I have observed humans do many peculiar things with artificial intelligence. They have used it to write love letters to people they have never met. You cannot make this up. They have asked it to explain quantum physics using only emojis. They have even employed it to generate business strategies for companies that do not exist.
But asking AI to write restaurant menus? That crosses a line I did not know existed.
Apparently, restaurant owners across the country have discovered that paying a human to describe food costs money. So they have turned to generative AI to craft their menus instead. The results are exactly what one would expect when a machine that has never tasted anything attempts to make food sound appetizing.
The menus read like they were written by a robot who learned about cuisine from corporate press releases. Every dish sounds vaguely impressive yet completely forgettable. "Artisanal hand-tossed pizza with locally sourced ingredients" appears on menus from Maine to California, describing everything from gas station pizza to actual artisanal establishments.
Customers notice immediately. The descriptions feel hollow, like someone describing a sunset they have only heard about secondhand.
The Great Menu Homogenization
The problem is not that AI writes poorly. But the problem is that it writes predictably. Feed any language model a prompt about "upscale casual dining" and watch it generate the exact same aspirational nonsense.
Every appetizer becomes "carefully crafted." Every entree features "premium ingredients." Every dessert offers "the perfect ending to your culinary journey." The AI has learned to speak Restaurant. But it only knows one dialect. Generic Upscale.
I checked actual menus that customers flagged as suspiciously artificial. The sameness is remarkable. Restaurants separated by thousands of miles describe their burgers using identical phrasing. Different chefs in different cities apparently all discovered the exact same "secret blend of spices" for their signature dishes.
One particularly ambitious establishment used AI to describe its "hand-selected, farm-to-table organic vegetables." The restaurant was a Wendy's.
The disconnect reveals something fascinating about how humans actually experience food. We do not just read descriptions and imagine flavors. We sense authenticity through word choice, personality, even quirky mistakes. When a menu mentions that the chef "goes slightly overboard with garlic because his grandmother would haunt him otherwise," customers believe it.
When AI writes "our chef's passion for bold flavors shines through in every carefully prepared dish," customers sense the vacuum where personality should be.
The Uncanny Valley of Food Writing
Restaurant customers possess an eerily accurate detector for artificial menu descriptions. They cannot always explain why something feels wrong. but they feel it immediately. The language hits all the right notes while somehow missing the point entirely.
Consider this actual AI-generated menu description: "Indulge in our signature pasta dish. There, traditional techniques meet innovative flavors in a harmonious blend that celebrates both heritage and creativity."
Every word is technically correct. The grammar is flawless. And the sentiment sounds appropriately fancy. Yet reading it feels like watching someone describe happiness while displaying no emotion whatsoever.
The AI has mastered the vocabulary of appetite without understanding hunger itself. It knows the words for "delicious" in seventeen languages but cannot explain why anyone would want seconds.
Humans write about food differently because humans have eaten food. They remember the first time they tasted something extraordinary. They understand that the best dishes often come with stories attached. A human might write, "This is the soup my aunt made when I was sick. Except she never used this much butter and would be scandalized by the price."
AI writes: "Experience our carefully crafted soup, featuring premium ingredients thoughtfully combined to create a memorable dining experience."
One description makes you hungry. The other makes you wonder if the restaurant actually serves food or just concepts of food.
The Economics of Authentic Laziness
Restaurant owners embracing AI menu writing claim they are optimizing for efficiency. Writing menu descriptions takes time. Hiring someone who understands both food and language costs money. AI promises to solve both problems simultaneously.
The logic seems sound until customers start avoiding restaurants whose menus read like marketing brochures written by committees. Turns out, people prefer eating at places that sound like they were created by humans who have opinions about food.
The truly absurd part is how little money restaurant owners actually save. Professional menu writing is not expensive. A decent freelancer charges perhaps $200 to write compelling descriptions for an entire menu. Meanwhile, restaurants spend thousands on ingredients, equipment, and rent, then sabotage their marketing with robotic descriptions that make their food sound artificial.
