Plato Was Right: Artificial Intelligence Makes the Philosopher-King More Urgent

By Matthew Parish

Thursday 22 October 2026

Plato has been out of fashion for about two thousand years. That may be a sign that he had something important to say. His most famous political idea was that states would not stop suffering from problems until philosophers became kings or kings truly became philosophers. Modern liberal societies have often treated this idea as an embarrassing leftover from the ruling class of ancient Athens. We prefer elections, opinion polls, focus groups, television debates, political advisers and politicians whose main intellectual skill is knowing which camera to look at. The idea of a philosopher-king seems dangerous, elitist and a little ridiculous. But artificial intelligence may make Plato's idea seem important again.

This is not because philosophers are especially pleasant people. Often, they are not. Nor does it mean that philosophy departments should control nuclear weapons, central banks and intelligence agencies. Trying that might quickly prove every argument against Plato right. The real reason is more worrying. Humanity is building tools with almost unimaginable intellectual power, while many people who govern advanced societies still show very ordinary human limits in their thinking. The gap between the power of our tools and the wisdom of the people using them is becoming dangerous.

Artificial intelligence does more than give political leaders access to more information. It also gives them greater power to use that information, influence populations, predict people's behaviour, organise government offices, wage wars, manage economies and, eventually, make decisions faster than human institutions can properly follow. We are moving towards a world where ordinary or unwise people may control extraordinary intelligence. Plato would have understood the problem immediately.

The Ancient Problem of Stupid Rulers

Plato did not believe that clever people should rule simply because they were clever. Athens had plenty of clever people. Sophists could make convincing arguments for almost anything, politicians could stir up crowds, merchants could work out profits and generals could plan clever military campaigns. Cleverness was common, but wisdom was not. This difference is central to The Republic.

A philosopher is not simply someone who knows a lot. A philosopher is trained to ask what knowledge is for. Philosophers learn to tell appearances from reality, desires from values, and short-term gains from what is truly good. Most importantly, Plato's ideal ruler does not desperately want power. Wanting political power too badly may itself be a sign that a person is not suited to hold it.

Plato would probably have found modern election campaigns confusing. Candidates spend huge amounts of money trying to prove that they want power more than their rivals. They shake thousands of hands, repeat slogans chosen through opinion polls, hire experts to improve their public image and spend months explaining why they alone can save civilisation from disaster. For Plato, this would be a very good way to identify the people who should not govern.

Yet modern constitutional democracy has a powerful answer to Plato's concerns. We do not need rulers to be exceptionally wise because we can limit their power. Elections can remove them, courts can block them, legislatures can challenge them, journalists can expose them and government offices can slow them down. Constitutions set limits on what leaders may do, even when those leaders believe their ideas are excellent. The liberal state is built on a less idealistic view of human nature than Plato's republic. Rather than trying to find a perfectly wise ruler, it asks how we can live safely with an imperfect one.

This system has worked surprisingly well. But artificial intelligence creates a new problem: it may make imperfect rulers much more powerful. The constitutional rules designed to limit ordinary human mistakes may not be enough when those mistakes are backed by enormous computing power.

Give a Fool a Supercomputer

In the past, foolish leaders faced practical limits. A leader might want to reorganise the economy overnight, arrest thousands of supposed enemies, control the information every citizen received or find the exact weaknesses of millions of people. But carrying out such plans required slow, expensive administrative systems. Government workers had to gather information, process paperwork, send orders and enforce decisions. Sometimes, the very incompetence of a government accidentally helped protect people from its power.

Artificial intelligence removes many of these practical obstacles. A government with sufficiently advanced AI could potentially analyse huge populations in real time. It could connect financial records, movements, communications, facial recognition data, political views and social relationships. It could create personalised propaganda for each citizen instead of sending everyone the same message. It could detect political opposition before people had organised themselves, while improving taxation, military mobilisation, policing and economic intervention on a scale that was once unimaginable.

