The Critique of Artificial Beauty: What Kant’s Third Critique Can Teach Us About Artificial Intelligence

By Matthew Parish, Associate Editor

Monday 3 August 2026

The arrival of artificial intelligence has unsettled far more than economics, politics and warfare. It has begun to disturb one of humanity’s oldest assumptions: that creativity belongs uniquely to ourselves. Machines now compose symphonies, paint portraits, write poetry, produce films and design architecture with a fluency that would have seemed miraculous only a decade ago. They have entered territories once regarded as the exclusive domain of imagination.

If Immanuel Kant were alive today, he would almost certainly recognise that the deepest questions raised by these developments are not technological but aesthetic. They concern neither intelligence nor morality, but judgement itself.

His Critique of Judgement, published in 1790, sought to explain how human beings recognise beauty without reducing it to either objective science or subjective preference. It remains perhaps the most profound philosophical investigation ever undertaken into the nature of art, taste and creativity. Surprisingly, it also offers one of the most penetrating frameworks through which to understand the artistic achievements and limitations of artificial intelligence.

Kant’s starting point is deceptively simple.

When we describe something as beautiful, we do not merely report our personal preferences. Nor do we identify measurable properties in the object itself. Beauty occupies a curious middle ground. It gives pleasure without serving a practical purpose, yet it also carries an implicit claim that others ought to recognise what we see.

A beautiful landscape does not become beautiful because it is useful.

A beautiful melody is not valuable because it solves a practical problem.

Beauty possesses what Kant famously called “purposiveness without purpose”.

Objects appear perfectly ordered, perfectly harmonious, perfectly suited to themselves, even though no practical objective explains their form.

Artificial intelligence now creates precisely such objects.

Image generators produce astonishing paintings.

Language models compose elegant prose.

Music generators write convincing concert works.

Increasingly, audiences find themselves emotionally moved by works whose creator possesses neither heartbeat nor childhood.

Would Kant therefore conclude that artificial intelligence has become an artist?

The answer is likely to be more complicated.

Central to Kant’s aesthetics is the concept of genius.

Genius cannot simply follow rules.

Indeed, genius creates the very rules that later generations attempt to imitate.

The great artist does not assemble beauty mechanically according to pre-existing formulae. Rather, nature itself seems to speak through the genius, generating forms that surprise even their creator.

This presents an immediate difficulty for contemporary artificial intelligence.

Large language models and image generators are constructed precisely through the analysis of existing works. They discover statistical regularities across enormous bodies of human creativity. Their extraordinary power lies in recognising patterns that individual human beings cannot consciously perceive.

Yet recognising patterns is not identical with originating them.

One might say that artificial intelligence possesses immense taste but uncertain inspiration.

It recombines.

It synthesises.

It transforms.

Whether it genuinely originates remains much less clear.

Kant might argue that today’s artificial intelligence resembles an exceptionally gifted apprentice capable of mastering every established style while remaining dependent upon the historical tradition from which those styles emerged.

This observation should not be mistaken for criticism.

Human artists themselves rarely create ex nihilo.

Shakespeare borrowed plots.

Mozart developed inherited musical forms.

Picasso absorbed African sculpture and Iberian art before revolutionising modern painting.

Creativity has always involved transformation rather than spontaneous creation from nothing.

Artificial intelligence merely performs this process upon an unprecedented scale.

The more interesting Kantian question therefore becomes whether originality itself admits of degrees.

Can novelty emerge statistically?

Or must genuine genius involve something irreducibly personal?

The Critique of Judgment repeatedly emphasises the role of the imagination.

For Kant, aesthetic experience arises through the free play between imagination and understanding. Neither faculty dominates the other. Concepts fail fully to capture what imagination presents, while imagination continually exceeds conceptual determination.

This dynamic interplay produces the distinctive pleasure of beauty.

Artificial intelligence seems remarkably adept at simulating precisely this process.

It generates images that surprise.

It invents metaphors that appear fresh.

It constructs fictional worlds of extraordinary richness.

To observers these outputs often produce the same aesthetic response as human works.

Yet whether the machine itself experiences any corresponding free play remains entirely unknown.

The distinction matters.

A violin does not hear the concerto it performs.

A camera does not admire the landscape it records.

