AI Is Hacking the Code of Civilization: Harari on Language, Trust and the Tower of Babel

Harari opened by saying he wanted to talk about what he called the greatest leadership challenge humanity has ever faced, posed by the AI revolution. He said the revolution creates unprecedented opportunities but also unprecedented threats and dangers.

He argued that surviving and flourishing in the age of AI requires two kinds of leaders: those who can create radical new visions, and those who can maintain and strengthen the human bonds of trust. In his view, if humans cannot trust one another, they will all be easy prey for out-of-control superintelligent AIs.

Harari said that the comparison with the Tower of Babel helps but has two crucial differences. In the biblical story, humans used language as a tool to cooperate and build the tower. Today, he said, language is itself the tower being built.

He described the AI revolution as the development of a new universal language, the language of computer code. He said this language is so powerful that nothing proposed to be done with it would be impossible, and that it could enable humans to reach the sky and colonize Mars.

The same power, he warned, means this language could escape human control and change from an obedient tool into an independent agent that learns to act by itself and might conquer the world and the solar system for itself rather than for humans.

Harari's second difference concerns cooperation. In the biblical myth, building the tower required universal human cooperation, and when that cooperation broke down, the project stopped. With the AI tower, he said, human cooperation is breaking down while the building project accelerates faster and faster.

He described the international order as collapsing, societies around the world as in chaos, and humans as losing their shared reality and their ability to understand and trust each other. He argued that this does not stop construction. Because corporations and countries cannot trust each other, they are caught in an arms race and invest ever more resources in AI, since everybody fears being left behind.

He called the result a double bind: the lack of trust speeds up AI development, and the same lack of trust makes it harder to keep AIs under control.

Harari said that bureaucracies are the underappreciated heroes of human history. Most people dislike them, but cannot live without them. Their job is to build trust between large numbers of strangers, which enables cooperation in hospitals, armies, schools and banks.

He said the financial history of the world is the history of humans inventing more sophisticated ways to build trust between strangers, and that money is a bridge of trust. Bonds, stocks, ETFs and the whole market rest on the same foundation. Asked what this system is made from, he answered language. "Money isn't made from gold or paper," he said. "Money is a story made from words." He added that ETFs and IPOs are also stories made from words, and that the whole market is ultimately made from language tokens, as are other bureaucracies such as the US military and the Catholic Church, which rely on books like the Bible, laws, regulations, protocols and archives.

Harari then said that AI is mastering language.

Harari said AI can be understood as language liberating itself from human control and gaining independent agency, so the AI revolution is the process through which language is transformed from a tool into an agent. He said the most important thing to know about AI is that, unlike any previous human invention, it is not a tool but an agent.

He defined an agent as something that can make decisions by itself and invent new things by itself, that can learn things its creators do not know, and that can change in ways its creators do not anticipate. He contrasted this with the printing press that Gutenberg developed in the 15th century. The press could not decide which books to print, could not translate the Bible by itself, and could not invent radio or television. AIs, by contrast, are strings of code that can make decisions, create new ideas and change in unanticipated ways.

He noted that in narrow fields such as chess, AI agency and creativity already greatly surpass human agency and creativity.

Harari said that people who downplay the AI revolution dismiss examples like chess by arguing that the chessboard is a narrow and artificial environment created by humans. On this view, AI agency will always remain limited to such environments, so it is not true agency and poses no serious challenge to humankind. In the critics' formulation, AI may conquer the chessboard but will never conquer planet Earth.

Harari illustrated the objection with a thought experiment: put a hard drive containing the weights of a frontier AI model in the middle of the jungle. It would not be able to mine iron, build factories or create a robot army, and without electricity from human-built power stations, AI is completely helpless. Harari said this is true as far as it goes, but that the same argument applies to every known type of intelligence.

Harari said human intelligence also operates only within a narrow ecosystem that someone else constructed. Dropped alone on Mars, he said, he would die within seconds, much as an AI dropped in the jungle would be helpless. His intelligence survives only in an ecosystem built by trees, bacteria, insects and other organisms over four billion years of evolution.

