In case you haven’t seen it, here’s the “burrito” that horrified a nation, linked out because I don’t want to get any emails saying Ravenous triggered their gag reflex. It looks like it’s filled with macaroni, or maybe maggots, and it briefly became the avatar of everything wrong with the recent onslaught of AI food imagery. No matter your stance on AI, it seemed that everyone could agree this was the most disgusting thing they’d ever seen.
More and more restaurants seem to be relying on generative AI “art” to promote themselves on social media and for signage, despite the fact that it could legally violate consumer protection laws. But the backlash is firmly here, at least in the media. Food & Wine, The Verge, Wired, and the New Yorker have all complained about the aesthetics of AI food slop. “Most of the cheap menu slop I see day-to-day still looks puke-inducing,” writes Reece Rogers at Wired. Mashable says these images are “haunting” social media, while CNN brought in a pro to explain why these images are so off-putting. Alexander Diel, a psychologist at Germany’s University of Duisburg-Essen who is studying AI food images, says that humans are deeply attuned to how food looks, both because it can nourish us or make us sick, “so we are very sensitive to when something is even slightly off.”
This is what triggers most people when they look at an image of fried chicken that somehow looks fuzzy. We eat with our eyes first, determining what is delicious and nutritious and what may poison us. And that skill, first vital to our evolution, now partially informs how we eat in typically less extreme circumstances. Scientists have shown that just changing the color of different foods and beverages can affect how much people like them, even changing the perceived flavor. These images have the ability to produce a profound reaction because we rely on them to survive.
I, too, find AI slop of incongruous sandwiches gross and unsettling. But it turns out that disgust is still in some part in the eye of the beholder. As every publication railed about how bad these images looked, I began to wonder if we were focusing on the wrong issue. Because the problem is, no matter how obviously disgusting you think something is, your opinion is subjective. And focusing on how ugly these images are only reveals that some people just do not notice or care what they look like. Some may even like them.
Often, these images are meant to be only halfway noticed in the first place. Most people don’t need an accurate photograph of a cheeseburger to know that’s what they want to order. And on that metric, these images often succeed. “I’m not sure I notice or care that much,” says Patrick Dennis, who says that he only started noticing the contours of AI food slop when people on social media started complaining about it. Even then, he says the images felt like a different version of the “artificiality” in much food photography.
Arguing that restaurant AI slop is bad because it looks bad is a losing game.
When I started asking if anyone just wasn’t bothered by AI food images, multiple people brought up all the old tricks, real or rumored, of food styling: turkeys sprayed with food coloring, mixing frosting and powdered sugar for “ice cream” that won’t melt, spritzing meat with oil to make it look juicier. On some level, we’ve already been trained not to trust what we see in food. AI imagery, they argue, is just the next evolution.
The people I spoke to didn’t really see the big deal with a deli posting AI sandwiches on Instagram. It didn’t bother them aesthetically, and for some, it’s even an improvement. “I don't think they look that good, but they certainly look better than 50-90% of flyers and posters that a small business owner would design in Canva or PowerPoint,” says Gabe Berlin, a graphic designer who says he is acutely aware of the way AI is encroaching on creative design work. And he’s right that even before generative AI, there was a lot of restaurant signage that just didn’t look good. As we long for the days when a business owner posted a photo of their offerings that looked like it was straight out of the Fyre Festival, we should remember it wasn’t long ago that we were making fun of restaurants for using stock imagery and bad photography.
For Berlin, the work that may have been lost and replaced with generative AI wasn’t worth the hassle, as he notes small business owners tended to underpay for art. And Sean Yamada, who says he’s against AI to the point that he’s been organizing to oppose data centers, notes that many of these businesses would not ever have been employing artists in the first place. Independent restaurants struggle with thin profit margins, and the profit motives of capitalism incentivize doing what is cheapest. AI food slop is likely replacing work that had been done by an owner themselves, which would probably also suck. “Have you ever seen that picture of the upside down peas?,” asks Yamada. “Not everything is a labor of love.”
