Dare to Know: The Enlightenment in an Age of Algorithms

By Matthew Parish

Thursday 27 August 2026

There are moments in history when old ideas acquire an unexpected new urgency. The Enlightenment is usually presented as something safely concluded: an eighteenth-century movement of philosophers, scientists, pamphleteers and political reformers whose intellectual victories eventually produced constitutional government, religious toleration, modern science and the liberal state. We inhabit, supposedly, the world that the Enlightenment created. Yet the strange conditions of the early twenty-first century suggest that this confidence was premature. The Enlightenment was never a completed historical event. It was an argument about how human beings ought to think and govern themselves — and that argument is now being reopened.

Social media has fractured the public sphere. Confidence in governments, universities, courts, journalism and other institutions has declined across much of the democratic world. Political debate increasingly resembles competing assertions of identity rather than rational disagreement about facts. Meanwhile artificial intelligence is acquiring an unprecedented capacity to mediate between human beings and the accumulated corpus of human knowledge. We are entering an era in which machines may become our principal interlocutors in deciding what is true.

Under these circumstances, the Enlightenment injunction Sapere aude — dare to know, or dare to think for yourself — may be more important than at any time since Immanuel Kant wrote his celebrated 1784 essay What Is Enlightenment?

Kant described Enlightenment as humanity’s emergence from its “self-incurred immaturity”. His point was not that people were stupid. It was that they were accustomed to allowing others to think for them. Priests determined religious truth, monarchs determined political truth and inherited authorities determined intellectual truth. The Enlightenment demanded that authority submit herself to reason.

This was an extraordinarily radical proposition. It did not mean that every person was equally knowledgeable about everything, nor that expertise was worthless. It meant that no proposition should ultimately be protected from rational examination merely because somebody powerful had pronounced it true.

The great Enlightenment thinkers disagreed about almost everything. Locke, Hume, Voltaire, Rousseau and Kant held substantially different views about religion, human nature, government and morality. What united them was something more fundamental: the proposition that disagreement itself need not be dangerous. Human beings could examine competing propositions, compare them against experience and argument and gradually distinguish stronger explanations from weaker ones.

From this intellectual disposition emerged many of the characteristic institutions of modern civilisation: constitutional government, independent courts, universities, scientific academies, a free press and eventually universal political participation. Their underlying principle was epistemological before it was political. Nobody possesses an infallible monopoly upon truth.

That principle is now under unusual pressure.

The digital counter-Enlightenment

The internet initially appeared to represent the ultimate Enlightenment technology. Knowledge that had once been restricted to universities and libraries became available almost everywhere. Political dissidents could communicate across borders. Citizens could scrutinise governments. Scientists could collaborate internationally. An individual possessing nothing more than an inexpensive computer could gain access to resources that would once have been available only to scholars at the world’s greatest universities.

Yet the internet subsequently developed in a direction that its early enthusiasts did not anticipate. The dominant institutions of the contemporary internet are not principally libraries. They are attention markets.

Social media companies discovered that human attention could be captured most effectively by appealing not to reason but to emotion. Anger, fear, sexual attraction, tribal loyalty and moral outrage produce powerful engagement. Algorithms consequently learned to promote material that stimulates these reactions.

This has created something resembling an industrialised counter-Enlightenment.

The problem is not merely that social media contains false information. Falsehood has existed throughout history. The more profound transformation is that social media encourages people to organise themselves into epistemological tribes. People increasingly encounter facts through communities that already possess conclusions about what those facts must mean.

Political identity thereby precedes reasoning.

A conservative and a progressive may observe the same event and inhabit entirely different informational worlds surrounding it. Russians, Ukrainians, Americans and Europeans may encounter radically different accounts of the same international crisis. Facts become badges of tribal membership. Changing one’s mind can consequently feel less like intellectual development than betrayal.

This is almost precisely the condition against which the Enlightenment rebelled.

The crisis of institutional legitimacy

There is nevertheless an important mistake in attributing the contemporary epistemological crisis entirely to social media. Traditional institutions contributed substantially to their own declining legitimacy.

Governments have lied. Intelligence agencies have made catastrophic mistakes. Banks and financial regulators presided over the crisis of 2008. Universities have sometimes allowed ideological conformity to masquerade as intellectual virtue. Newspapers have blurred the distinction between reporting and advocacy. Public bureaucracies have become remote from the populations they govern. Scientific institutions have occasionally communicated uncertainty poorly while expecting public deference nevertheless.

Citizens noticed.

It is therefore insufficient for established institutions to complain that the public no longer trusts experts. Enlightenment values do not require obedience to experts. They require experts to explain themselves.

