Criticisms of AI-generated art are now familiar, ranging from the unauthorized use of artists’ works to train models, to the aping of art styles that border on plagiarism, to claims that AI enthusiasts fail to understand that creating art requires intention and purpose and is antithetical to being produced automatically by a program. While these criticisms remain as important as ever, AI programs continue to evolve, and with new capabilities come new issues.
Case in point: Google’s new Genie 3, which allows users to create “interactive, playable environments” on the basis of prompts and images. To demonstrate the technology, a Google researcher showed how one could “walk around” famous paintings, such as The Death of Socrates by Jacques-Louis David and Nighthawks by Edward Hopper. The AI program generates a 3D world, allowing users to see characters and objects in the original painting from different angles, essentially creating a kind of videogame, albeit one in which there’s not much to do (at least for now).
I think there is good reason to be critical of AI that makes rudimentary videogames out of works of art. Here I’ll consider three such criticisms. The first two can be commonly found in the comments on social media: first, that using AI to digitally manipulate art is disrespectful to the artist or artwork itself; and second, that choosing to interact with the videogamified artwork represents a failure of imagination on the part of the user. I’ll also consider a new version of an old criticism from the philosophy of art: that AI-generated creations lack the original artwork’s aura.
There is a sense in which manipulating art in this way isn’t new. After all, so-called “immersive experiences” have been popular for a while now, such as Immersive Van Gogh, where visitors can walk among projections of some of Van Gogh’s most recognizable artworks. These experiences are sometimes criticized for being tacky and a tourist trap, but few would consider them egregious crimes against art. It’s also long been accepted by all but the stodgiest scholars that videogames are capable of being aesthetically valuable, so it’s not as though we should think that only oil paintings in ornate frames hanging in galleries are worthy of our aesthetic appreciation.
So what’s wrong with using AI to create a virtual world out of a painting? First off, we might worry that using these programs disrespects the original artist, who likely did not intend their work to be a virtual environment to be walked around in. Part of the problem is that AI programs struggle with generating environments that are coherent, producing artifacts and noise that detract from the original composition of the work of art. In the example of the world created out of Hopper’s Nighthawks, for example, AI-generated faces and words became garbled messes, with the end product feeling akin to vandalism.
This first criticism is an aesthetic one: AI programs that videogamify art ruin the artist’s vision, taking something beautiful and making it grotesque. We might also be tempted to criticize the person who chooses to engage with an artwork via its AI-generated videogame form. Commenters on social media are particularly liable to sling this kind of mud, accusing AI art fans of exhibiting a wide range of personal failings. While social media tends not to feature the most careful debates, one criticism that is worth singling out is that engaging with AI-manipulated versions of artworks represents a failure of imagination.
Why think this? Part of what’s involved in appreciating an artwork is to engage with it on its own terms, which requires interpreting what the artist has put in front of you and what they have left out. We might argue that getting an AI program to fill in the blanks by creating a navigable 3D environment is like taking a shortcut, where you are getting a program to do the work required to appreciate a work of art for you.
We’ve seen this kind of criticism when it comes to people using chatbots to write for them: writing is meaningful when it is intentional and effortful, and it loses that meaning when we offload our cognitive functions to programs. In the same way, using an AI program to generate a world out of a painting offloads your imagination and prevents you from being able to meaningfully appreciate a work of art.
So, the first criticism of AI videogamified art pertains to how a person treats an artist or artwork, and the second is a criticism of the person who uses such programs. The last argument I’ll consider is a bit different: that turning an artwork into a 3D virtual environment provides a subpar aesthetic experience because it fails to capture the original artwork’s aura.
This argument (or at least a form of it) comes from the philosopher Walter Benjamin, who wrote on art and aesthetics in the first half of the 20th century. Benjamin was concerned with a practice that was becoming more and more frequent at the time: that artworks were being reproduced, sometimes on a massive scale. An original painting, Benjamin argued, is unique, and when experienced in a certain place and time, has a presence about it, or what he calls an “aura.” It is a concept perhaps better experienced than described: there is some feeling that you get when encountering an artwork in a gallery as opposed to seeing a picture of it online, or as a postcard in a gift shop.
Benjamin’s worry was that copies of artworks fail to capture something that can only be possessed by the original. He did not, of course, have a conception of modern AI tools, or virtual 3D environments, or videogames. But Benjamin’s complaint still feels apt when experiencing new AI creations today: you’re no longer interacting with the original, but instead something that has been manipulated, and in doing so you fail to have the same kind of aesthetic experience. This criticism is not the charge that you’re necessarily lacking in imagination by engaging with the AI-generated version of a painting instead of the original; it’s just that it’s a shame that you’re missing out on having a more meaningful aesthetic experience.
How serious these criticisms are is up for debate, and many online have argued that new ways for AI programs to create and manipulate artworks really amounts to little more than cool new technology. Regardless, something of value does seem to be lost when interacting with the videogamified version of artworks instead of engaging with them on their own terms. When it comes to having a meaningful aesthetic experience, AI continues to feel like little more than a novelty.