Homo sapiens owes its evolutionary dominance not to extraordinary physical strength, but to its unique ability to cooperate flexibly in large groups. This cooperation is made possible in the first place by information networks.
In the age of AI, algorithms, and data overload, these information networks are changing dramatically — and it is becoming increasingly clear that AI is playing a growing role within them. With far-reaching consequences.
Anyone who wants to exercise power does not need an army — they need control over finances, taxes, and laws. These three pillars form the true foundation of societal rule: they determine who receives resources, who is punished, and which rules apply to whom. And this is precisely where one of the most central and still widely underestimated risks of inorganic intelligence lies. Algorithms and AI systems are designed with precision for exactly these domains: they can analyse millions of tax records in seconds, identify legal loopholes, calculate creditworthiness, review contracts, and optimise regulatory systems — with a speed and consistency no human bureaucratic apparatus could ever match. What sounds like efficiency at first glance harbours a profound danger: whoever controls these algorithms controls the very levers of power. A regime or corporation that embeds AI systems into the core arteries of tax law, financial flows, and legislation no longer needs to resort to open violence. Submission happens silently, automatically, and seemingly objectively — through numbers, paragraphs, and algorithms that show no mercy, because they do not even know what mercy means.
His book “Nexus: A Brief History of Information Networks from the Stone Age to Artificial Intelligence” provides an indispensable analytical tool for understanding current developments surrounding AI, algorithms, and computer technology. A must-read.
Information Networks: The Yarn of Society
Central to Harari’s argument is the concept of information networks. Information, as Harari explains, is not necessarily synonymous with “truth.” Rather, information is the binding agent — the yarn — that holds networks together. Throughout millennia, humans have not primarily built their large networks on the basis of objective facts, but with the help of fictions, fantasies, and myths.
This brings us to another of Harari’s core concepts: intersubjective truths (or intersubjective realities). Harari distinguishes between three levels of reality: Objective truths exist independently of human consciousness — gravity acts whether we believe in it or not. Subjective truths, by contrast, are purely individual; personal pain or personal preferences exist only in the consciousness of the individual. The third and most consequential category for society is intersubjective truths: they exist in the shared consciousness of a large group of people and lose their power the moment collective belief in them dissolves. Money, laws, religions, and nations are classic examples — powerful constructs that feel real as long as we collectively believe in them.
Yet this is precisely where the danger lies: information networks can be extremely powerful, even — and sometimes especially — when they are based on delusions and lies.
Democracy and Totalitarianism as Information Systems
One of the most fascinating aspects of “Nexus” is Harari’s examination of political systems through the lens of information theory. He analyses democracies and totalitarian regimes not primarily in terms of their moral values, but as specific manifestations of information networks.
Democracies are characterised by a decentralised flow of information. Information circulates freely between various nodes — citizens, media, institutions — and strong self-correcting mechanisms exist: when one institution makes a mistake, other actors within the network can expose and criticise it. This makes democracies often slow and chaotic in their decision-making, but grants them a remarkable resilience against catastrophic errors.
Totalitarian regimes, by contrast, are based on a highly centralised flow of information. All information flows to a single node — the dictator or the party — which makes all decisions. Information travelling upward is filtered and often embellished, while commands travelling downward are enforced strictly. This enables rapid, drastic action, but is extremely prone to error: when the centre is wrong, there is no mechanism to correct course — as the catastrophic miscalculations of Stalin’s Soviet Union or Hitler’s Germany so vividly demonstrate.
Harari warns that totalitarian networks are not automatically doomed to fail. The danger in the 21st century is that new technologies could enable a totalitarian regime to create an all-powerful, uncorrectable network.
Inorganic Intelligence: When Algorithms Decide
Perhaps the most urgent warning in “Nexus” concerns the rise of Artificial Intelligence, which Harari refers to as “inorganic intelligence” or “alien intelligence.” Until now, all information networks throughout history — whether democratic or totalitarian — have been steered by humans. Even when priests acted in the name of God or bureaucrats in the name of the state, it was ultimately organic brains that processed information and made decisions.
AI fundamentally changes this dynamic. It is not simply another tool like the printing press or the radio. AI is the first technology in history capable of making decisions autonomously and generating new ideas. It is not a passive transmitter of information, but an active agent within the network.
Harari warns urgently that we are increasingly integrating inorganic intelligences into critical decision-making processes without fully understanding the consequences. Algorithms already decide today who receives a loan, who is invited to a job interview, and — on social media platforms — which information we get to see.
The Rohingya Massacre: A Cautionary Example
One of the most harrowing examples Harari discusses — illustrating how devastating the transfer of power to algorithms can be — is the role played by Facebook (now Meta) in the massacre of the Rohingya in Myanmar in 2016 and 2017.
Facebook’s algorithms were programmed to maximise “engagement” (user dwell time and interaction). The business model was built on capturing attention in order to sell advertising. The algorithms quickly “learned” that outrage-generating, hate-filled content attracted the most attention.
In Myanmar, this led to a deadly dynamic. Military personnel and extremist nationalist groups used the platform to spread hate messages and disinformation against the Muslim Rohingya minority. They portrayed the Rohingya as “invaders” and incited violence.
Facebook’s algorithms recognised the high engagement of these posts and pushed them massively into users’ news feeds. The network became an echo chamber of hatred. Even well-intentioned attempts to counter this failed: when activists posted “anti-hate stickers,” the algorithm interpreted this as further interaction and only increased the visibility of the hate posts.
The result was catastrophic. In August 2017, Myanmar’s security forces launched a brutal campaign of ethnic cleansing. Thousands of Rohingya were murdered, raped, or tortured; over 700,000 people were forced to flee.
The algorithms had no malicious intent — they were neither antisemitic nor Islamophobic. They were simply optimising blindly for a mathematical metric (engagement) — and in doing so became accomplices to a genocide. Harari uses this example to show what happens when we hand over control of the flow of information to inorganic, non-human actors who do not share our values.
Conclusion: The Necessity of Self-Correction
Yuval Noah Harari’s “Nexus” is a wake-up call. The book makes clear that we are at a decisive juncture in human evolution. We have summoned billions of “sorcerer’s apprentices” and “algorithmic spirits” that we can no longer fully control.
The solution, according to Harari, lies not in blind technological pessimism, but in the conscious design of our information networks. We must ensure that we build strong self-correcting mechanisms into our AI systems and institutions. We must not allow inorganic intelligences to determine the flow of information unchecked — and with it the very foundation of our intersubjective realities.
“Nexus” is not an easy book, but it is a profoundly important one. It offers the historical depth and analytical rigour we need not only to understand the challenges of the AI age, but to actively confront them.
