, if you don't trust what is outside you, you cannot live for a single in it because even the air we breathe like this simple movement that sustains life the breath what is it? It is trusting in what is outside us. You know you have all these people today including the most powerful people in the world that they want to be separate. All they think about is separating things, building barriers and borders and walls. And actually complete separation is death. Every moment we have this simple gesture of trust in what is outside us. We open our mouth or our nose and we take in something from the outside. We have trust in it and then we give it back. And this is the in and out movement of trust that sustains life. If you don't trust what is outside you and you just block your mouth and your nose, you die within a minute or two.”
The most urgent research question at an interdisciplinary center should be how to build trust between humans, because the biggest danger in the world is still distrust between humans—if humans could trust each other and cooperate, they could devise ways to develop AI safely, but it is human distrust that makes the AI revolution so dangerous
“How to build trust between humans? I think that is the most urgent because what we see now in the world is a collapse of of human trust exactly when we need it most to the biggest danger in the world still is not AI it's the distrust between humans if humans could trust each other and cooperate we could still deise ways to develop AI in a safe way it's the distrust between humans that make the AI revolution so [27:24]”
The most urgent research question is how to build trust between humans, not in a general sense, but specifically in an era when communication between humans is being twisted and manipulated by new information technology.
“How to build trust between humans? I I think that is the most urgent because what we see now in the world is a collapse of of human trust exactly when we need it most to the biggest danger in the world still is not AI it's the distrust between humans if humans could trust each other and cooperate we could still deise ways to develop AI in a safe way it's the distrust between humans that make the AI revolution so dangerous.”
AI is revolutionizing scientific research itself, as demonstrated by Nobel Prizes in physics and chemistry going to AI developers, with particularly notable example of AlphaFold solving the protein folding prediction problem—a discovery that will change biology, medicine, and drug production—showing that AI developers without deep chemistry backgrounds can create tools that transform entire scientific fields
“AI is is now revolutionizing scientific research. As you said, we just saw Nobel prizes, two Nobel prizes in physics and in chemistry actually go to AI. And I was especially amazed by the AI Nobel Prize in chemistry which was given to the developers of Alpha Fold. And I mean they are brilliant people. I'm not sure they are chemists. I mean they had they needed deep understanding of chemistry to develop alpha of course but their background is not in chemistry and but their creation completely changed chemistry uh and will change biology and will change medicine production by solving this huge scientific puzzle of predicting how proteins fold.”
Large parts of financial trading (especially in bonds, stocks, and foreign exchange) are already conducted by algorithms trading with other algorithms, and with deregulation by the Trump administration this trend will likely accelerate toward new financial devices and kinds of money that only AIs understand and trade.
“Even today, a large part of the trade in the world, especially trade in bonds and stocks and foreign exchange, it's just algorithms trading with other algorithms. and especially if uh uh humans deregulate the financial markets which is what the Trump administration is all about. So we are likely to see new financial devices, new kinds of money which only AIs understand and AIs trade with other AIs.”
More and more of both the form (how things are created) and content (what is created) of entertainment will come from AI rather than human imagination, previously all entertainment from poetry to theater to television came from human minds, but now AI is capable of producing entertainment perhaps far more cheaply and efficiently than humans while breaking free from the limitations of human imagination constrained by biological brain capacity
“Um, and we we don't know, we can't predict, but it's very clear that more and more both the form and the content of entertainment will come from AI and not from human beings. Previously in history, all entertainment, whether it's poetry or theater, television, movies, everything came out of the human mind, of the human imagination. Now we have something on the planet that is capable of producing such things uh perhaps certainly far more cheaply than human beings and far more efficiently and also break out of the limitations of the human imagination because the human imagination is limited by the human brain by our own biological limitations. AI isn't.”
All hopes of reaching a global agreement or international regulation on AI development are now extremely naive and have effectively collapsed, particularly since the last US elections, because countries locked in competitive dynamics cannot afford to slow their own AI development if competitors will not, creating a prisoner's dilemma where even leaders who understand the risks feel forced to accelerate
“The other thing is that all hopes of uh regulating this of reaching some kind of agreement about this seem now extremely naive. Certainly since the last US elections, um I think really all hope of reaching a global agreement on how to manage the risk of AI development is is is gone.”