The machine itself does not need to become tyrannical for any of this to happen. Humans can provide the tyranny. The popular nightmare is that AI becomes evil and decides to control humanity. But a more familiar danger from history is that a human being decides to control other people and discovers that AI makes this much easier. A malicious superintelligent machine would certainly be frightening. But a foolish politician with access to superintelligence may be a more immediate danger.

Intelligence Is Becoming Cheap

For most of human history, producing sophisticated intellectual work was expensive. Writing a good legal report required lawyers. Designing a bridge required engineers. Creating propaganda required writers, artists and printers. Studying military intelligence required large teams. Governments built specialised offices full of people with expert knowledge because intelligence was limited, slow and costly to reproduce.

Artificial intelligence is making many kinds of intellectual work much cheaper. It can already produce convincing legal analysis, computer code, economic summaries, strategic plans, speeches, translations and administrative documents in seconds. These abilities will improve. This creates an unusual situation: intelligence, or at least many things that look like intelligence, may become widely available, while wisdom does not.

This difference may shape politics in the age of AI. The world may soon have more analysis than anyone can read, more predictions than anyone can judge and more arguments than anyone can answer. Every government minister may have machines ready to produce highly sophisticated reasons for almost any decision. The challenge will no longer be finding someone who can build an argument. It will be finding someone who can recognise when an impressive argument is nonsense. That requires, in large part, philosophical skill.

Plato's Cave Acquires Wi-Fi

There is another reason Plato suddenly seems so relevant. His best-known story describes prisoners chained inside a cave, watching shadows on a wall. Because they have never seen anything else, they mistake the shadows for reality. Over the past thirty years, modern civilisation has built what may be the most sophisticated cave in history.

We increasingly experience the world through screens. Algorithms choose which news stories we see, which people appear in our feeds, which advertisements follow us and which arguments dominate public discussion. Search engines shape how we find knowledge. Social media shapes how we make friends. Dating apps shape how we find romance. Navigation systems shape how we understand places. Artificial intelligence is increasingly shaping language itself. Generative AI completes Plato's metaphor because the shadows can now create themselves.

A photograph no longer has to show something that really happened. A voice no longer has to belong to the person who seems to be speaking. A convincing essay does not have to be written by someone who believes its argument. Millions of artificial personalities may eventually exist alongside real people in digital spaces, creating political arguments, product recommendations and emotional responses that are impossible to distinguish from those of actual humans. At least Plato's prisoners knew they were watching a wall. We may eventually forget that the wall exists.

The political effects could be enormous. Democracy needs citizens to share at least a basic understanding of reality. People may strongly disagree about what should be done, but political debate becomes impossible if they cannot agree on whether something happened at all. AI threatens to do more than add lies to public discussion. Politicians managed that well enough before computers. It could make it much harder to know where any piece of information came from. Plato's old challenge of telling appearances from reality is therefore no longer just an academic exercise. It is becoming an essential part of civic life.

The Political Mediocrity Problem

There is an uncomfortable fact that democratic societies prefer not to discuss. Political systems do not always choose leaders based on intelligence, good judgement or independent thinking. Success often depends on a mix of ambition, energy, charm, fundraising, loyalty to a political party, media skills, agreement with the party's beliefs and luck. Some outstanding people succeed in this process, as do many decent people. But some very ordinary thinkers also reach the top.

This mattered less when political leaders had fairly basic tools. It matters much more when a head of government may have access to computing systems that can analyse entire economies, simulate wars or monitor whole populations. A ruler's poor thinking can have much greater consequences when technology greatly increases their ability to turn impulses into action.

Traditionally, expert advice helped protect against poor political judgement. Ministers did not need to understand nuclear physics because nuclear physicists could advise them. They did not need to understand disease outbreaks because epidemiologists were available. They did not need to understand how money and economies worked because central banks employed economists. AI changes the relationship between leaders and experts because a machine may eventually know more than every specialist in the room. The politician will then face an adviser whose conclusions they cannot work out independently and whose abilities they may not fully understand.