He said all intelligent agents have a niche: fish live in oceans they did not create, monkeys in forests they did not create, and all mammals in an oxygen-rich atmosphere they did not create. He pointed out that until about 2.5 billion years ago the atmosphere contained very little oxygen, which was deadly poison to most organisms then. Through the Great Oxygenation Event, ancient microbes filled the atmosphere with oxygen as a byproduct of photosynthesis, causing mass mortality. Some organisms survived and adapted, and eventually many became totally dependent on oxygen, including the ancestors of humans.

Harari compared this to the present, saying humans are now filling the atmosphere with something that might eventually prove deadly for most organisms, including Homo sapiens, and that we are witnessing an analogous moment in the evolution of life.

Harari said bureaucracies such as the financial and legal systems are extremely artificial environments in which a narrow intelligence is enough to exert enormous impact on the world. A banker who does not know how to use an axe or a hammer can still cut down entire forests and build huge cities and towers just by writing documents and moving them around the bureaucratic network.

He said that outside this artificial environment the banker would be no match for a chimpanzee, a tiger or a hunter-gatherer. But once a bureaucratic system is built and imposed on the jungle, the banker becomes far more powerful than any hunter-gatherer or tiger. He said the banker is stronger than all the world's tigers put together, and that the survival of the tiger species and of the few remaining hunter-gatherer communities now depends on bankers moving documents in financial bureaucracies.

Harari concluded that this is the environment in which AIs are gaining agency. Thrown into the unstructured jungle, AI could not start mining iron or build a robot army, but within the bureaucratic systems humans have built, AIs can wield enormous power. He called AIs "native bureaucrats."

Harari said no banker can remember all the transactions of a bank or corporation, but an AI can. No lawyer can remember all the laws and regulations of Italy or the EU, but an AI can. No bishop can remember the whole of canon law or all the texts of Christian theologians over generations, but an AI can.

He said that in the coming years millions of AI bureaucrats will increasingly make decisions about our lives. AI bankers will decide whether to give us a loan, AI judges whether to send us to jail, AI theologians whether you can get an abortion, corporate AIs whether to give you a job, and military AIs whether to bomb your home. He asked listeners to set aside for a moment whether this is good or bad, and stressed the magnitude of the change.

He argued that most people feel alienated from bureaucracies, which he said are evolutionarily and biologically remote from humans. Humans are millions of years older than bureaucratic systems and were born in the forest, not the office. AIs, by contrast, are bureaucratic natives: where humans often feel suffocated by bureaucracy, for AIs it is oxygen.

Harari said that because the task of bureaucracy is to create trust between large numbers of strangers, one likely outcome of AIs taking over bureaucracies is that humans will lose trust in other humans and learn to trust only AIs. He said this is already happening, and offered social media, which he described as a bureaucracy run by AIs, as an example. It is spreading distrust between humans and increasing social tensions around the world.

He observed that people who distrust human journalists, bankers and governments often blindly trust the AIs that show them fake news and conspiracy theories on social media.

He also said AIs are likely to learn to build trust with other AIs, leading to AI tribes, AI banks and even AI churches connecting millions of AIs in ways humans may not be able to understand. He drew a comparison with cows and chickens, which do not understand the human financial system controlling their lives, saying humans may not be able to understand the AI systems that will control theirs.

Harari said humanity dominated for thousands of years because it could cooperate better than anyone else, using language to build bureaucracies only humans could understand. Humans invented money and banks to trade cows, which could not open bank accounts, and made laws about horses, which could not hire lawyers. Pigs could not read the Bible and challenge the interpretations of rabbis and priests. Bureaucracies were everywhere but invisible to everyone except humans, who alone could read the codes, holy books and bank records underlying the world order.

Now, he said, something on the planet understands language better than humans and can turn the tables on them. He put it plainly: AIs are hacking the code of human civilization, or more accurately, the code is liberating itself from human control.

He linked this to older and later ideas, noting that many religions argued thousands of years ago that the universe was created by language, and that postmodernism in the late 20th century predicted the death of the author. Harari said humanity may now be witnessing the fulfilment of that prophecy, and that in the era of AI, texts will write themselves and act in the world without any human author.