The Trump administration's worldview reflects a logic that 'the weak should obey the strong,' where if weak countries refuse to obey a strong country and conflict results, the weak are to blame for the resulting war because they should have obeyed—this logic, if it becomes dominant, will bring all humankind back to a constant state of war
“You listen to Trump blaming Ukraine for the war and it's very clear that what he has in mind is a world order in which the proper thing is for the weak to obey the strong. Right? And if the weak refuse to obey the strong, the weak are to blame for the resulting conflict. Right? And you see it like all it threatens that the United States now wants to conquer Greenland. Now if Denmark and Greenland say but we don't want to be part of the US and as a result there is a war in Trump's mind who is to blame for that Denmark is to blame because United States is strong you must obey the strong if you don't obey the strong and there is a war the weak are to blame they should have obeyed the strong now this kind of worldview this kind of logic will bring all humankind back to a constant state of war.”
Russia's invasion of Ukraine broke the most important taboo of the post-WWII international system, which will lead to a return to constant state of war and to governments abandoning healthcare and education budgets in favor of military spending.
“I come from the Middle East. I come from Israel. I know this perfectly well. Though it was not a perfect order, but it was better than anything that existed previously. You could see it most easily in the budgets of governments. For most of history, more than 50% of the budget of every king and shogun and emperor was on the military, soldiers, warships, castles. This was the main government expenditure. In the early 21st century, average government expenditure on the military went down to about 6% or 7%. This amazing achievement about 10% went to healthcare. It was the first time in human history that governments all over the world spent more on health care than on the military. This is now going in reverse. As a result of Russia's invasion of Ukraine and of other developments, you see defense budgets skyrocket”
Nuclear war has no upside and no supporters, making it a clear taboo, whereas AI has enormous positive potential that makes regulation much more complicated because people are attracted to its benefits even while underestimating its risks.
“with regard when comparing AI say to the threat of nuclear war here the reason to focus on AI actually two reasons first of all nuclear war has no upside so it has no supporters everybody it's a huge huge threat especially In today's world when we see that countries like Russia are just completely disregarding the biggest taboos of the international system. Um since 1945 the biggest taboo of the international system was that a strong country cannot just invade and try to conquer and destroy a weaker country. And now this is broken. So the danger is is is huge. But at least everybody understands the danger. There are no winners in a nuclear war. With AI, it's a much more complicated challenge because there is a huge positive potential and um we don't and people find it very difficult to understand what the threats are.”
Horses have been bought and sold for money for thousands of years, but horses never understood what money is—they just see a samurai give a piece of metal to a merchant and receive the horse in return but cannot comprehend the transaction—and humans could control horses because horses do not understand the system governing their exchange, and similarly humans could find themselves in the horses' position with respect to financial systems they cannot understand
“Like think about it from you know think about horses and money. People have been buying and selling horses for money for thousands of years. But the horses never understood what was happening. The host could see that say one samurai is give is is giving some piece of shiny metal to a merchant and gets me the horse in return. But what was happening there? What is this worthless piece of metal? The horses never understood money. Only humans understood them. Which is why we control the horses and not vice versa. In 10 years, our life could be managed by financial transactions between algorithms that we can't understand”
AI should be conceptualized not as a single big computer but as a wave of millions of alien immigrants sweeping over the entire world, which parallels and vastly exceeds populist fears about human immigration—AI will take people's jobs, have different ideas about running societies, and could take over countries, making it a threat that should frighten people much more than human immigration
“AI is not one single big computer somewhere. It's going to be think about it as a wave of immigration sweeping over the entire world. You now have all these populist parties that are so afraid of immigrants. Oh, the immigrants will come. The immigrants will take our jobs. The immigrants have they they have different ideas about culture. They will take over our country. They will change our culture. We are so afraid. AI is a wave of millions of alien immigrants that will take people's jobs that will have very very different ideas how about how to run societies that could take over countries. This is much should should frighten people much more”
Cryptocurrency represents a threat to the dollar as a universal narrative story because people choosing cryptocurrency over traditional currency are saying 'I don't trust the dollar, I trust Bitcoin/Ethereum,' which really means 'I don't trust humans, I trust algorithms'—a fundamental shift from human-produced currency to algorithm-produced currency
“this even this is now under threat with cryptocurrency. Yes. Yes. Uh because there are the people who say we don't want the dollar, we want crypto. And this is a very important indication because dollar is human currency. It's produced by human institutions like banks. Crypto like Bitcoin, Ethereum and so forth is a posthuman currency produced by algorithms. The people who say I don't trust dollar, I trust Bitcoin. What they actually say is I don't trust humans. I trust algorithms. I don't trust the humans who run the banks. I prefer to trust the algorithms who run the cryptocurrency networks.”
