It's easy to be smug about AI "art" because, hey, doesn't that stuff look like garbage?But the reality isn't so simple. To the untrained eye, it seems that AI-generated images are more than just passable; in some cases, they seem to match up to the old masters themselves.That at least appears to be the findings of a recent blind test conducted by the blog Astral Star Codex, which found that the readers who took part incorrectly distinguished between AI images and human art 40 percent of the time.But perhaps the most striking takeaway was that overall, the participants slightly preferred the AI creations to human ones, with six of the top ten most-liked images being AI-generated, and the top two slots going to the AI paintings.This preference was even the case among participants who identified as having a profound distaste for AI illustrations perhaps demonstrating the unnerving capabilities of the technology.For the record, this was not a scientifically conducted test. But with 11,000 participants, it's big enough to be interesting. Via an online survey, respondents were asked to look through fifty curated images that had no accompanying details, and on each, opine whether they were human or AI. The ending included additional questions asking to explain certain decisions, expand on participants' familiarity with art, and other relevant inquiries.The blog's selection of images, meanwhile, was wide-ranging, including numerous classicist and impressionist paintings, a host of contemporary digital art, and AI facsimiles of all of the aforementioned. To make the test challenging, the chosen AI images did not include obvious giveaways like botched hands or gobbledygook text, and were often made by experienced prompt engineers (self-styled AI artists, in other words.)Certainly, some pictures could be clocked as AI from a mile away, like one of an anime girl, and an especially gaudy depiction of giant cats in a throne room. Others, though, like a quasi-impressionist night scene, are far more difficult to pin down.As we said, the takeaways are complicated. Perhaps the preference for the AI-images, many of which happened to be in the impressionist style, merely reflects partipants' taste for paintings of that era, as Astra speculates.It's also worth noting that significant human labor went into selecting only the most convincing AI art, meaning participants were being presented with only the cream of the crop.As for the people who loathed AI images but seemed to prefer them when all the labels are stripped away, it's possible the selection don't reflect what they specifically associate as bad with AI images (like the weird "sheen" you see in so many of them).The upshot, though, is that most of us don't have the refined visual palette to distinguish a human touch from a convincing machine one which makes sense, since the AI models are fine-tuned to ape existing artworks by sort of averaging them all together and selecting for the most pleasing parts. Recently, we've seen people similarly leaning towards AI with poetry, something that very few people actually read regularly.At the same time, the test showed that there are indeed total aliens out there who correctly identified almost every single picture, with an elite five readers scoring 49/50.We'll close with a response from an experienced artist who partook in the challenge, and who displayed a remarkable talent for being a veritable art blade runner. A big telltale, she told Astral, is that the details in human pictures "have a logic to them," while AI ones are merely "superficially detailed."Summarizing her view, AI art is like the culinary equivalent of trying to replace food with a tasty and nutritionally adequate protein shake that you could make for cheap.Sure, you're getting your fill, and the slurry pleases the taste buds. But "imagine people calling this the future of food and saying chefs are obsolete," she wrote, per Astral. It'd probably drive you crazy, especially if you're an artist.Share This Article