I first started noticing them without looking for them at all: a restaurant menu would appear between photographs from a relative’s holiday and a friend’s graduation dinner, and for a split second the food would look perfectly convincing. I’d scroll back up. I would then look a little longer. Perhaps zoom in a bit.
Suddenly, the noodles would seem glossy beyond belief. The garnish would sit in an oddly improbable place. Then, the texture of the rice would resemble that of Miyazaki’s magical dishes straight out of his Studio Ghibli movies*.*
Once I began noticing those details, I started seeing the same visual vocabulary everywhere.
Different restaurants and dishes morph into the same polished atmosphere, the same calculated abundance, and the same sense that the image had been assembled rather than photographed.
The Promise in the Performance
It is easy to understand the irritation, and I get it.
There has been a growing discomfort around the use of AI in food and beverage marketing, particularly as businesses begin relying on generated imagery for posters, menus, and social media campaigns. Much of the criticism focuses on how artificial the food looks, or how repetitive these materials have become.
But when I think about why these images bother me, I am not sure that the technology itself is the thing I object to.
I think what bothers me is the promise.
We Have Always Staged Food
Anyone who’s hung out with friends at some fancy restaurant knows that food photography has always involved a certain amount of theatrics.
We arrange a dish before photographing it. We move the plate toward the light, remove the less attractive pieces, and consider which angle makes the food look most inviting.
A photograph has never been a completely neutral record of what arrives at the table. It is a carefully constructed invitation to imagine the experience of eating.
It is an interpretation of the food as much as it is a record of it.
AI Just Expanded the Set
AI does not fundamentally change that relationship. AI simply gives that interpretation a much wider imagination.
There is something wonderfully creative about that possibility, particularly when the intention is to explore an idea rather than pretend that something imaginary is sitting in a restaurant kitchen.
The difficulty begins when an image is presented as a representation of a dish that a customer is actually going to receive, particularly when the generated version bears little resemblance to the food itself.
At that point, the frustration surrounding AI begins to look less like a rejection of new technology and more like a very old frustration with misleading advertising.
The Gap Between Acts
If a customer sees a photograph before ordering, that image is already shaping an expectation. They are imagining the portion, the texture, the colour, and perhaps even the experience of eating it.
When the food arrives looking substantially different, the disappointment is not really about the software that made the picture.
It comes from the distance between the picture and the plate.
For that reason, I find myself thinking that AI food photography should be held to a higher standard rather than treated as something that gets a free pass simply because it vows convenience.
If anything, this convenience makes thoughtful direction even more important. I’ve written before about how lighting, composition, and styling can still guide the way we create food photography with AI, and I keep coming back to those principles because they remind us that the technology does not replace the eye behind the image.
Someone has to recognise when an image has stopped belonging to the food and started belonging only to the capabilities of the tool.
That responsibility still belongs to the person directing it.