At that point, character becomes more important than intelligence. Does the politician understand uncertainty? Can they tell the difference between confidence and evidence? Do they know when to ask another question? Can they accept an answer that goes against their instincts? Do they understand that every attempt to find the best solution depends on assumptions about what matters most? Can they tell the difference between what is possible and what is right? These are not mainly technical questions. They are questions about how a person thinks and judges the world.

The Machine Cannot Choose the Good

Imagine that an AI could calculate the most cost-effective healthcare system. That would not tell us whether saving money should be healthcare's most important goal. Imagine that it could find the best surveillance system for preventing terrorism. That would not tell us how much privacy people should give up for safety. Imagine that it could find the military plan most likely to win a war. That would not tell us how many civilian deaths could be morally justified to achieve victory. Machines may become remarkably good at finding ways to achieve goals, but choosing the goals is a different problem.

Every attempt to optimise a political decision contains a hidden moral choice. Maximising GDP puts greater value on production. Maximising lifespan puts greater value on living longer. Maximising equality may require giving up some freedoms. Maximising security may require more surveillance. Maximising military success may require actions that a civilised society finds unacceptable. Science cannot make these choices for us. It can describe what may happen, but it cannot decide what we should value.

Making AI more intelligent will not automatically solve this problem. Greater intelligence may help us predict the results of our choices with amazing accuracy, but it will not settle which results we should prefer. A machine may tell a king how to get whatever he wants. Philosophy asks whether he should want it in the first place. The difference between these questions lies at the heart of the political problem.

Socrates May Be More Useful Than Plato

There is an obvious danger in bringing back the idea of the philosopher-king. Every dictator in history has believed that he was unusually wise. Give a ruler enormous computing power and convince him that he alone understands what is good, and Plato's ideal state could become a kind of digital prison. A philosopher-king with AI could turn into a philosopher-tyrant with terrifying speed and efficiency.

The answer cannot simply be to let professors govern. Universities provide plenty of evidence that advanced education does not remove vanity, group rivalries, pettiness or foolishness. Academic politics backed by artificial superintelligence might produce disasters with exceptionally detailed footnotes. The more useful figure from ancient Athens may therefore be not Plato's ruler but his teacher, Socrates.

Socrates did not walk around Athens claiming to have every answer. Instead, he annoyed people by asking how they knew that their answers were right. Part of his wisdom came from recognising what he did not know. He asked people to explain what they meant, reveal their assumptions and follow their arguments until contradictions appeared. Political leaders will need exactly this kind of thinking when dealing with AI.

When a machine says Policy A is the best choice, the first question should be: best according to what standard? When it predicts a 93 per cent probability, someone must ask how that number was calculated. When it recommends surveillance, someone must ask what idea of freedom was built into its calculations. When it advises war, someone must ask how the model counted the human costs. When everyone in the room is impressed by the machine, someone must have enough independence of mind to ask whether it could simply be wrong.

The New Philosopher-King

Plato's idea should therefore be changed, not adopted exactly as he proposed it. We do not need philosophers to become kings. We need kings, presidents, prime ministers, ministers, generals, judges, regulators, business leaders and everyone else with significant power to develop some of the philosopher's habits of mind. They must learn to doubt easy certainty, especially when technology makes it easy to create convincing appearances of certainty.

They must understand that intelligence is not the same as wisdom, prediction is not the same as morality, and finding the most efficient solution does not automatically make it legitimate. They need enough knowledge of how we know things to recognise weak evidence, enough understanding of ethics to see that different values cannot always be reduced to numbers, and enough knowledge of history to understand how often intelligent societies have convinced themselves to do terrible things. Above all, they need intellectual humility.