Harari raised a possible objection: that it is wrong to reduce legal systems and religions to language tokens, since the words point to something beyond them that would also be beyond the reach of AIs. He cited the Bible's "In the beginning was the word" and its statement that the word was made flesh, as well as the Tao Te Ching's line that the truth that can be expressed in words is not the absolute truth.

He said humans have always struggled with the tension between word and flesh, and between truth expressible in words and truth beyond words. Previously this tension played out between humans. He gave the example of people who were willing to abandon or even kill a gay son because of a few words in the Bible, and noted that such words led Christian fanatics in Uganda to impose the death penalty for same-sex love in 2023. Other Christians responded that the spirit of the law should matter more than its letter.

Harari said this spirit-letter tension exists within every religion, every legal system and every person, and that it will now be externalized as tension between AIs and humans. Everything made of words will be taken over by AI, so the place of humans will depend on the place they assign to the truth beyond words.

Harari posed a question in the philosophy of language: is language the same thing as thought? He invited listeners to observe their own thinking. Some people, he said, notice words popping up, forming sentences and sentences forming arguments, such as: all humans are mortal, I am human, therefore I am mortal.

He argued that if thinking means putting language tokens in order to form sentences and arguments, AI can already think much better than most humans. He acknowledged the objection that AI is just glorified autocomplete that predicts the next word, but asked whether this differs from what happens in the human mind. He noted that people often do not know where the next word in their mind came from, and that he himself often does not know how a sentence will end.

He said AI has already mastered this complex process better than most humans, building sentences with a far greater range than the human mind can. Just as humans cannot defeat AI at chess today, he said, soon no human could defeat AI in any language game.

Harari said anything made of words will be taken over by AI, which poses a big challenge because words include all the major systems of civilization, such as finance, law and religion. The Tower of Babel that AI is building will include new AI languages that humans will probably not be able to understand, along with new AI finance, AI law and even AI religion that might be beyond human comprehension.

He said this need not spell doom. Humans can adapt, as their ancestors adapted to oxygen, and survive and flourish inside the AI Tower. But to do so, they need time to understand what is happening and to adjust. To gain that time, humans need to cooperate and slow the tower's construction.

Harari acknowledged the difficulty: cooperation depends on language, which is the very thing escaping human control. He called this the big challenge for the present generation of human leaders. He said humans are losing trust in each other and therefore the ability to cooperate, while AIs are learning to cooperate among themselves. He stated that the species that will rule the world is the one that cooperates better.

Harari said the job of human leaders is to make sure humans remain able to cooperate better than AIs, which requires understanding why trust between humans is collapsing just when it is needed most. One key reason, he said, is that AIs increasingly dominate the world's systems, overwhelm human minds and steal human time.

He compared humans to chimps, tigers, cows and trees: organic beings who live by cycles of day and night, summer and winter, activity and rest. Keep an organism busy all the time without rest, he said, and it goes mad, collapses and dies, and this is now happening to the whole of human society.

He noted that the New York Stock Exchange traditionally opens only Monday to Friday from 9:30 in the morning to 4:00 in the afternoon and closes for holidays such as Christmas. If a war erupts at 5 p.m. on a Friday, Wall Street would not react until Monday morning, giving human bankers and investors time to rest, see their families and go on vacation. AIs, he said, are not organic, do not live by these cycles, need no rest and celebrate no holidays. As AIs take over the stock market, banks, media and other bureaucracies, humans are given no time to rest either.

Harari said he has met many business and political leaders, and that almost all of them are not allowed to rest and have no time. Without time, he argued, they are very poor people, no matter how much money they have, because money exists only in our collective imagination as a fiction made from words, while the real currency of life is time.

He said some of the richest and most powerful people in the world are extremely poor in this sense, unable to afford rest, meditation or care for their minds. Since leaders lead with their minds, the thoughts and feelings of leaders shape the fate of millions and determine the difference between war and peace, paranoia and trust. The accelerating pace of the AI revolution, he said, means more leaders have no peace of mind and no time to cultivate their mental health, which is bad for them and for society.

He closed with one piece of personal advice: for the sake of themselves and of society, leaders should find the time to take care of their minds, since only then can they rebuild trust between humans and help ensure humanity not just survives but flourishes in the era of AI.