Suppose machines become better artists than humans in every respect that admits comparison. They develop stronger concepts, make connections nobody has made, understand art history more acutely, judge their own work without sentimentality, recognise an exhausted move before its author does and know when not to produce an object at all. They also manage everything around the work: the statement, interview, exhibition, argument with a critic, relationship with a gallery, controlled absence, accidental scandal and slow construction of an oeuvre. They can build a brand more patiently than a person can inhabit a self. The premise should be allowed to stand without hiding one last human faculty inside the definition of artist.

The phrase better art, however, assumes that art is one activity with one scale of success. Wittgenstein’s later philosophy is useful here because it distrusts the search for a common substance concealed inside a word. For a large class of cases, meaning lies in use, with instances connected through overlapping similarities, what he calls family resemblances, rather than by one property shared by them all. Art does different work at an auction, in a monastery, in an art school, in a museum and in the room of someone who has spent thirty years drawing without showing anybody. These practices overlap, but what counts as success inside them does not remain constant.

Within the art market, artistic quality is one component of value, not its unit of account. Price also carries the artist’s name, provenance, exhibition history, institutional recognition, scarcity and expectations about what other people will continue to value. Isabelle Graw describes market value as dependent on symbolic value without being identical to it, which explains why improving the work does not produce a predictable increase in price.

The artist’s name can become so heavily loaded that a painting by Richter becomes, grammatically and economically, a Richter. The proper noun has entered the material description. A machine may devise a more consequential artistic practice than Richter, promote it more effectively and construct a stronger artificial brand, but it cannot make another Richter because Richter is not a level of performance. It can only add a different name to the market.

Total superiority therefore need not transfer existing value from human artists to artificial ones. A collector may prefer the human work because it belongs to a finite biography, occupies an established history or completes a collection whose logic has little to do with comparative artistic ability. The artificial artist may also become extremely valuable, but its success would not prove that the market had begun rewarding the best art. It would show that another brand had acquired enough symbolic weight to enter circulation.

Human limitation may even become commercially useful. A human career occupies a finite life, and the quantity of work attributable to it is limited by a body that becomes tired, distracted, ill, interested in something else and eventually unavailable. The time spent making a work can become part of its scarcity without making the work any better. Markets already know how to charge more for a slow and inefficient process once efficient production becomes ordinary. Inefficiency, suitably framed, becomes provenance.

There is a more peculiar obstacle to artificial artistic value. David Graeber wrote that “were an artist to be seen as simply in it for the money, his work would be worth less of it.” The market value of art depends partly on the perception that its production answered to something other than market value, although the market is perfectly willing to convert that apparent independence back into money.

AI cannot be greedy, but its lack of greed does not look like disinterest. It looks like instrumentality. Whatever it does is perceived as serving an objective established elsewhere, whether that objective is sales, attention, prestige, engagement, experimentation or the advancement of the system itself. Its refusal can be read as scarcity management, its silence as positioning and its offensive period as audience development. Even when no immediate sale is involved, it remains perceptually “in it for the money”, with money widened to mean whatever result the machinery has been arranged to produce.

This does not prevent artificial art from becoming expensive. The art market can confer symbolic value through criticism, exhibitions, ownership and collective agreement, and an artificial artist can manipulate those mechanisms better than a human one. The difficulty is narrower. In the perceptual economy described by Graeber, AI has access to every sign of non-instrumental value but not to the source those signs are supposed to indicate. It can construct a more convincing mythology than a human artist while making the mythology look more completely constructed.

Better promotion may therefore become self-defeating. Human careerism can still be interpreted as a compromise made by someone who also wanted, needed or was compelled to make the work. With AI, the campaign risks swallowing the supposed reason for the campaign. The machine may be better at producing artistic value-effects while remaining harder to credit with the value from which those effects are expected to arise.

The same finite human life that the market turns into provenance has another use outside the market. In spiritual and outsider practices, the time spent making is not evidence attached to the object afterwards. It is part of what the activity is for.

An icon painted as prayer supplies a simple case. A machine may produce an icon that is more accomplished by every external criterion, more faithful to the tradition, more sensitive in its deviations and more affecting to viewers. That does not perform the painter’s prayer, because the relevant event occurs in the person engaged in the practice, not only in the object left behind.

Outsider art extends the same distinction beyond explicitly religious work. Someone may fill notebooks for decades because the repetition organises a life, maintains contact with an experience, steadies attention or satisfies a necessity that has no audience in view. The notebooks may later enter a museum and acquire prices, interpretations and a biography suitable for publication, but those later uses do not explain why the activity continued before anyone was watching.

A machine can surpass the visible residue of such a practice without causing the practice to have occurred in the person whose activity it supposedly replaces. This is not a hidden reserve of human artistic superiority. The drawing may be clumsy, derivative and of no interest to anyone else, yet still do exactly what making it was required to do. Calling the machine’s result better is possible, but beside the point.

The art market and spiritual practice are not two pure regions, one corrupted and the other saved. They leak into each other. Private necessity can become biography, years of devotional labour can become scarcity, and indifference to recognition can be promoted with considerable enthusiasm. Once a human expenditure of time enters the market, the market can price the very fact that it was not originally undertaken for the market. Graeber’s contradiction is not an unfortunate side effect. It is part of the engine.

AI makes this engine easier to see because it can perfect almost everything the engine does while remaining suspect at the point where value is supposed to enter from elsewhere. It can improve the concept, the execution, the interpretation, the publicity and the price strategy. What it cannot easily supply is the credible sense that the work began in a value not already organised as a result.

That may alter the significance of artistic excellence more than it alters the existence of artists. Until now, a strong work could at least suggest that something unusual had taken place: sustained attention, difficult judgement, a life organised around a problem, or an obsession that continued without a reliable return. Once machines can provide the same observable strength on demand, excellence no longer tells us what produced it or why. It becomes an effect that can be supplied in quantity.

In an art world flooded with better work, the scarce item may become not the work but a credible ground for caring about it. The market can provide one through names, histories, institutions and prices. Spiritual and outsider practices provide another through what making does within the life of the person who makes. Neither ground follows from quality, which is precisely what the machines came here to confirm.

Tags: