Nexus Explained: The Terrifying History Of Information Networks From Stone Tablets To Artificial Intelligence
How Human Civilisation Became A Network — And Why AI Could Break It
Information, Power, Democracy, Dictatorship, And The AI Crisis
Human beings like to believe information makes the world wiser.
Yuval Noah Harari’s Nexus is built around a colder idea: information does not automatically produce truth, wisdom, freedom, or peace. Information connects people, but it can also mislead them, organise them, control them, divide them, and destroy them.
That is the core warning of Nexus: A Brief History Of Information Networks From The Stone Age To AI. Harari’s official description frames the book as a long historical account of how information networks have “made and unmade” the world, written specifically to help readers understand the threats and promises of the AI revolution.
Nexus was published in 2024, after Sapiens, Homo Deus, and 21 Lessons For The 21st Century. It is best understood as the missing bridge between Harari’s earlier arguments: Sapiens explained how shared stories let humans cooperate; Homo Deus warned that data and algorithms could displace human authority; Nexus explains the information networks that made both outcomes possible.
The Big Idea Of The Book
The big idea of Nexus is that information is not the same thing as truth.
This sounds obvious, but much of modern culture forgets it. We assume that more information means better decisions. More data means more accuracy. More connectivity means more understanding. More speech means more freedom.
Harari says that is dangerously naïve.
Information’s primary function is not to represent reality accurately. Its primary function is to connect things. Sometimes it connects people through truth. Sometimes it connects people through lies, myths, fear, bureaucracy, identity, propaganda, or fantasy.
A tax record can organise a state. A myth can organise a tribe. A holy book can organise a civilisation. A newspaper can organise a democracy. A radio network can organise genocide. A social media feed can organise rage. An AI system can organise reality before humans even understand what has happened.
The central question of Nexus is therefore brutal:
Can humanity build information networks powerful enough to run the world without letting those same networks dominate, deceive, or replace us?
The Argument In One Flow
Nexus begins from a direct challenge to the modern faith in information.
Many people assume that history is a slow march from ignorance to knowledge. First humans had myths. Then they invented writing. Then they built libraries. Then printing spread knowledge. Then science improved truth. Then the internet connected everyone. Then AI arrived to process everything.
That story sounds optimistic.
Harari complicates it.
He argues that information has always had two faces. It can reveal reality, but it can also create order. These are not the same thing. A story can be false and still organise millions of people. A bureaucracy can be inaccurate and still control lives. A propaganda system can be delusional and still mobilise armies.
The earliest human societies depended on stories, rituals, symbols, kinship, memory, and trust. Before writing, information lived in human minds and social practices. People knew who owed what, who belonged where, which gods mattered, what the tribe remembered, what the ancestors demanded, and which rules preserved the group.
Writing changed the scale of power.
With writing, information could leave the human brain and enter durable external systems: clay tablets, scrolls, ledgers, archives, contracts, tax records, legal codes, and religious texts. This allowed states and empires to organise far more people than oral memory alone could manage.
But writing did not automatically make society wiser.
It made society more administratively powerful.
This distinction matters. A written tax record may be accurate or inaccurate. A legal document may be just or unjust. A holy text may inspire compassion or violence. The key point is that written information can outlive individuals, travel across distance, and become part of institutional machinery.
Bureaucracy becomes one of Harari’s major examples. Modern people often mock bureaucracy as boring, slow, and soulless. Harari treats it as one of history’s most powerful information technologies.
A bureaucracy turns messy human reality into categories, forms, numbers, files, certificates, permissions, and classifications. It decides who owns land, who pays tax, who is married, who is a citizen, who has status, who is guilty, who receives benefits, and who disappears into a system.
That gives bureaucracy enormous power.
It can make societies legible. It can also make people smaller.
The next major step is mythology and religion. Harari does not treat religion merely as private belief. He treats it as a large-scale information network capable of organising behaviour across time and space.
A religion can bind strangers together through shared stories, sacred texts, rituals, moral rules, institutions, calendars, authorities, and punishments. It can preserve wisdom and create solidarity. It can also preserve dogma, justify violence, and block correction.
This brings Harari to one of the book’s deepest themes: self-correction.
An information network is dangerous when it has no reliable mechanism for correcting itself. If a system treats its own outputs as sacred, unquestionable, or perfect, then mistakes become structural. A falsehood can become permanent. A bad rule can become divine. A corrupt leader can become untouchable.
Democracy matters to Harari because it is not merely a voting system. It is an information system with self-correcting mechanisms. A functioning democracy spreads information across many institutions: elections, courts, free media, opposition parties, universities, civil society, public debate, local government, and independent regulators.
