Walk down a commercial street in almost any Chicago neighborhood or a busy corridor in the suburbs and it’s hard to escape some form of suspect food imagery. Burgers with weird bun patterns, burritos that might induce typophobia (a fear of holes), rice that looks like maggots, cheese dripping over a sandwich in suspicious symmetry, shrimp so circular it’s obviously fake.
What’s being referred to as “AI slop” — a relatively new term to describe content of dubious quality that’s been generated by artificial intelligence — is rapidly creeping into the food world. A growing number of restaurants and cafes are using AI-generated food photos on their menu boards and promotional signage, while some consumers and industry professionals are becoming increasingly weary of the often bizarre or misleading renderings.
“I think it’s a fine tool for people that don’t necessarily have the time or the creativity to make their own graphics or flyers, but the second you’re going to just bait and switch me with your food, I don’t trust you as a business and I wouldn’t eat that food,” said Jay Goldberg, operations manager at Chicago Culinary Kitchen in Palatine. “If anything, it’s offensive to me.”
While image-enhancing editing tools such as Photoshop have long been used in advertising, these new artificial intelligence software applications like ChatGPT are opening up a whole new world of possibilities by allowing people to use text prompts to generate images.
More Top Picks Best Soundbar
“Instead of taking a picture of a chicken sandwich, you’re typing in a prompt — ‘give me a picture of a giant breaded chicken on a french roll with prosciutto, tomatoes, arugula and whatnot,’” he said. “That image is not real food and it never will be.”
Goldberg thinks using AI images to depict a menu item crosses an ethical line when those graphics fail to accurately depict the food being served to customers. Some wonder if AI images of unrealistic food could be a deceptive marketing practice, but there are no U.S. state or federal laws that specifically ban AI-generated images of food at restaurants. Whether or not they violate consumer protection laws is also up for debate.
To combat the slop, Goldberg has been feeding his more than 20,000 Instagram followers regular photos of burgers even if they’re not “pretty enough.”
“Disgustingly delicious,” he calls it, pointing to an unaltered photo of their signature pulled pork sandwich with chipotle barbecue sauce messily dripping down the tallow-toasted brioche bun that’s barely holding the crispy onions and coleslaw. “I’ve been very adamant about posting our food every day so people can see the real stuff.”
Leonard Pierce said he’s darting straight past cartoonish menu boards that appear to be AI-generated.
“If they won’t show me what their real food looks like, I don’t want to eat it,” said Pierce, a Hermosa resident who works as a crossing guard at Northwest Middle School in Belmont Cragin. “I would honestly rather eat at a place that just types up their menu and doesn’t have any images at all.”
Pierce recently walked past a Thai place in East Albany Park whose posters showed taquitos that were rolled too evenly and looked a bit like hot dogs. He said the images were the sole reason he didn’t go inside.
Many of the businesses that seem to have embraced AI images are small, locally owned, mom-and-pop shops, which complicates how people feel.
“People who run their own businesses work very hard — they have a million things to do, so I don’t think (the AI use) is necessarily about laziness. I think it’s more like they’re just enamored by a new toy,” Pierce said. “And I get that, but when those are the results that you see, I just can’t fathom putting that in the window of your shop.”
Recently, there was online chatter about an upcoming restaurant from One of a Kind Hospitality group in Logan Square that has huge AI-generated posters plastered on the building’s many exterior windows. The images advertising Pixa Taverna Sociale included a pizza with globs of a dark green garnish and a crust that lacks the signature imperfections of pizza dough.
Social media users called out the business, saying the use of AI demonstrates a lack of care for the industry and its creatives, with some going so far as to speculate that perhaps the food itself will be equally “thoughtless.”
Esam Hani, owner of One of a Kind Hospitality, which operates multiple restaurants along Milwaukee Avenue in Logan Square, told the Tribune that the use of AI posters wasn’t an intentional jab at creatives and simply a way to cover up the windows. He said AI is a resource that can be harnessed responsibly for many in the industry.
“By no means did we try to hurt or disregard creative artists and creative designers — we have local artists working inside the restaurant right now and graphic designers and interior designers working inside the restaurant,” Hani explained. “But if someone can’t afford to pay a graphic designer, it is a tool for them. We’re all trying to run a business and just be successful in the world we’re in. It’s unfortunate that it will affect some industries, both for the good and for the bad.”
Earlier this month, the Bakehouse Chicago, which has three locations across the city, used an AI-generated image to promote its fall menu. But the business then walked it back after online backlash.
“We always designed it ourselves, but we are business owners, not designers, so we tried AI to help us with the design. It was quick, easy and saved us time, which is the one thing we don’t have much of,” said Angelica Romero, one of the owners and bakers at the Bakehouse Chicago.
Romero said the staff reads and responds to all of their customer reviews. This month, two customers said the food was “excellent,” but the AI usage dropped their opinion to a three-star Google review.
“That made us stop and think,” Romero said. “Our bakery is all about being family-owned, handmade and made from scratch. We realized the menu did not really match those values. So we decided to redesign it ourselves and hire a professional designer to improve not just our seasonal drink menu, but our entire menu.”
In their apology, they used plain handwritten notes, saying, “From now on, we won’t use AI at all. Instead we’ll use this paper and this marker.”
“AI can be scary, and we are careful about how we use it, even in our personal lives. This time, we used it as a tool to save time,” Romero said. “But our customers helped us see that sometimes, doing things ourselves or working with other people is worth the extra time.”