This may sound like a modest demand, but it is not. Humility may become the most important political virtue in the age of AI because artificial intelligence can make arrogance extremely powerful. A politician who thinks too highly of his own genius might once have given bad speeches and created foolish policies. His successor may have the same inflated view of himself, along with machines that can put his ideas into practice across an entire society at extraordinary speed. A fool with a bureaucracy is a problem. A fool with artificial superintelligence creates a constitutional emergency.

The Last Human Decision

People often frame the political questions about AI in the wrong way. We ask whether machines will become conscious, take people's jobs, become more intelligent than humans or eventually take control of governments. These are important questions, but they distract us from something that is already happening: people are choosing to hand their judgement over to machines because it is convenient.

Navigation apps choose our routes. Algorithms choose our entertainment. Recommendation systems decide what we read. Automated systems flag financial transactions as suspicious, decide which job applications receive attention and select which information seems relevant. Each decision may seem unimportant on its own, but together they create a habit: when a machine gives an answer, accept it. Governments face the same temptation, especially because government offices value efficiency and politicians are always short of time.

The danger may not be a dramatic morning when AI announces that it has taken power. The transfer of power could happen quietly. Ministers would still sit in their offices, parliaments would still vote, courts would still issue rulings and presidents would still give speeches. The outward signs of constitutional government would remain, while real decision-making power gradually moved into systems whose recommendations humans increasingly approve without question. The king would remain on the throne, but he would have stopped thinking for himself.

This would turn Plato's idea upside down. Instead of philosophers becoming kings, kings would become operators of machines whose intelligence they could not understand. Human beings would still officially hold political power, but more and more of the actual judgement would be done by computers. Responsibility would remain with people who no longer played a meaningful part in making the decisions they were supposed to be responsible for.

Plato's Revenge

The great irony of artificial intelligence is that the more intelligent our machines become, the more important it becomes for humans to develop kinds of intelligence that machines cannot take off our hands. We need good judgement because calculations are becoming cheap and plentiful. We need scepticism because convincing answers are becoming easy to produce. We need to take moral questions seriously because using power is becoming easier. Philosophy is therefore moving from the edges of political life towards its centre.

This does not have to mean academic philosophy, and certainly not the professional study of tiny disputes that only specialists can understand. What matters is philosophy in its older sense: carefully questioning what we know, what we value and what we should do. These questions become more important, not less, when a machine can give apparently authoritative answers to almost everything else.

Plato believed that political disasters happened because societies gave power to people who did not understand themselves or what was good. We rejected his solution because constitutional democracy offered a better approach: instead of trusting rulers to be wise, we limited their power because we knew they might not be. Artificial intelligence means we must now do both.

We need stronger constitutional limits because rulers are gaining more powerful tools. But we also need better rulers because no constitution can predict every decision in a world changed by technology whose effects may unfold faster than lawmakers can respond. The philosopher-king must therefore return, but without his crown. Instead, he becomes a standard of thought for everyone who holds power. They must question assumptions, tell knowledge apart from confidence, understand competing values, accept challenges to their beliefs and recognise the limits of their own understanding.

Artificial intelligence may become better than humans at almost everything we once thought made us intelligent. It may calculate faster, remember more, recognise patterns we cannot see and eventually reason across more knowledge than a human mind could absorb in a lifetime. None of this makes political philosophy outdated. Instead, it makes philosophy more urgent. The more impressive a machine's answer becomes, the more important it is for a human being to ask whether it is answering the right question.

The frightening future may not be one in which machines become our rulers. It may be one in which our rulers remain entirely human, with all their vanity, fear, ambition, loyalty to their own groups, impatience and occasional spectacular stupidity, but now have access to machines with almost godlike abilities. Plato could hardly have imagined such a combination, but he would have understood its danger immediately.

The intelligence behind the throne may grow more powerful every year. That does not mean we should hand the throne to philosophers, much less to machines. It means that the people sitting on it must develop the philosophical habits needed to understand the limits of power, knowledge and themselves. The more powerful the intelligence behind the throne becomes, the more urgently we need a philosopher sitting upon it.