No single centre knows everything.
That is the strength.
A democracy can be noisy, slow, contradictory, emotional, and frustrating. But those weaknesses are also connected to its resilience. Because information is distributed, errors can be exposed. Leaders can be criticised. Governments can be replaced. Journalists can investigate. Courts can challenge executive power.
Totalitarian systems work differently.
They centralise information. They try to force society into one channel of truth: one party, one leader, one ideology, one official reality. This can make them look strong. They can move fast, silence opposition, control narratives, and mobilise resources.
But centralisation creates blindness.
If nobody can safely tell the truth upward, the centre becomes stupid. If every official reports success because failure is dangerous, the leadership loses contact with reality. If every institution repeats the same lie, the system may look unified while becoming informationally rotten.
This is one of Nexus’s strongest warnings. Dictatorships do not merely suffer from moral evil. They suffer from bad information architecture.
They punish correction.
Modern media intensifies the problem.
Printing presses, newspapers, radio, television, and the internet all transform information networks. Each technology creates new forms of power and new forms of distortion. Print can spread scientific knowledge, but also conspiracy and religious war. Radio can educate a nation, but also coordinate hatred. Television can expose injustice, but also turn politics into theatre.
The internet appeared to many people as the ultimate democratising force. It lowered barriers to publishing, connected billions, enabled new communities, and gave ordinary people access to vast information.
But Harari’s broader argument suggests that more connection does not automatically mean more truth.
The internet also created an attention economy, misinformation networks, outrage machines, algorithmic amplification, surveillance systems, and social fragmentation. It connected people at scale before societies had built strong enough correction mechanisms.
Then comes artificial intelligence.
For Harari, AI is not just another information technology. It is categorically different because it can make decisions and generate new ideas by itself. Earlier media transmitted human-created information. AI can produce information, interpret information, and act on information at speed and scale.
A printing press does not decide what to print.
A radio does not invent propaganda.
A television does not autonomously tailor a message to each viewer.
AI can.
That is the rupture.
AI can write texts, generate images, imitate people, design strategies, analyse behaviour, recommend actions, trade assets, identify targets, rank citizens, manipulate emotions, and create synthetic realities. It can become an active agent inside the information network rather than a passive channel.
This is why Harari worries about AI’s political and social consequences. The Guardian’s review of Nexus notes his concern that AI can make decisions independently and that the book repeatedly returns to the dangers of digital surveillance, fake news, and the erosion of democratic trust.
The danger is not only that AI may become “smarter” than humans in some abstract sense. The danger is that AI may flood society with persuasive synthetic information, destroy shared trust, and make it impossible for citizens to know what is real, who is speaking, or which institutions can still be believed.
A democracy depends on conversation.
AI can poison conversation.
A market depends on trust.
AI can fake trust.
A legal system depends on evidence.
AI can manufacture evidence.
A human relationship depends on confidence that another mind is present.
AI can simulate presence.
This is why Nexus is darker than a normal technology book. It is not just about automation or productivity. It is about whether human societies can preserve reality-testing when machines can manufacture personalised realities at industrial scale.
Harari also worries about surveillance. AI makes it possible to process immense amounts of data about human behaviour, speech, movement, emotion, health, and social networks. Totalitarian states in the twentieth century wanted total surveillance, but lacked the technological capacity to watch everyone all the time. AI makes that old dream more feasible.
A secret police officer cannot read every message, watch every citizen, and analyse every facial expression.
An AI system can move much closer to that.
This changes the relationship between citizen and state. It also changes the relationship between consumer and corporation. A platform does not need to imprison you to shape your life. It can rank what you see, predict what you want, test what angers you, and sell access to your attention.
Nexus therefore links ancient bureaucracy to modern AI.
Both systems classify people.
Both convert human reality into information.
Both can serve society.
Both can erase the person inside the file.
The book’s historical sweep matters because Harari is not saying “technology bad.” He is saying that every information network needs correction, accountability, and humility. Without those, the network becomes a machine for producing power without wisdom.
The climax of the argument is the problem of alignment between intelligence and truth.
AI systems may optimise for engagement, profit, power, efficiency, persuasion, or institutional goals. None of those are identical to truth. A system can be brilliant at achieving a target and disastrous for civilisation if the target is wrong.
That is why “more intelligence” is not enough.
A lie can be intelligent.
A dictatorship can be efficient.
A propaganda system can be coherent.
A recommendation engine can be accurate about what keeps you watching while destructive to your wellbeing.