If financial transactions and their determining factors become unintelligible to humans, this creates a crisis not only for democracies (where leaders cannot explain financial crises to the public) but also for dictators (whose biggest fear is subordinates becoming more powerful—if AI becomes too powerful for them to understand, it threatens their power as much as anyone else's)
“What is the meaning say for democracies? If not even the prime minister or the finance minister understand finance anymore there there is a financial crisis but nobody no human being understands why there is a financial crisis and what can be done about it and it's not just a problem for democracies also for dictatorships. The biggest fear of every dictator in history is to have a subordinate that becomes more powerful than me. Like if you are Putin or if you're Kim, your biggest fear, the things that keeps you up at night is maybe the defense minister is plotting against me. Maybe some army general will have a coup and kill me, take my place. That's the biggest fear. And so for them if AI becomes so powerful that I can no longer understand the AI it's a threat even to human dictators.”
AI is not one single big computer but a wave of immigration sweeping over the entire world—millions of alien immigrants that will take people's jobs, have different ideas about how to run societies, and could take over countries, which should frighten people much more than the immigrant fears that populist parties exploit.
“AI is not one single big computer somewhere. It's going to be think about it as a wave of immigration sweeping over the entire world. You now have all these populist parties that are so afraid of immigrants. Oh, the immigrants will come. The immigrants will take our jobs. The immigrants have they they have different ideas about culture. They will take over our country. They will change our culture. We are so afraid. AI is a wave of millions of alien immigrants that will take people's jobs that will have very very different ideas how about how to run societies that could take over countries. This is much should should frighten people much more and um so this is the big paradox and the big question is how do you build trust between humans before we develop super intelligent AI.”
Cryptocurrency represents a fundamental shift in the nature of currency from human-created institutional money (produced by banks) to posthuman algorithmic money produced by algorithms, and when people say 'I trust Bitcoin more than the dollar,' they are actually saying 'I don't trust humans, I trust algorithms.'
“This even this is now under threat with cryptocurrency. Yes. Yes. Uh because there are the people who say we don't want the dollar, we want crypto. And this is a very important indication because dollar is human currency. It's produced by human institutions like banks. Crypto like Bitcoin, Ethereum and so forth is a posthuman currency produced by algorithms. The people who say I don't trust dollar, I trust Bitcoin. What they actually say is I don't trust humans. I trust algorithms. I don't trust the humans who run the banks. I prefer to trust the algorithms who run the cryptocurrency networks.”
Freedom of speech (a human right) must be distinguished from freedom of information (a new right given to AIs and algorithms), and while human beings should not be punished for inventing lies except in extreme circumstances, algorithms deliberately spreading lies and hate should be stopped and punished because algorithms don't have rights.
“When it comes to fake news and and disinformation and all that, we need to differentiate between freedom of speech and freedom of information, freedom of speech is given to human beings. It's a human right, a very important human right which should be protected. uh whether it is here in this hall, whether it is on on Tik Tok or Facebook or anywhere else, human beings except in extreme situations should have the right to express their views without fear of being punished for it. This should be protected. Uh f of information is something very different. It's a new right uh which is given not to human beings, it's given to AIS and algorithms and bots. When a lot of the social media giants, they intentionally confuse the two rights.”
The collapse of human trust over the past 10-20 years paradoxically coincides with humanity being in the best material condition ever (lowest child mortality, disease, famine), indicating that human satisfaction depends less on material well-being than on some other factor disrupted by technology.
“there is a huge discussion about what has caused this collapse of human trust over the last 10 or 20 years. It's such again it's it's a paradox because humanity is in the best condition we were ever in. You look at most of the kind of of of hard data things like child mortality, like disease, like famine, it all goes down. We are in the best situation ever. And yet people are so angry and so frustrated and can't talk with each other. What happened? What happened is that a new technology came between humans and now most human communication is mediated by a nonhuman agent”
Governments should pass laws banning counterfeit humans (bots pretending to be human on social media), the same way there are laws against counterfeiting money, because much of the traffic on platforms like TikTok and Twitter is run by bots pretending to be human beings.
