The rise of generative AI models has led to equal amounts of clapping and handwringing. One worry is that, as Kevin Kelly put it, “artificial intelligence can now make better art than most humans.” So where does that leave us?
The mistake is to assume that the meaning of “better” will stay the same. What’s more likely is that the goal posts will shift because we will move them. We have changed our collective tastes in response to technological progress in the past. We’ll now do it again, without even noticing that it’s happening. And if history is any indication, our tastes will evolve in a way that rigs the game in favor of human artists.
It’s not surprising that in conceiving of a new world awash with AI art, we haven’t accounted for a society-wide change in taste. We tend to assume that in the future we will want the same things we want now, and that only the ability to achieve them will evolve. One famous study dubbed this the “end of history illusion”: People readily agree that their most strongly held tastes have changed over the past decade but then insist that from this point on those tastes will remain as they are. Having presumably reached some peak level of refinement, they can now rest idly in their self-assurance.
In truth, what turns us on and off is constantly being reshaped by a range of powerful social forces, mostly beyond our awareness. Technological progress tops the list because it changes what is easy and what is difficult, and our running definitions of the beautiful and the vulgar are instantly affected by these criteria. When new advances expand the confines of what is possible, collective tastes respond—by wanting to partake of the new abundance and by wanting nothing to do with it.
I think of this as the William Morris effect. Morris was the bushy-bearded figurehead of what came to be known as the Arts and Crafts movement, which emerged in Victorian England in the 1870s. The timing was no coincidence: Britain had reached the peak of the industrial revolution. It had become the fastest-growing country on the planet, and London its largest city. For the first time, tableware, jewelry, and furniture could be made in factories, at scale. Such a quantity of goods had never been so accessible to so many.
Morris and his acolytes denounced the new abundance. They decried the soulless homogeneity of the machine age. In response, they looked to the past, seeking inspiration in medieval patterns and natural forms. Their designs were all intricate leaf patterns, elegant ferns, and curving flower stems. It was a radical move for the time, and the “medievalists,” as they were called, were mocked at first. But they quickly found a receptive audience. Just as technology was bringing mass-produced goods within reach of the middle class, under the influence of Morris and his acolytes, elite tastes turned to block-printed floral wallpapers and furniture purposefully left unfinished, the better to hint at its handmade origins. Soon, this fancy spread through English society. By the end of the 19th century, Arts and Crafts interiors had become the dominant style in British middle-class homes.
William Morris shaped British tastes, spawning imitators throughout Europe and across the Atlantic. But he was also a product of his time. The zeitgeist was waiting for a figure like Morris. The general unease with Victorian factory conditions and the dense London smog expressed itself through a sudden appreciation for intricate hand-drawn floral patterns. Time and again, technical advances change our sense of what is appealing or valuable. And as in 19th century Britain, the change often runs against the grain of technology, rather than with it.
So what does the William Morris effect imply for our newly expanded possibilities? How will the ability to spontaneously generate images by dictating a prompt to an AI reshape our idea of what is beautiful? Forecasting trends is a tricky business, but in this case, we don’t lack for clues.
A full 15 years ago, a team of researchers from University College London and the University of Copenhagen put people into an fMRI machine and showed them a series of abstract images. They told them the images were either made by a human or by a computer. A clear winner emerged. People not only claimed to prefer the (identical) human-made pictures, their brains’ pleasure centers actually lit up more brightly. What the researchers didn’t anticipate, but which is likely to happen, is that this visceral preference for human over robot makers might grow stronger with time, just as technology closes the gap between them. Think of it as humanity’s collective defense mechanism.
Why would people derive more aesthetic pleasure from an otherwise identical illustration, painting, or poem, merely because of how it was made, or by whom? It’s one of the distinctive quirks of modernity. Consider what a recent invention the notion of a “fake” is. Art historians estimate that half of all commissioned artworks in the 16th century were copies of originals. As long as they were competently executed, those copies were thought to be almost as valuable as the real thing. You see it in the prices each would fetch: In the Renaissance, an original painting went for about 2.5 times the price of a good copy. Now, that ratio might be closer to 10,000:1. A perfect replica of an Old Masters painting valued at millions might fetch a few hundred dollars, at most. The market for copies is so dismal that few painters even bother. Our tastes have long been shifting in a direction that might strike any AI as baffling, not to mention deeply unfair. It’s a learned preference, and we’ve been working hard at our lessons.