Institutional legitimacy in an Enlightenment society must always be provisional. A court deserves respect because it gives reasons for its judgments and permits appeals. A scientist deserves respect because his conclusions can be challenged by experiments conducted by other scientists. A newspaper deserves respect because its claims can be checked against evidence. A university deserves respect because arguments can be advanced there without fear that an unpopular conclusion will be punished merely because it is unpopular.

Once institutions begin demanding trust instead of earning it, they become anti-Enlightenment institutions even when they retain Enlightenment names.

The solution to institutional illegitimacy therefore cannot be a restoration of unquestioning deference. That world is gone, and perhaps it ought to be. The solution must instead be radical institutional openness: reasons, evidence, transparency, tolerance of criticism and willingness to admit error.

Authority must once again become persuasive rather than compulsory.

Artificial intelligence and the new intellectual dependency

Artificial intelligence makes these questions vastly more important because AI may transform the relationship between human beings and knowledge itself.

The printing press reduced the cost of distributing ideas. The internet reduced it almost to zero. Artificial intelligence goes further. It reduces the cost of interpreting ideas.

This distinction is profound.

A person confronted with a difficult book traditionally had to read it. A person confronted with conflicting arguments had to compare them. A researcher faced with a large body of evidence had to organise it. Increasingly an artificial intelligence system can perform these tasks instantaneously.

There is enormous promise in this development. AI may become the greatest educational technology ever invented. A child in a poor country may obtain something resembling an infinitely patient private tutor. Scientists may interrogate enormous bodies of research. Lawyers may compare thousands of judgments. Doctors may synthesise medical literature. People previously excluded from elite educational institutions may gain access to intellectual resources that once required enormous wealth or privilege.

This is Enlightenment democratisation on an extraordinary scale.

Yet precisely the opposite outcome is also possible.

If artificial intelligence becomes the authority to which people reflexively turn whenever confronted with intellectual difficulty, humanity may recreate technologically the very condition Kant described as immaturity. Instead of priests telling us what to believe, algorithms will tell us. Instead of monarchs determining permissible opinion, computational systems may determine which explanations appear credible. The intellectual guardian will have returned, except that this time the guardian will speak pleasantly, instantly and apparently knowledgeably about virtually everything.

This would be a peculiar historical irony. Humanity might construct the greatest instrument for the dissemination of knowledge ever devised and use it to abolish the habit of independent thought.

The machine that sounds certain

There is a further difficulty. Artificial intelligence systems can express uncertain propositions with extraordinary confidence.

Human beings instinctively associate articulate language with intelligence and authority. A hesitant speaker appears less knowledgeable than a fluent one. Yet linguistic fluency and truth are different properties. An artificial intelligence may produce a beautifully reasoned explanation containing a factual error, an invented authority or an unjustified assumption.

The danger becomes greater as the technology improves. Primitive errors encourage scepticism. Plausible errors encourage trust.

The Enlightenment response cannot sensibly be to reject artificial intelligence. That would resemble rejecting printing because books sometimes contain lies. Instead we must develop a culture in which AI outputs are treated as propositions to be examined rather than revelations to be accepted.

The most valuable artificial intelligence may ultimately be not the machine that gives us answers but the machine that helps us reason.

Such systems should ideally reveal uncertainty, distinguish evidence from inference, identify competing explanations and encourage users to interrogate conclusions. They should make intellectual disagreement easier rather than silently eliminating it.

AI ought to become an instrument of Enlightenment rather than an oracle.

Truth in the age of synthetic reality

This becomes particularly important as artificial intelligence acquires the capacity to manufacture convincing synthetic evidence.

Photographs once possessed an evidential authority because creating a convincing false photograph was difficult. Audio recordings and video subsequently acquired similar authority. Generative artificial intelligence is rapidly destroying these assumptions.

A politician can apparently be shown saying something he never said. A battlefield atrocity can apparently be recorded even if it never occurred. A fraudulent telephone caller can reproduce another person’s voice. Entire communities of synthetic online personalities can be created to manufacture the impression of political consensus.

The immediate fear is that people will believe false things.

The deeper danger is that people will cease believing anything.

Once every inconvenient photograph can be dismissed as artificial intelligence, genuine evidence loses authority alongside fabricated evidence. The liar’s greatest achievement is not necessarily persuading everyone of a particular falsehood. It is persuading everyone that truth itself is inaccessible.

Authoritarian political systems understand this very well. Modern propaganda frequently does not attempt to establish a coherent alternative account of reality. It overwhelms the informational environment with contradictory explanations until citizens conclude that objective truth cannot be established. Cynicism then replaces belief, and politically exhausted populations retreat into private life.