In 2021, Jiayang Fang wrote about the Disgusting Food Museum in Malmö for the New Yorker, and the question of who gets to define what “disgusting” food even is. As an example of pure gustatory revulsion, she reminisces about immigrating to America from China and encountering her first non-Chinese meal, which she describes as square dumplings in a thick white sauce filled with a “gooey, creamy substance” that made her gag. It was cheese ravioli.
A heavy percentage of the “disgusting” foods in the museum hail from Asia, where they are not disgusting at all to the people who created them. “To be a new immigrant is to be trapped in a disgusting-food museum, confused by the unfamiliar and unsettled by the familiar-looking,” she writes. “The firm, crumbly white blocks that you mistake for tofu are called feta. The vanilla icing that tastes spoiled is served on top of potatoes and is called sour cream. At a certain point, the trickery of food starts to become mundane. Disgusting foods become regulars in the cafeteria, and at the dinner table.” This is all to say that what looks off-putting, what drives us to rear our heads back and hold our breath, is mutable.
We deserve to live in a beautiful world. Dunking on doner kebab that looks like a horrifying kitchen sponge feels righteous, because it’s just so obvious to many of us how horrible it looks. And we don’t even have to worry about insulting a hard-working artist — there isn’t one! But something that has gone unsaid is that restaurant owners using these visual tools don’t just use them because they have been made cheap and convenient, but because on some level they like the way the results look, whether that’s because they think they read as more “professional,” or they just think it looks cool. At the very least, they don’t believe the output is so offensive as to turn people away. If these business owners found the results disgusting, if they truly thought they were so gross that customers wouldn’t come, they wouldn’t be doing it. And clearly, the disgust at these images is not universal.
Arguing that restaurant AI slop is bad because it looks bad is a losing game. It pegs the argument to aesthetics, which are inherently subjective, and which mires the conversation in personal visual taste. The focus becomes the images themselves, not the mechanisms by which they are made. It’s like arguing that Shein’s biggest sin is that it makes ugly clothes, instead of the massive labor violations and environmental costs. And keeping the argument’s focus on aesthetics, while understandable, opens the door to the reasoning that, should the technology get good enough to no longer look so obviously creepy, there would be no problem.
That future is possibly imminent. One of the main arguments about AI food slop is that it’s immediately recognizable, and indeed it’d be hard to ignore a perfectly circular shrimp. But even when it’s not some uncanny visual horror, there’s something about the shading and contrast in these images, or the too-perfect roundness of the edges, that reveals its origins to the discerning eye. The artists in my life can spot it in a second, but not all of us have developed that skill. I’ve been in multiple conversations with friends trying to figure out if an image is AI or not, with some emphatically insisting that what, to others, is clearly AI must be a photograph. If you can’t tell, is it still slop?
These arguments also ignore that aesthetic values change, and what is considered ugly one day may be considered cool and countercultural the next. New York tour guide Sharghzadeh argues that he likes seeing AI images in the city’s bodegas, because “it says it’s owned by an immigrant who is totally unplugged from the discourses online…if they know the AI is bad, that means they’re too American.” He goes on to argue that corporate-owned restaurants are attuned to the AI backlash, but immigrant-owned ones are just doing what they can to get by. Other folks I spoke to similarly brought up sympathy for restaurants using AI, saying we shouldn’t blame them for doing what they must to compete (sure). It’s not hard to imagine AI slop aesthetics becoming synonymous with a kind of lowbrow authenticity, that the shitty images, just like the bad photographs that came before them, somehow make the place more real (even though I have seen three-starred restaurants and upscale cocktail bars use them just the same).
As Berlin emphasizes, generative AI is built off the work of countless uncredited, unpaid artists. The problem with generative AI art isn’t that the diner posted a video with misspelled menu items, or nachos that look like sea urchins. It’s the unpaid artists. It’s the stolen work. It’s the data centers that governments and corporations prioritize as communities go without power. It’s that corporations have spent trillions of dollars on this technology (or many technologies, as “AI” is basically a marketing buzzword at this point, which is a whole other issue) who are shoving it into every part of daily life, whether we want it there or not. All of that was built on the backs of actual artists and workers who never consented to be part of this experiment. That would still be true even if the sandwich looked like a sandwich.
Ravenous promises to never use generative AI in the creation of our stories or art. Subscribe today for more human-powered journalism!