Likewise, an artificial intelligence model may generate aesthetic objects without ever participating in aesthetic experience.

Kant would likely insist that the creation of beauty and the appreciation of beauty, though intimately connected, remain philosophically distinct.

Even more striking is Kant’s discussion of the sublime.

Beauty concerns harmony.

The sublime concerns overwhelming magnitude.

Towering mountains.

Violent oceans.

Infinite stars.

These experiences initially defeat our imagination before reason ultimately reasserts its superiority by recognising ideas greater than sensory experience alone.

Artificial intelligence increasingly confronts humanity with a technological analogue of the sublime.

Its computational scale exceeds ordinary comprehension.

Its training data encompass much of recorded civilisation.

Its reasoning often appears simultaneously familiar and alien.

Human observers alternate between admiration and apprehension, much as travellers once contemplated glaciers or thunderstorms.

The machine becomes a new object of sublimity.

Not because it is beautiful, but because it exceeds ordinary cognitive expectations.

Yet Kant’s account of teleology may prove even more relevant.

The second half of the Critique of Judgment investigates organised living systems.

Living organisms display a peculiar characteristic: each part exists both for and because of the whole. A tree grows leaves that sustain roots which nourish branches which produce further leaves. Every component simultaneously serves and creates every other.

Artificial intelligence currently lacks this organic self-production.

It does not reproduce itself biologically.

It does not develop through intrinsic purposes.

It depends upon engineers, electricity, data centres and continual human maintenance.

It remains an artefact rather than an organism.

Nevertheless, future developments could complicate this distinction.

Autonomous systems capable of designing successor systems already exist in rudimentary form.

Artificial intelligence increasingly contributes to its own improvement.

The boundary between tool and evolving system becomes progressively less obvious.

Kant would almost certainly caution against drawing premature conclusions.

Apparent self-organisation need not imply genuine life.

Nevertheless, he would recognise that our conceptual vocabulary must evolve alongside technological reality.

Perhaps the most enduring lesson of the Critique of Judgement concerns the nature of criticism itself.

Kant believed aesthetic judgment could never be reduced entirely to objective rules.

Taste requires cultivation.

Education.

Conversation.

Historical awareness.

Communities gradually refine their capacity to distinguish profundity from novelty, permanence from fashion and genuine beauty from superficial ornament.

Artificial intelligence cannot replace this process.

Indeed, it may make it more necessary than ever.

As machines generate virtually unlimited quantities of competent art, the scarcity shifts elsewhere.

Beauty itself becomes abundant.

Judgment becomes precious.

The critic assumes greater importance than the creator.

The curator becomes as significant as the artist.

Human civilisation increasingly depends not upon producing more cultural objects but upon discerning which among them deserve enduring attention.

In this respect Kant’s Third Critique acquires an unexpected contemporary relevance.

Artificial intelligence may eventually compose music indistinguishable from Mozart.

It may paint landscapes rivaling Turner.

It may write novels worthy of international prizes.

Whether these achievements constitute genius, however, cannot be determined merely by technical excellence.

The question is philosophical.

It concerns intention.

Meaning.

Imagination.

Experience.

Above all, it concerns whether beauty ultimately resides in the object created, the mind creating it or the dialogue established between creator and audience.

Kant never claimed to possess final answers to these mysteries.

His ambition was more modest and, for that very reason, more enduring. He sought to explain the conditions under which aesthetic judgement becomes possible.

Artificial intelligence has not invalidated that project.

It has transformed it into one of the defining philosophical enquiries of the twenty-first century.

The first Critique asked whether reason could know the world.

The second asked whether reason could govern the will.

The third asked whether reason could recognise beauty.

Artificial intelligence now compels humanity to revisit all three questions simultaneously.

We have created machines that appear to reason, that increasingly influence our moral choices and that now participate in artistic creation. Yet the ultimate acts of judgment remain ours. We must still decide what counts as knowledge, what deserves to be called virtue and what merits the name of beauty.

Kant’s three Critiques therefore emerge not as monuments to an eighteenth-century intellectual world that has disappeared, but as an unexpectedly coherent philosophical guide to the civilisation artificial intelligence is rapidly bringing into being. They remind us that technology may alter the landscape of human achievement beyond recognition, yet it cannot absolve us of the responsibility to judge wisely what we have created.