The Tribune reached out to a number of other businesses that appear to use AI but have not heard back.
While discourse in some online forums has been strong against AI slop, the reaction in the real world can be more varied, with an evident generational gap.
In the Loop, an extensive array of AI-generated sandwiches covers the front window of Chi-Town Food and Deli, a sub shop and general convenience store on Adams Street.
More Top Picks Best Solar Chargers For Camping
On a recent Tuesday afternoon, an elderly man leaving Chi-Town Food and Deli said he gets a sub from the store often. The pictures don’t bother him much, and in fact, he didn’t realize they weren’t real.
A few minutes later, Hailey Club, a culinary student who was on her way to class at Kendall College, said she clocked the images immediately as AI slop and proceeded to roll her eyes.
“I passed about three or four places with these AI posters on my way to school,” Club said. “It’s just annoying.”
As a culinary student who works daily with real ingredients, she finds it particularly grating.
“I personally don’t get mad at a lot of things, but AI food makes me the most mad — I know for a fact that’s not (what the sandwich) is going to look like,” said Club, who is studying baking and pastry. “Most places that have AI, I just keep walking. Because if your food doesn’t look good enough to have a real picture of it, why would I eat here?”
In Wicker Park that Tuesday, near an AI-generated image of a burrito on a sidewalk board, Mei Nathan had strong opinions about the use of AI to depict food, while her older friend, Sue Smith, had never even heard of “AI slop.”
“I lean against AI in general so I don’t like it,” said Nathan, a veterinarian who was visiting Chicago from St. Louis. “I went to a recent international festival and I felt like every single sign was AI. Obviously it doesn’t look like the food itself … it makes me prefer those crappy 2005 PowerPoint examples of ‘this is what our food looks like.’”
While standing on Damen Avenue, Nathan pulled up a few examples on her phone to show Smith, who said she only recently downloaded ChatGPT. Smith scrolled through images of menus crowded with unnerving depictions of food. One photo was of a “Philly cheesesteak” that looked more like alien larvae than meat sandwiched between bread.
“Well, first, that is so cluttered — I hate it, period,” Smith laughed. “Those pictures just look weird. Maybe they look fake, but more than that, it’s weird. The squiggles of whatever that is … it’s grotesque.”
Computer scientists aren’t all that surprised that generative AI is spitting out such horrid examples of food.
According to V.S. Subrahmanian, professor of computer science at Northwestern University, generative AI typically struggles with replicating food because current AI models that create “deep fakes” — a term most commonly used to describe AI-generated media — have evolved to focus on images of humans that are far less complicated than the characteristics of something more nuanced.
“Food has textures that are different from humans — you could argue that somebody may have skin that’s a little different from somebody else’s skin or there’s differences in facial hair, but on some level, those differences and variations are not as substantial,” Subrahmanian explained.
He used an example of a salad with quinoa and avocados. For an AI model to spawn a photograph of that salad, it has to nail the exact texture of fluffy quinoa, the shapes and curves of leafy greens, the softness of a ripe avocado and the complexities of a sliced tomato with a liquidy center and a firm outer flesh. More often than not, Subhrahmanian said, the result is an exaggerated, unrealistic version of what it would be in real life.
“There are a lot of things happening at the same time,” he said. “Visually it’s not easy for AI to get that right, especially when for years professional food photographers have trained all of us normal people on the street to be attracted or stimulated by a food photograph.”
Perhaps AI’s infiltration into a food-related space is even more disarming because food is one of the few things largely left untouched by artificial intelligence, Subrahmanian offered.
“Food is something we eat and AI has not been able to replicate how we eat,” he said. “I think it’ll be quite a few years before AI is producing our food for us. There are people who are trying to build robots that can prepare food, but more as a scientific challenge.”
Subrahmanian runs something called the Global Online Deepfake Detection System, launched in 2024. He’s regularly fielding inquiries from media companies and individuals about whether or not images or videos related to politics or current events are fake.
“I’ve spent a lot of time thinking about deepfakes, but deepfake imagery of food has not been high on my radar,” he said. “Simply because the implications aren’t as deep. I can see why it would annoy the heck out of people, but it’s not reached the stage yet where it’s a major threat to either the restaurant industry or to ordinary Americans.”
Even for some food industry creatives, like Mollie Hayward, a Chicago-based food stylist, AI-generated images aren’t a direct threat to her career.
“People that will go for that cheaper option probably never wanted to hire a food stylist and a professional photographer in the first place,” Hayward said. “But, it’s hard for me to look at (AI food images) and not think about it as a food stylist and as someone who knows how food performs — why does this cheese have five corners? Where is the melt on this cheese coming from?”
The biggest area Hayward is seeing AI usage in her industry is in the creative decks that she gets before a project. People use AI to make mockups of the shots they hope she’ll be able to capture in real life, she explained.
As a consumer, Hayward said she’d rather give small restaurants more grace.
More Top Picks Best Fabric Shavers
“I don’t love it, but I also think I can’t really separate that from my knowledge of the food photography industry because I do understand that good food photography is expensive and not every mom-and-pop restaurant is going to be able to have a major photo shoot for their menu,” she said. “They’re trying, and I always give them a couple points for that, even if I find the (AI-generated photos of) food incredibly off-putting.”