This is why the defence of objective truth is not an academic luxury. It is a political necessity.

Enlightenment without naïveté

The Enlightenment nevertheless requires revision if it is to survive the twenty-first century. Human beings are not purely rational creatures and never have been.

We seek belonging, dignity, affection, status and meaning. Social media platforms are powerful precisely because they understand these needs better than many Enlightenment institutions do. An algorithm showing somebody material that confirms his identity may be psychologically more compelling than a professor presenting contradictory evidence.

A renewed Enlightenment therefore cannot consist merely of telling people to be more rational.

Reason requires a culture in which changing one’s mind is considered a strength rather than a humiliation. Intellectual uncertainty must cease to be interpreted as weakness. Political opponents must once again become people who may be mistaken rather than enemies who must be destroyed. Institutions must acknowledge that scepticism towards authority can be healthy. Citizens must understand simultaneously that scepticism does not entail believing every conspiracy theory encountered online.

The distinction between scepticism and cynicism is crucial.

The sceptic asks: What is the evidence?

The cynic declares: Everybody lies.

The first proposition is Enlightenment. The second is intellectual surrender.

The education of free citizens

This suggests that education itself must change.

For much of the twentieth century, education concentrated upon transmitting information. Students learned historical dates, scientific principles, literature, mathematics and geography because access to organised knowledge was comparatively scarce.

Information is no longer scarce. Attention and judgement are.

Education in the age of artificial intelligence must therefore increasingly teach people how to distinguish knowledge from assertion. Logic, probability, rhetoric, scientific method, source evaluation and basic epistemology should cease to be specialist academic subjects and become foundations of civic education.

Young people should learn how algorithms manipulate attention. They should understand why emotionally satisfying propositions may nevertheless be false. They should know how statistics can mislead, how photographs can be fabricated and how apparently authoritative language can disguise uncertainty.

Most importantly, they should learn that saying “I do not know” is not an intellectual failure.

It is frequently the beginning of wisdom.

Liberty as an epistemological principle

Freedom of speech also acquires renewed importance in this environment.

The traditional argument for free expression is often presented as a matter of individual liberty. Yet freedom of speech is equally an epistemological technology. Societies permit people to say foolish, offensive and unpopular things partly because nobody can reliably identify error in advance.

Truth is discovered through contestation.

This does not mean that every opinion is equally valuable. Nor does it imply that deliberate fraud, threats or incitement must be tolerated without limit. It means that societies should be extraordinarily cautious about creating institutions empowered to determine which disputed propositions may be expressed.

The Enlightenment insight is that authorities themselves can be wrong.

Indeed, history repeatedly demonstrates that they are.

The unfinished Enlightenment

We may therefore be entering not a post-Enlightenment age but a second Enlightenment crisis.

The first Enlightenment confronted monarchies, ecclesiastical authority, inherited hierarchy and superstition. The second must confront algorithmic manipulation, informational tribalism, institutional illegitimacy and synthetic intelligence.

The principles required are surprisingly similar.

Evidence matters.

Arguments matter.

Institutions must justify themselves.

Authority must remain open to challenge.

Individuals must retain responsibility for their own judgments.

Truth exists even when discovering it is difficult.

And human beings must never surrender the obligation to think.

Artificial intelligence will test these principles more severely than any previous technology because it enters directly into the cognitive relationship between individuals and the world. Properly developed and wisely used, it could become an extraordinary Enlightenment instrument — spreading knowledge, exposing weak arguments and making sophisticated intellectual inquiry available to billions of people.

Used carelessly, it could create the most intellectually dependent civilisation in history.

The choice will not be made by technology alone. It will be made culturally, institutionally and ultimately individually.

Every civilisation rests upon certain habits of mind. Liberal civilisation rests upon an unusually demanding one: the willingness to believe that nobody — neither king nor priest, professor nor politician, newspaper nor algorithm — possesses an unquestionable entitlement to determine reality for everybody else.

That conviction produced some of humanity’s greatest achievements. It produced modern science, constitutional government, religious toleration and the extraordinary expansion of individual freedom.

We should not assume that these achievements are irreversible.

The great danger of the contemporary era is not simply that artificial intelligence might become too intelligent. Nor is it merely that social media might become too powerful or institutions too distrusted. The deeper danger is that human beings, exhausted by complexity and overwhelmed by information, might gradually abandon the habit of judging things for themselves.

Against that possibility the old Enlightenment motto remains remarkably sufficient.

Sapere aude.

Dare to know.

And, perhaps more importantly now than ever before, dare to think.