At the Edge of the Abyss

Artificial intelligence no longer merely assists artists. It produces whole works. This is not only a technical shift; it puts a vertiginous question to us. What is left of the human creative act once the algorithm becomes the author?

More than two thousand years ago, in Plato’s Phaedrus, King Thamus receives the invention of writing with suspicion. The god Thoth presents it as a remedy for forgetting. Far from assisting memory, the king warns, the technique will be its simulacrum, producing the appearance of wisdom at the expense of real knowledge, the kind born of living dialogue. Writing, that pharmakon, remedy and poison at once, would externalise thought, for better and for worse.

A new technology places us before the same dilemma. Generative artificial intelligence, able to compose a symphony, write a poem or paint a picture in seconds, presents itself as the ultimate auxiliary to creativity. But does it not risk emptying the creative act of its substance, automating what until now belonged to effort, to doubt, to existential commitment?

The question is no longer whether AI can create. The works exist, often startlingly coherent in style. The deeper interrogation is what this new power does to our own creativity. If AI becomes a partner, or even the principal agent in the artistic process, how does that reconfigure the identity of the artist, the meaning of the work, and the very foundations of originality?

At the root of the creative act

To grasp the scale of the tremor, we must first map what it is we are trying to defend. The notion of creativity has a history of its own. First perceived in antiquity as a divine madness, an inspiration exceeding the human, it was progressively interiorised during the Renaissance and became, with Romanticism, the attribute of genius: an innate force, singular and inimitable. For Kant, genius is that “natural gift which gives the rule to art”, a talent that can neither be learned nor imitated.

It is precisely here that AI appears to fail. Margaret Boden offers an illuminating typology, distinguishing three forms of creativity. The first, combinational, assembles familiar ideas in new arrangements. The second, exploratory, works systematically through the possibilities a style or a set of rules affords. In both of these AI excels. Fed on millions of works, it is a virtuoso of recombination and of stylistic extrapolation.

The third form escapes it. Transformational creativity means breaking the rules themselves, changing the paradigm, making possible what was previously unthinkable. A rupture of that kind is not a matter of statistics but of a radical decision, and that decision remains, for now, proper to the human.

This transformational capacity rests on three things algorithms struggle to simulate. First, intentionality: the creative act is about something. It is directed by a vision, by the desire to express an idea or an emotion. The point is not merely to produce a result but to orient it toward a meaning. Second, emotional resonance: art draws its force from lived experience, from joy, suffering, wonder. AI can analyse and reproduce the codes of pathos, but it feels nothing. It manipulates the syntax of emotion without knowing its semantics. Third, cultural embeddedness: a work is never born in a vacuum. It converses with a history, with traditions, with social struggles. Punk rock is not a matter of saturated guitar chords; it is a revolt against social and musical norms. AI can ingest cultural corpora, but it does not inhabit the world that gives them their weight.

From the hand to the command

That philosophical distinction made, we should look at how AI actually inserts itself into the creative process. Examined closely, that process reveals particularly sensitive zones of friction. AI is a formidable assistant for technical tasks, for sorting or suggesting. Its influence becomes a problem when it reaches the heart of creation: the generation of the idea itself. In that sensitive zone, the potential of the machine and human subjectivity come into direct conflict.

More troubling still is chaining: the capacity of recent models to take charge of several successive stages of the process at once, from the search for inspiration through to final production. The artist, yesterday an artisan bent over their material, risks becoming a curator, a conductor of algorithms. The role shifts from direct making toward the engineering of prompts and the selection of one option among an infinity of machine proposals.

This mutation redefines the nature of talent itself. Mastery of an instrument or a technique gives way to dexterity in conversation with the system. In music the interface is no longer the neck of a guitar or the keyboard of a synthesiser but, increasingly often, language. Make me a track, melancholic and hypnotic at the same time. Creation becomes an act of verbal specification, an abstraction that holds at a distance the bodily commitment, the unforeseen gesture and the resistance of matter that have always nourished art.

The perils of a creation without friction

This apparent ease carries its dangers. The first is the erosion of craft. The sociologist Richard Sennett has shown how far patient engagement and the confrontation with material difficulty are essential to the construction of meaning and of motivation. A frictionless creative process, in which every obstacle is smoothed away by the algorithm, risks producing works that are polished and soulless, and depriving the artist of the deep pleasure that comes from overcoming a constraint.

The second peril is cultural homogenisation. Theodor Adorno, in his critique of the culture industry, was already warning of the standardisation of art for commercial ends. AI could carry that tendency to its extreme. Learning from enormous databases, models identify the patterns that succeeded in the past and reproduce them. A feedback loop sets in: AI generates works resembling yesterday’s successes, platforms promote them because they match what audiences expect, and those works in turn feed the next generation of models. The result is an aesthetic dead end, a world of infinite variation on the same themes, in which genuine rupture, the avant-garde, struggles to appear.

Bernard Stiegler saw in this process the risk of disindividuation: the loss of the capacity, for individuals and for collectives, to forge their identity and their future through singular symbolic works. Culture would no longer be reinvented, merely reproduced.

But the deepest danger, and the most subtle, may be the loss of the abyss. Creating has always been a confrontation with the unknown, a leap into a void of possibilities. The artist stands at the edge of an abyss, and each decision, each note, each colour, is an irreversible commitment that closes off other paths. That existential weight, that risk, is what gives a work its density. AI replaces the abyss with a latent space: a universe of possibilities that is vast, certainly, but entirely mapped by statistics. The unknown is replaced by the probable. The irreversible decision gives way to an infinite continuum of reversible options. Creation, once an adventure, becomes optimised navigation across marked terrain. What remains when there is no longer any void to face?

Toward meta-creativity?

Should we then see in AI the end of art? Perhaps we are witnessing instead the emergence of a new paradigm: meta-creativity. In this model the creative act no longer consists in shaping the work directly, but in conceiving, guiding and constraining the system that generates it. Creativity becomes an epistemic virtue, a capacity to ask the right questions, to find problems rather than to solve them.

The vision is not without nobility. After all, as the philosopher Andy Clark suggests, the human being is a natural-born cyborg, one who has always incorporated technologies to extend their cognitive reach. AI could be the most powerful of these mental exoskeletons.

The tension remains, even so. The real danger is not so much a super-intelligent AI that would replace us but, as the philosopher Anne Alombert writes, “the industrialisation of thought and the automation of alterity”. The risk is that, in the name of efficiency and engagement, we delegate to machines not only the tedious tasks but the very capacity to think differently, to produce that rupture, that radical alterity which drives cultural evolution.

We stand at the beginning of this era. The myth of Thoth does not tell us to refuse writing, but to remain alert to its double nature. In the same way, the challenge is not to reject AI but to bring it into our practices in a way that enlarges our creativity rather than diminishing it. It is a matter of preserving spaces of friction, of slowness and of doubt, of continuing to cultivate the irreplaceable dialogue between a body, a mind and matter. For it is perhaps in that struggle, in that confrontation with the abyss, that what will always remain beyond calculation in us has its place.