AI can create art that feels human. Whose fault is that?

This was the year—ask your stockbroker or the disgraced management of Sports Illustrated—when artificial intelligence went from a dreamy projection to an environmental threat and a perpetual sales pitch. Does it feel like the future to you, or has AI already taken over the boring and scamming the now worthless, non-fungible token?

After all, artists have been using AI technologies for some time: Ed Atkins, Martine Syms, Ian Cheng and Agnieszka Kurant have been using neural networks and large language models for years, and orchestras played AI-produced Bach variations in the 1990s. . I suppose there was something useful the first time I tried ChatGPT – a slightly more advanced grandchild of Eliza, the 60s therapist chatbot – although I’ve barely used it since; ChatGPT’s hallucinatory falsehoods make it worthless to journalists, and even its tone seems like an insult to my humanity. (I asked: “Who was the better painter, Manet or Degas?” Answer: “It is not appropriate to compare artists in terms of ‘better’ or ‘worse’, as art is a very subjective field.”)

Yet the explosive growth of text-to-image generators like Midjourney, Stable Diffusion, and Dall-E (the latter named after the corniest artist of the 20th century; that should have been a clue) created the fear that AI came in for. culture – that certain capabilities once considered uniquely human were now confronted by computational rivals. Is this really the case?

Without specific cues, these AI images default to some common aesthetic characteristics: highly symmetrical composition, extreme depth of field, and sparkling and radiant edges that pop on a backlit smartphone screen. Figures have the waxy fruit skin and deep-set eyes of video game characters; they also often have more than 10 fingers, but let’s wait for a software update. There’s little here that I’d call human, and each of these AI photos is aesthetically irrelevant in itself. But together they point to a danger we are already facing: the devaluation and trivialization of culture to just one type of data.

AI cannot innovate. All it can deliver are rapid approximations and reconstructions of pre-existing materials. If you believe that culture is an imaginative human endeavor, then you have nothing to fear except that – what do you know? – many people have not imagined anything more substantial. When a TikTok user posted an AI-generated song in the style (and voices) of Drake and the Weeknd in April, critics and copyright lawyers said that nothing less than our species’ self-definition was under threat, and that a simpler kind of listener You can you wonder: was this a ‘real’ song? (A soulless engine that strings together a bunch of arbitrary formulas could pass as Drake – hard to believe, I know….)

A more appropriate question is: why is the music of these two confident Canadians so algorithmic in the first place? And one more: what can we learn about human art, human music, human writing, now that the good-enough approaches to AI have fully shown their barrenness and thinness?

As early as 1738, as the musicologist Deirdre Loughridge writes in her fascinating new book “Sounding Human: Music and Machines, 1740/2020”, the Parisian audience marveled at a musical automaton equipped with bellows and pipes, which could play the flute. They loved the robot and happily accepted that the sounds it produced were ‘real’ music. An android flautist did not in itself pose a threat to human creativity, but it prompted philosophers to understand that humans and machines are constantly intertwined, and to push artists to improve their playing. To do the same in the 21st century, we will have to take seriously not only what capabilities we share with machines, but also what sets us apart, or should do so.

I remain deeply relaxed when it comes to machines masquerading as humans; they are terrible at it. People who behave like machines – that is a much more likely danger, and one that culture, as the supposed guardian of (human?) virtues and values, has been unable to combat in recent years.

Each year, our arts and entertainment has further devolved into recommendation engines and rating structures. Every year our museums, theaters and studios have further internalized the technology industry’s reduction of human consciousness into simple sets of numbers. A score out of 100 for joy or fear. Love or pain, surprise or anger: all the same metadata. To the extent that AI threatens culture, it doesn’t come in the form of some cheap HAL-meets-Robocop fantasy of out-of-control software and killer lasers. The threat is that we limit ourselves to the extent of the limited capabilities of our machines; the threat is that human thought and life will be sanded down to fit increasingly standardized data sets.

It certainly seems that AI will speed up or even automate the composition of elevator music, the production of striking colors, festive portraits, and the writing of scenarios for multiverse coming-of-age quests. If so, then as Cher Horowitz’s father says in ‘No idea’, I doubt anyone would miss you. These were already, in every way, the results of ‘artificial’ intelligences – and if what you write or paint has no more depth or humanity than the creations of a server farm, then you certainly deserve your obsolescence.

Instead of worrying about whether bots can do what humans do, we would do much better increase our cultural expectations of people: expecting and demanding that art – even and especially art made using new technologies – testifies to the full extent of human powers and human aspirations. Ukrainian composer Heinali, whose album “Kyiv Eternal” I have kept close to me throughout 2023, reconstructed the wartime capital through beautiful reconciliations of medieval Gregorian chant and contemporary synthesizers. Nairy Baghramian’s sculptures, which I chased this year in Mexico City, in Aspen, in the garden of the MoMA and on the facade of the Met, use the most contemporary fabrication methods for the most fragile and tender forms. These artists are not afraid of technology. They are also not replaceable by technology. Technologies are tools for human flourishing.

I’ve been thinking a lot this year about stylistic exhaustion and the pervasive sense that culture in digital times is going nowhere fast. The concerns that accompanied artificial intelligence in 2023 reaffirmed these fears: that we have lost something essential between our screens and our databases, that content has conquered form and that novelty has had its day. If our culture has become static, can we call our cringe-worthy chatbots and insta-kitsch image machines what they are: mirrors of our diminished expectations?

Viewed this way, I might even ask myself whether AI is the best thing to happen to culture in years – that is, if these perpetual mediocrity machines, these supercharged cliché engines, will eventually force us to revalue the things that humans alone are . can do. After “a narrow fixation on how human machines can perform,” as Loughbridge writes, it is now time to figure out “what it means to work with them and exist in relation to them.”

To make anything count, you’re going to have to do more than just rearrange precedent images and words, like any old robot. You’ll have to put your back into it, your back and maybe your soul too.