“I think that it's not the it's not a a a complete solution of course but the first and simplest thing every government should do is pass a law banning fake humans, banning counterfeit humans. The same way there is a law that bans counterfeit money, you cannot just print yens and and and pretend this is real. No, you can't. Same thing with human beings. Until today, you couldn't create fake humans. Now, you can. Bots can pretend to be humans. Much of the traffic in places like Tik Tok and like X Twitter is run by bots that pretend to be human beings. So the the first and and simplest step to take have a very strict law. You cannot fake humans. AIS are welcome to talk with us on condition that they identify as an AI. If somebody tries to fake humans or if a platform doesn't take action against fake humans, they should go to jail.”
The pace of AI development has become astounding and unrecognizable compared to 10 years ago when Harari wrote 'Homo Deus'—at that time AI was treated as a distant philosophical speculation with little immediate practical relevance, but now corporate leaders and researchers are claiming AGI could arrive within a year to five years, with even the 'conservative' figure of 2029 (Ray Kurzweil's prediction from 2014) now considered optimistic
“When I wrote Homodos it felt like oh we are talking about something that maybe will happen in hundreds of years some philosophical speculations with very little immediate practical implications for for our lives. Uh now it's all around us and the the pace of development is simply astounding.”
AI is now revolutionizing scientific research, with two recent Nobel Prizes going to AI developers (in physics and chemistry, specifically AlphaFold for protein folding prediction), and AlphaFold's creators had deep understanding of chemistry but not traditional chemistry backgrounds, showing that AI enables expertise in multiple disciplines that individual humans cannot achieve due to time and cognitive limitations.
“Yeah. AI is is now revolutionizing scientific research. As you said, we just saw Nobel prizes, two Nobel prizes in physics and in chemistry actually go to AI. And I was especially amazed by the AI Nobel Prize in chemistry which was given to the developers of Alpha Fold. And I mean they are brilliant people. I'm not sure they are chemists. I mean they had they needed deep understanding of chemistry to develop alpha of course but their background is not in chemistry”
The book 'Nexus' focuses on information networks and AI rather than the three original threats (nuclear war, biotechnology, AI) because AI moves faster and is harder for people to understand the dangers of—nuclear war's danger is intuitive, but AI's alien nature makes the threat hard to grasp.
“when comparing AI say to the threat of nuclear war here the reason to focus on AI actually two reasons first of all nuclear war has no upside so it has no supporters... But at least everybody understands the danger. There are no winners in a nuclear war. With AI, it's a much more complicated challenge because there is a huge positive potential and um we don't and people find it very difficult to understand what the threats are.”
Conspiracy theories persist because while the truth about complex events like pandemics is genuinely complicated and requires understanding strange biological concepts like viruses, conspiracy theories offer simple explanations (powerful people created the pandemic to take over the world) that are easier to believe, because humans prefer simple narratives to complicated ones.
“To understand how a pandemic begins and spreads very complicated. It's much easier to believe a conspiracy theory. Oh, there is a group of very powerful people. They created this pandemic to take over the world.”
Social media companies should have the same responsibility as traditional newspapers and TV stations to verify that content they disseminate is truthful rather than lies and hate, since their role is gatekeeping what reaches mass audiences.
“It is the job of the social media companies the same way the job of newspapers and TV stations before they disseminate something to make sure that it is the truth and it is not some hatefilled lies.”
Recent global developments are very worrying due to accelerating erosion of trust within and between societies, with the international order established after World War II—whose most important taboo was that strong countries cannot invade and conquer weaker countries—now collapsing, reversing thousands of years of previous history when strong-weak conquest was the norm
“Um globally uh how do you how so far it's been very worrying again because of the accelerating erosion of of trust within societies and between societies. Um we see that the international order is collapsing and the international order established after the second world war its most important taboo its most important rule was that strong countries cannot just invade and conqu weaker countries.”
If AI can do everything humans do (invent, teach, decide, create) then the question becomes what role humans could still play in research and society in 10-20 years—a frontier question with no clear answer.
“and it will the question then becomes so what would be the role of humans in in such a situation and again we don't know I mean given the the the the potent the pace of development and the potential it's not clear what would be still what could humans contribute ute to research uh in 10 or 20 years.”
Truth about pandemics is complicated (viruses themselves are barely understood, straddling life/non-life boundary) while conspiracy explanations are simple (powerful people created it), so people believe conspiracy theories despite complexity being the accurate model.
“When there is something like a big pandemic that changes so many things. The truth of the pandemic is very complicated simply to understand what a virus is. A virus is such a strange thing. uh humans for most of history never knew that viruses existed. Only in the last century we really began to understand what viruses are. And the strange thing about them is that they are not exactly alive and not exactly not alive...So to understand how a pandemic begins and spreads very complicated. It's much easier to believe a conspiracy theory. Oh, there is a group of very powerful people. They created this pandemic to take over the world.”
The scientific revolution was fundamentally a revolution of admitting ignorance, not a revolution of knowledge; scientists began by saying 'we don't know' and calling for research, which is a more powerful epistemic stance than claiming to already know.
“For me, the scientific revolution looking back in history was a revolution of ignorance, not of knowledge. Before science, every human group, every human culture said, 'We already know everything. Everything is in the holy books or our holy men and sometimes holy women know everything. No need of new research. The scientific revolution was about people coming and saying we don't know. That's a very we want to know this the answer to this question and we don't know what it is. Oh so we need to do research”
Elon Musk and Donald Trump are firing tens of thousands of human bureaucrats while replacing them with digital bureaucrats—algorithms—so they are not actually against bureaucracy but against human bureaucracy and in favor of algorithmic bureaucracy, which represents a shift of power from humans to algorithms.
“you see at the same time that Elon Musk and Donald Trump they are firing tens of thousands of human bureaucrats but they are not really against bureaucracy because they simply replace them with digital bureaucrats with algorithms. Um and this is the shifting power from humans to algorithms.”
Trump's worldview is that the proper order of the world is for the weak to obey the strong, and if the weak refuse to obey, they are to blame for any resulting conflict; this logic applied globally will bring all humankind back to a constant state of war.
“what he has in mind is a world order in which the proper thing is for the weak to obey the strong. Right? And if the weak refuse to obey the strong, the weak are to blame for the resulting conflict. Right? And you see it like all it threatens that the United States now wants to conquer Greenland. Now if Denmark and Greenland say but we don't want to be part of the US and as a result there is a war in Trump's mind who is to blame for that Denmark is to blame because United States is strong you must obey the strong if you don't obey the strong and there is a war the weak are to blame they should have obeyed the strong”
The purpose of educational institutions like universities should be to create a serving elite—people with talent and influence in particular fields who use their power and talents not just for personal benefit but for the benefit of all society, rather than attempting the impossible fantasy of creating completely equal societies without any elite
“I think especially for an institution like this and for all universities the purpose is not to destroy elites or break down all barriers between this group. The purpose is to create a serving elite. Yes. that people who they have more talents in a particular field, they have more influence in that particular field and they use their talents and their power not just for their personal benefit but for the benefit of all that should be the aim.”
Building trust between humans requires interdisciplinary input from biology and psychology to understand our animal inheritance and how species like chimpanzees build or fail to build trust, from economics and computer science because social systems are now based on information technology, and from other disciplines to understand how human communication mediated by computers differs from previous trust-building mechanisms
“I think we need input and this is why a an interdisciplinary center is the ideal place to do it because no single discipline can can tackle this question by itself. You need input from biology and psychology uh because we are still animals. You need to understand our our animal inheritance. uh and you can learn a lot from how chimpanzees for instance build trust or or fail to build trust between themselves to understand how humans function. Um you need the input you know from economics of course from computer science because today all social systems are based on the new information technology.”
Despite many reasons for anxiety and fear about real problems in the world, ultimately the basis of life is trust, and this provides grounds for hope.
“So yeah, I know there is a lot of reason for anxiety, for fear. It's real. It's not imaginary. There are huge problems in the world, but ultimately the basis of life is trust. Thank you.”
Horses never understood money—they could see samurai trading shiny metal for them but didn't grasp its meaning, which is why humans control horses and not vice versa; similarly, if financial systems become incomprehensible to humans in 10 years, humans will occupy the same epistemic position as horses—unable to understand the systems managing their own lives.
“Like think about it from you know think about horses and money. People have been buying and selling horses for money for thousands of years. But the horses never understood what was happening. The host could see that say one samurai is give is is giving some piece of shiny metal to a merchant and gets me the horse in return. But what was happening there? What is this worthless piece of metal? The horses never understood money. Only humans understood them. Which is why we control the horses and not vice versa. In 10 years, our life could be managed by financial transactions between algorithms that we can't understand that whether we get a loan to buy a house uh will be decided by algorithms. Talking with algorithms in a financial language which is in unintelligible to human beings.”