Today, the William Morris effect is once more upon us. The first-wave craft revival that Morris brought about was the precursor to our current yearning for “authenticity” in every guise. Just as an unprecedented expansion of international trade has made cheap goods manufactured abroad widely accessible, the Western consumer has become enamored of locally made small-batch mustard with handwritten labels. The distinction comes down to the presumptive identity of the maker, and what we like to assume of their intent.
In my book, Beyond Self-Interest: Why the Market Rewards Those Who Reject It, I argued that the incentives of capitalism lead us to value disinterested makers over self-interested ones. In a crowd of greedy profit-seeking market actors, the only ones we can trust are the obsessives who care more about their craft than their bottom line—or at least claim to. One paradoxical result is that doing things for their own sake has become a profitable move. We find individual passion reassuring, and this preference is not limited to farmers’ markets: Experimental evidence suggests that corporate managers view passionate workers as more competent, and they promote them more rapidly—even when those passionate employees’ output is actually lower. Artists face an extreme version of this whim; their market success depends on being seen as oblivious to market success.
The advent of AI models will only accelerate this trend. We will place ever more value on works that seem made for their own sake, rather than ours. That’s bad news for the AI robots, which are explicitly designed to please us. Engaging in a task for its own sake is one thing that, by construction, is beyond any AI’s ability. Trained on what has appealed to us in the past, they offer it back in new colors.
We will look upon these pastiches with increasing suspicion, scrutinizing the provenance of words and images. Books and movies will tout their full-blooded bona fides. We will regard these as “better,” just as we convince ourselves that small-batch mustard tastes more “real” than its supermarket equivalent. We will develop increasingly sophisticated means of telling the two apart, and technology will itself be enlisted in the effort.
The ground has already been set, which is often the case. It turns out the gothic revival had been in the air for over a decade by the time William Morris offered British elites hand-painted tiles from his atelier. Similarly, the AI revolution will elicit a further elevation of “authenticity” from consumers, which painters and illustrators and writers will pounce on. Far from signaling a decline of original art made by humans, the advent of AI will render it more precious by contrast. The gap between the artists and the robots will grow wider, just as their technical abilities continue to converge.
What actual form might our newfound preferences take? William Morris provides some additional clues. His greatest influence was the art critic John Ruskin, who was 15 years his senior and can be credited with launching the gothic revival that Morris capitalized on. Ruskin was a polemical thinker who united a set of aesthetic preferences with a zealous social philosophy. He not only had set ideas about church stonework, but also strong beliefs about social institutions. Inveighing against what he saw as the dehumanizing division of labor in Victorian factories, he held that makers should be involved at every stage of the making. “The painter,” he claimed, “should grind his own colors.” Morris himself embodied this idea, and it proved good business. Though he eventually found himself at the head of a thriving firm, he never stopped grinding his own colors; he remained obsessively involved at every stage of production.
Expect the trend to continue. We will demand works that can be attributed to an identifiable individual vision. The AI age will lead to a doubling down on biography, which happens to be another thing robots are notably short on. Already, there is grumbling over how major contemporary artists, from Damien Hirst to Jeff Koons, rely on vast studios of assistants to do the actual painting and sculpting as a way of keeping up with the demand for maximal scale and output. Expect the complaints to grow deafening, and the rote response, according to which even Renaissance artists delegated tasks to dozens of apprentices, to lose its potency. That may have been fine for Titian’s time, but we now have painting robot upstarts to contend with, and our tastes have grown fickle.
This isn’t to say that artists will not take up AI as a new tool. Even the Impressionist painters, who responded to the advent of the daguerreotype in the 19th century by going to places where photography couldn’t follow, relied on photographs as a sketching device for their own work. But AI creations will only be rescued by tethering themselves to an individual human vision.
It turns out we’ve been prepping for the AI revolution for decades, developing quirky tastes for the very kind of symbolic values—individual passion, purpose, lived experience—that robots will not exhibit anytime soon. That’s why AI is unlikely ever to produce “better” art than humans. Instead, it will transform our sense of sweet and sour. Our collective defense mechanism will kick in. It’s the robots who should be wringing their little grippers.