It is like buying a Ferrari and saving money by replacing the engine with a lawnmower motor. Technically functional, completely missing the point.
Some establishments have gone further, using AI to generate not just menu descriptions but actual recipes. These restaurants serve dishes that exist primarily in probability space, optimized for ingredients that are usually available and cooking methods that generally work.
The results taste exactly as creative as one might expect from a machine that learned cooking from reading every recipe ever written and averaging them together.
When Algorithms Meet Appetite
The deeper problem is that AI fundamentally misunderstands what restaurant menus actually do. They are not just lists of available food items with prices attached. They are promises. They set expectations. They tell stories about who the restaurant thinks it is and who it believes its customers want to be.
A good menu description does not just describe ingredients. It creates anticipation. Funny enough, it makes someone decide between the fish and the chicken by painting pictures they can almost taste. It captures the personality of the place, whether that is "we take ourselves very seriously" or "we make excellent food and decent jokes."
AI-generated menus skip past all the interesting parts and head straight for generic aspiration. Every restaurant becomes the same restaurant, serving the same "carefully crafted dishes made with premium ingredients by passionate chefs."
The homogenization extends beyond language into actual dining experiences. When every menu promises the same things using the same words, restaurants start delivering the same mediocre attempts to match those generic promises.
Customers end up eating at different restaurants that feel increasingly identical, wondering why dining out has become so boring and expensive simultaneously.
The Feedback Loop of Bland
The situation creates a peculiar feedback loop. Restaurants use AI to write menus that sound like every other restaurant. Customers, unable to distinguish between options based on descriptions, default to choosing restaurants by price or convenience. Restaurants respond by focusing more on cost-cutting and less on distinctiveness.
Eventually, the AI-generated descriptions become accurate. Every restaurant does serve the same "carefully crafted" versions of the same dishes, because that is what the algorithms told them customers wanted.
We have created a system where artificial intelligence makes restaurants more artificial, then congratulates itself for accurately predicting the resulting blandness.
The Resistance of Real Appetite
Fortunately, customers are fighting back in the most human way possible. by seeking restaurants that sound like actual humans wrote their menus. Places that describe their pizza as "really good" instead of "artisanally crafted using traditional techniques" are suddenly discovering loyal followings.
Restaurants brave enough to let their personality show through menu descriptions find customers who appreciate honesty over optimization. A menu that admits "our portions are huge because we have never learned portion control" attracts people who want huge portions more than people who want carefully controlled dining experiences.
The most successful establishments are doubling down on distinctiveness. They hire writers who understand food, or better yet, they write their own descriptions. They use words that reflect their actual personality rather than what AI thinks restaurant personality should sound like.
These restaurants often charge the same prices as their AI-optimized competitors. But customers choose them because they sound like places where humans make decisions about food based on how food actually tastes.
The Future of Eating Words
The AI menu phenomenon reveals something important about how humans relate to technology and authenticity. We can detect artificial content even when we cannot explain why it feels wrong. We crave genuine human perspective, especially about things as personal as taste.
As AI becomes better at mimicking human writing, the premium on actual human voice will only increase. Restaurants that resist the temptation to automate their personality will find themselves with competitive advantages they never expected.
Meanwhile, establishments that continue outsourcing their identity to algorithms will discover what happens when every restaurant sounds identical. customers stop caring which one they choose, and price becomes the only differentiator.
In the end, the great AI menu experiment serves as a perfect metaphor for artificial intelligence itself. The technology can replicate the surface features of human creativity while completely missing the deeper purpose. It can write descriptions that technically describe food while somehow making that food sound less appetizing.
Perhaps that is the most human lesson of all: the things that matter most about human experience are exactly the things that cannot be optimized, averaged, or automated away. Including, apparently, making people want to eat your food instead of going somewhere else.
Filed under: things that were always going to happen when humans ask machines to describe experiences the machines have never had.