The book’s final movement is a warning that humanity must build self-correcting institutions before AI becomes too deeply embedded in the nervous system of civilisation. This aligns with broader AI-risk literature: a 2023 paper co-authored by Harari and leading AI researchers warned that advanced autonomous AI systems could produce large-scale social harms, malicious uses, and even irreversible loss of human control if governance and safety preparation remain inadequate.
Nexus does not end by saying AI must be stopped entirely.
It says AI must be governed as a civilisational force, not treated as a normal product launch.
The book’s deepest warning is that humanity’s problem has never been lack of information. It has always been the difficulty of turning information into truth, truth into wisdom, and wisdom into power that can correct itself.
The Main Concepts Inside The Argument
The first key concept is the difference between information and truth.
Information connects. Truth represents reality. Sometimes they overlap. Often they do not. A rumour, myth, database, law, scripture, conspiracy theory, or propaganda broadcast can connect millions while being partly false or morally disastrous.
The second concept is the information network.
A network is not just a collection of facts. It is a structure that moves information between people, institutions, machines, and authorities. Families, religions, states, markets, universities, newspapers, platforms, and AI systems are all information networks.
The third concept is self-correction.
This is the heart of Harari’s political argument. Good information networks need mechanisms that detect error, punish corruption, allow criticism, and update beliefs. Science, democracy, independent courts, free media, peer review, and institutional checks all matter because they help systems correct themselves.
The fourth concept is centralisation versus distribution.
Centralised systems can act quickly but often suppress correction. Distributed systems can be chaotic but often detect mistakes better. Democracy is messy because it distributes information. Dictatorship is brittle because it concentrates it.
The fifth concept is AI agency.
AI is different from earlier media because it does not merely carry human messages. It can produce messages, make decisions, pursue goals, and shape environments. This makes it an active participant in history.
The Central Conflict Inside The Book
The central conflict in Nexus is between connection and correction.
Humanity needs information networks because large societies cannot exist without them. But every network that connects people also creates opportunities for manipulation, distortion, control, and delusion.
Too little connection means chaos.
Too much uncorrected connection means mass error.
Democracy, science, journalism, and law are attempts to solve this problem. They are not perfect truth machines. They are correction machines. Their value lies not in never being wrong, but in allowing wrongness to be challenged.
AI threatens this balance because it can accelerate information production faster than correction systems can respond.
If lies can be generated at scale, if fake humans can flood public debate, if synthetic evidence can overwhelm courts, if personalised propaganda can target each citizen privately, then democratic self-correction weakens.
The conflict is not humans versus machines in a simple science-fiction sense.
It is accountable truth-seeking versus automated persuasion.
The Turning Points Inside The Argument
The first turning point is writing.
Writing allows information to scale beyond memory. It creates states, archives, contracts, scriptures, tax systems, and bureaucracies. It lets humans organise complexity but also turns people into records.
The second turning point is bureaucracy.
Bureaucracy makes society governable by converting reality into categories. It can deliver order, but it can also flatten human complexity into boxes, numbers, and files.
The third turning point is mass media.
Print, radio, and television make it possible to coordinate entire populations through shared narratives. This can support education and democracy, but also propaganda and mass hatred.
The fourth turning point is the internet.
The internet connects billions and decentralises publishing, but it also overwhelms correction. It creates too much information for old trust systems to process.
The fifth turning point is AI.
AI turns the information network into something partly autonomous. It can generate, filter, rank, imitate, persuade, decide, and adapt. That makes it more powerful than any previous information technology.
Nexus ends as a warning rather than a prophecy.
Harari does not argue that catastrophe is inevitable. He argues that catastrophe becomes more likely if societies continue treating information technology as neutral infrastructure or private entertainment.
The solution is not blind anti-technology. It is better information architecture.
Democracies need stronger correction mechanisms. AI systems need accountability. Platforms need responsibility. Institutions need transparency. Citizens need media literacy. Governments need coordination. Societies need to defend trust as seriously as they defend borders.
The ending also reframes the AI debate.
The question is not only whether AI becomes conscious.
The question is whether non-conscious intelligence can dominate the information networks that govern conscious beings.
That is more immediate and more disturbing.
The Story Anchor
The strongest story anchor in Nexus is bureaucracy.
It sounds dull until you see what it really does.
A human life enters a system and becomes a file. The file becomes a category. The category determines rights, duties, taxes, permissions, punishments, and opportunities. The person remains alive, complex, emotional, contradictory, and specific. But the institution sees the record.
That is the ancient version of the AI problem.
Modern algorithms do something similar at far greater speed. They classify people, predict behaviour, rank risk, recommend content, approve loans, flag threats, target adverts, and shape attention.
The file became the profile.
The profile became the prediction.
The prediction may become the decision.
If You Only Remember Three Ideas
The first idea is that information is not truth.