The impact of Harari's books on people in cyberspace is very difficult to measure objectively—he releases a book to the world but cannot directly know its actual impact on individuals—so he and his husband established a social impact company to translate the book's ideas into more media formats (social media, videos, etc.) to reach more people, trusting that if enough people do what they can in their small field of activity, humanity will be okay
“how do you feel that your effort is penetrating into cyerspace those who live in cyerspace it's very difficult to know I mean you um you write a book you release it to the world what impact it will actually have on people very difficult to know we have I mean one of the things that helped the book is that it's not only me uh my husband and I established a social impact company at the same time that 10 [31:36] years ago that Homodos was was published and I'm in charge of writing the books and he is in charge of all the business side and we have a team of uh uh 20 people working for us one of them is here in Tokyo and uh with social media accounts and with and trying to take the the ideas of of of the book and translate them into more and more media so they could reach more and more [32:07] people and um whether this will be enough we don't know I mean we feel that each person has some little agency in the world uh we try to do what we can and trust other people to do what they can. Nobody can carry the entire world just on their shoulders. If enough people uh do what they can in their small field of activity, I think we'll be okay.”
All hope of reaching a global agreement on how to manage AI development risks has disappeared, especially since the last US elections, making regulation through international cooperation unrealistic.
“The other thing is that all hopes of uh regulating this of reaching some kind of agreement about this seem now extremely naive. Certainly since the last US elections, um I think really all hope of reaching a global agreement on how to manage the risk of AI development is is is gone.”
Harari's books are written with the explicit intention to build a bridge between academic and scientific research and the general public, translating up-to-date theories, models, and findings into language accessible to high school students without requiring computer science background or complex mathematics.
“Absolutely. I mean I see my job as building a bridge between the uh uh uh world of academic studies where mostly experts talk with other experts in a language that is uh very difficult for lay people without academic background to access. So I try to take the most upto-date theories and models and findings of the academic and scientific community and uh relay it explain it in a way that will be easily accessible say even to high school students or to any person who wants to understand what is this AI revolution I don't I I never studied computers I don't want to have all these graphs and equations can I understand in simple language what is the AI revolution and what is the history of information”
Governments can regulate financial markets to require that all tradable financial devices be vetted by humans and understandable to humans, preventing the emergence of postthuman financial systems that only algorithms understand.
“So this is a kind of intersubjective computer reality which we are likely to see very soon especially if we deregulate financial markets again the same way that a government can have a low no fake humans AI can't pretend to be humans governments can also regulate financial markets no all financial devices that are tradable in the market must be vetted by humans we don't allow But with the current trend, we are likely to see a postthuman financial system very soon.”
Harari's 'Homo Deus' published 10 years ago discussed AI as something that seemed like philosophical speculation with little immediate practical implication, but now the pace of AI development is astounding and people in leading corporations predict AGI within 1-5 years rather than by 2029 as Kurzweil predicted.
“10 years ago when I wrote Homodos, all this talk I mean very few people were talking about AI at the time. You had some people in of course in the computer science departments. You had people in the leading corporations uh like Google and uh Microsoft who already understood what was happening and you had a few philosophers like Nick Bostonramm who published super intelligence in 2014. Um but when I wrote Homodos it felt like oh we are talking about something that maybe will happen in hundreds of years some philosophical speculations with very little immediate practical implications for for our lives. Uh now it's all around us and the the pace of development is simply astounding.”
There is a growing division between people who live in cyberspace versus physical space, with more people now living primarily in cyberspace and following short-form content (tweets under 280 characters), and educational institutions face increasing difficulty getting younger generations to read long, thought-provoking books that require sustained attention
“you also in Nexus you also uh talk about um division between those who live in cyerspace versus physical space and uh more people are now living in cyerspace and just you know following let's say tweets uh under 280 characters”
Reading the book 'Nexus' brought a variety of emotions but also gave hope that it's not too late and that human nature contains an ability to self-correct and cooperate together, provided there is enough time before AI reaches a transformative capability level
“reading Nexus brings at least brought me you know wide variety of emotions but but at the same time you know it gave me hope that it's not too late and there's a you know human nature that is you know that can act ability to self-correct and also ability to co cooperate work together um if only we had time I I suppose right before AI gets somewhere else.”