To be fair, it’s easy to notice: the AI-generated picture of Alex Jones the Atlantic uses in their recent article does look off. A savvy observer might pick out the tell-tale gibberish text attempting to label the various objects around Jones, and any layperson would probably notice oddities in the image’s composition and detail.
But more surprising than the fact that perfect artistry yet eludes our most advanced machines is the fact that the Atlantic—a mainstream, establishment publication if there ever was one—decided to use one of them to begin with.
For years creative pursuits were seen as some of the least vulnerable to artificial replacement, even as the notion that such replacement would occur grew fashionable. “Sure,” the argument went, “AIs might augment the creative process, but they can never unseat humans as creative visionaries because they lack their own creativity and thus can never create ‘true’ art—artsits are safe, or at least won’t be replaced soon.”
And yet, the image works. Minor backlash on twitter1 notwithstanding, I doubt the median Atlantic reader even noticed the difference, let alone took offense to it. At worst it’s a neutral backdrop to the story, just like pretty much anything else the Atlantic might have used, but even on a thematic level it works pretty well.
The AI rendition of Jones is obviously distraught, even backed into a corner, reflecting his precarious legal situation and unknown future the article chronicles, and the abstract presentation compliments the articles pessimism and uncertainty. A haunting, paperlike background might evoke pending litigation or press coverage, and harsh lighting and the general composition also parallel the article, suggesting a deposition or even an interrogation. You could even take the jumbled mess at Jones’ left2 hand to be an allusion to his supplement-hawking business.
You might think this is a massive reach on my part, and yeah, a little. But no amount of reaching will change the fact that the composition of the picture reflects the general tenor of the article, and that it serves its purpose well.
More importantly, though, arguably the entire raison d être of art is to solicit reaches like this. Practically the definition of art is something that elicits an emotional response, and it’s not like this response has to artist’s intent. The actual content of the piece is still relevant—anyone would be hard pressed to interpret this very pleasantly—but more for suggesting general themes or moods than a specific interpretation; the viewer still has to do a lot of the work to fully appreciate it.
Indeed, this dynamic is precisely why people enjoy art to begin with—any art lover knows that the process of reflecting on an arts themes and how they relate to them can be intellectually engaging and deeply rewarding.
But practically this also means that an AI only needs to find something in roughly the right concept space to arrive at a final product that will be deemed appropriate by an audience willing to engage.
Thus, AI has completely sidestepped the pessimist’s argument. It might not be ‘creative,’ whatever that means, but as it turns out, the creative process that produces the work has no bearing on the creative process that interprets it, and the later is what consumers of art are seeking, making the former irrelevant, or at least forgotten.
That’s why the AI works so well here. As long as people are willing to engage with AI art the same way they do with human art3, AIs will have a much easier job making it than they will, say, writing a legal contract or designing an engineering schematic.
Sure, the later two pursuits can also be augmented by AI, and indeed already are, but ultimately they have to say something specific, and they have to say it precisely. An Atlantic reader will probably accept a header image to relate to the thrust of an article because of its vibes; a federal judge will not settle a case using similar framework.
As such, I expect AIs will continue their conquest of the creative fields at a blistering pace, faster than they make inroads in other fields. As the Atlantic illustrates, it’s already more than capable of replacing background art and stuff people don’t think about, but this will inevitably expand as AI continues its rapid development and develops more capabilities. Suspension of disbelief can extend very far, and it will always extend further for creative works than practical matters.
This isn’t to say humans will be muscled out of the art world entirely. AIs still need to be directed and curated, and sometimes monkeys just want to see what other monkeys can do, so it’s hard to imagine ‘artist’ will die as a profession.
But the reality is, a server is cheaper than a person, and can also produce a lot more art—starting now (or maybe 3-5 years ago), artistic and creative jobs will slowly evaporate, and people won’t notice the difference save for the usage of more AI-friendly art styles in lieu of Corporate Memphis4.
Not that every other job will be replaced much more slowly, or that anyone will notice when that happens, either. But it’s at least a little amusing that the first domino to fall to the AI takeover will be the one long thought to be the safest.
I’m sure the powers that be have duly considered the implications of this and are carefully adjusting our global AI adaptation strategy to account for it.
Your right.
And certainly they seem to be willing to now, and I predict this dynamic will persist if for no other reason than because people won’t typically notice if a given piece of art had biological or digital origins.
Unless good old CM turns out to be a maximally AI-friendly art style, which I certainly concede might be the case.
Any thoughts on whether this is, you know, good?
I was here.