More information can create more confusion if the network rewards speed, emotion, loyalty, or power over correction. A society drowning in content can still lose contact with reality.
The second idea is that democracies are information systems.
Democracy is not just voting. It is a distributed method for processing reality. Courts, journalism, opposition, universities, public debate, and civil society all help prevent one centre from controlling truth.
The third idea is that AI is the first information technology that can become an agent.
Earlier tools stored or transmitted human information. AI can create information, make decisions, and interact with people as if it were a social actor. That makes it politically explosive.
The Sentence That Explains The Entire Book
Nexus argues that civilisation was built by information networks, but may be broken by networks that become too powerful to correct.
Why This Book Matters
Nexus matters because the AI debate is often too narrow.
People ask whether AI will take jobs, write essays, generate images, or become conscious. Harari asks a bigger question: what happens when AI enters the information bloodstream of civilisation?
That question is already practical. Elections, journalism, education, warfare, finance, policing, healthcare, dating, entertainment, and public trust are all being reshaped by digital information systems.
The book also matters because it refuses the lazy assumption that free speech alone solves the truth problem. Speech matters, but speech without trust, verification, accountability, and correction can become noise or manipulation.
Nexus is not perfect, but it gives readers a powerful frame: do not judge an information network by how much content it produces. Judge it by how well it corrects itself.
What Most People Misunderstand About This Book
The shallow reading is that Nexus is “Harari’s AI book.”
That is only partly true.
It is really his information book. AI matters because it is the newest and most dangerous stage in a much older story. The same problem runs from ancient myths to modern algorithms: who controls the network, how does it define reality, and can it correct itself when wrong?
The second misunderstanding is that Harari is anti-information.
He is not.
He is anti-naivety. He is warning that information without correction can become a weapon.
Misconceptions
Harari is not merely saying that AI might become powerful. He is saying that human societies have always depended on fragile information systems, and AI is arriving at a moment when many of those systems are already weakened.
The online version says: “AI might destroy us.”
The deeper version says: “AI may enter broken trust networks and make repair much harder.”
That is a more serious argument.
The Taylor Tailored Interpretation
The Taylor Tailored interpretation is this: Nexus is about the difference between being informed and being governed.
Modern people think they consume information. Harari shows that information consumes, classifies, directs, and organises them back.
A feed is not just a feed.
A file is not just a file.
A database is not just a database.
An AI assistant is not just an assistant.
Every information system trains behaviour. It decides what becomes visible, what becomes credible, what becomes normal, and what becomes impossible to ignore.
The real power struggle of the twenty-first century is not simply who owns land, oil, factories, or weapons. It is who owns the systems that decide what reality feels like.
The Real-Life Test
The real-life test is to ask which information networks currently shape your decisions.
Your workplace dashboard shapes what counts as performance.
Your bank account shapes what feels possible.
Your social feed shapes what feels urgent.
Your news sources shape what feels true.
Your private messages shape what feels emotionally real.
Your search results shape what you think exists.
Your AI tools may soon shape what you believe you decided yourself.
Nexus asks you to stop treating these systems as background. They are not background. They are the operating environment.
How To Apply The Lessons Without Turning Them Into A Fantasy
Do not respond to Nexus by declaring that everything is propaganda.
That is lazy.
Instead, build better correction habits.
Check source quality before emotional reaction. Separate primary evidence from commentary. Notice when a platform rewards outrage. Keep some information channels slow. Do not outsource judgement completely to feeds or AI summaries. Preserve relationships and institutions that can tell you when you are wrong.
At work, design systems that surface errors rather than hide them.
In public life, defend institutions that can correct power.
In private life, ask whether your strongest opinions are truth-seeking or identity-protecting.
The point is not to escape information networks. You cannot.
The point is to live inside them without becoming their easiest product.
Five Questions To Test Whether You Actually Understood This Book
What is the difference between information and truth?
Why does Harari think self-correction matters more than information volume?
Why are democracies better understood as information systems, not just voting systems?
Why is AI different from printing, radio, television, or the internet?
What happens when a society’s information network becomes powerful but loses the ability to correct itself?
The Final Lesson
Nexus is a warning against one of the modern world’s most dangerous beliefs: that more information will save us.
It might.
But only if our networks can still tell the difference between signal and noise, truth and loyalty, evidence and manipulation, intelligence and wisdom.
Human civilisation was not built by truth alone. It was built by stories, files, myths, records, commands, institutions, and machines for organising trust. Now those machines are becoming intelligent.
The final lesson is simple and severe: the future will not be decided by who has the most information, but by who builds the networks that can still correct themselves when power starts lying.
