What makes a piece of art? Its images? Its storytelling? Its realism? One crucial, but underappreciated element of any artwork is its limits. A painting’s canvas; a pop song’s three-minute structure; the action that remains off-stage in a play or film; the conclusion of a book or TV show. Great art is often just as much about what the artist has deliberately chosen to leave out as it is what we are shown.
As if to illustrate this fact, last week a series of tweets went viral that showed people using a new artificial intelligence tool, Adobe Firefly, to take famous photographs and paintings and create an extended version of them using generative AI – building out “more” of these works beyond their existing frames. “Ever wonder what the rest of the Mona Lisa looks like?” one user wrote in a particularly popular thread. “Here’s what the backgrounds of the most famous paintings in the world look like with AI.” What followed was images of paintings, such as Edward Hopper’s Nighthawks, Vincent Van Gogh’s The Starry Night and Hieronymus Bosch’s The Garden of Earthly Delights, expanded beyond the edges of the originals by Firefly.
1. Ever wonder what the rest of the Mona Lisa looks like?
— Kody Young / 💬 Your AI Interpreter (@heykody) May 26, 2023
Got @Adobe Firefly to help fill out the background for me with the power of AI
Here’s what the backgrounds of the most famous paintings in the world look like with AI: pic.twitter.com/2nkqESLrE9
While some praised these tweets as evidence of the “genius” of Firefly, the images were, for the most part, widely mocked. You can see why: this heavily-hyped, “cutting edge” technology generated the dullest, least imaginative new versions of these paintings (for the Mona Lisa, we get to see more river and trees; for The Starry Night, we get, of course, more starry night). This practice doesn’t yield any interesting, revealing or new results, and it certainly doesn’t qualify as “art”. But while it’d be easy to blame this trend on AI itself (and there is plenty to say about AI as “the future of creativity”), the development of this technology is reflective of deeper, pre-existing attitudes around how we view and value works of art.
The distinction between “art” and “content” is being erased. People see films, books, TV shows and visual art less and less as contained, intentional works, instead imagining them to be infinite and malleable. There is an overwhelming desire among audiences to constantly have more: more of the same; more of what is already in their comfort zone; more of what is familiar. The impact on art is that it has come to be expected that it should be safe and predictable, and crucially that it should affirm things we already know. This inclination contributes to the difficulty people have in seeing art as finite. There is not more of Nighthawks because Nighthawks is already a whole work – and yet people are awed by technology reimagining the most pedestrian parts of it.
Television is plagued by this attitude – shows are renewed, revived, remade. Some viewers might write their own epilogues to overcome their discomfort with open endings or finales they don’t like. (At the same time as these AI art pieces were going viral, the Succession actor Matthew Macfadyen gave an interview to Variety following the show’s finale, in which he was asked what he thought was happening to the characters now. He replied: “I haven’t given it any thought, is the honest answer. The story is stopped in my head… It’s stopped on that last image. That’s it.”)
What is art and what is merely content? Increasingly, it appears the most popular answer is that they are more or less the same. The former is being swallowed up by the latter; cultural works are stretched out to feed rabid audiences and meet companies’ bottom lines. This mindset leaves less room for works that are innovative, grotesque, flawed or surreal. In the world of content, everything must be relatable or familiar, designed to make us feel good, or feel nothing at all. But art isn’t supposed to exclusively deliver good feelings; it should be challenging, inventive and original.
There isn’t necessarily anything wrong with imagining an extended universe based on a piece of art – considering how a story might unravel further or what could lie just outside of frame. But in doing so, you aren’t extending the boundaries of a work of art you love but creating something else. Great art may be timeless, but it can never be endless.
[See also: Why Big Tech pretends AI is dangerous]