Through the Human Lens: The Limits of AI in Food Photography.
- Hermes Jimenez
- May 6
- 5 min read
By Hermes Jimenez
As someone who’s very familiar with AI, I’ve had a front-row seat to what this technology can do. I use it regularly in my business. I collaborate with it. I experiment with it. I even have entire conversations with it about how we can create stronger visuals for my clients, how to strategize better and how I can be a better marketer overall. So this isn’t a “AI is going to take your jobs!” post. I’m not afraid of AI, and I don’t think you should be either.
As we work with local restaurants and look forward to new trends, new technology and ideas, we fully understand the importance of AI and how that will impact almost every industry. Having said this, I’ve also learned something really important: no matter how advanced it gets, AI still can’t capture certain things. Especially in photography. Especially in food and really in anything that captures the "soul" of the moment.
It can make things look amazing. It can help me bring ideas to life faster. It can even mimic the mood of a photo I describe in detail. But what it can’t do, at least not yet, is recreate the imperfections, the instincts, and the raw little moments that come naturally to a human photographer. The stuff that makes a photo feel alive. The SOUL.
To show you what I mean, I ran two simple tests. One where I described a real food photo in detail and asked AI to recreate it without ever seeing the original. And another where I gave it the actual image and asked for about as perfect a copy as it could. The results were impressive but not without cracks. In both cases, AI got very close, in fact, in one it was about 98% identical. But what was missing was the thing that makes photography feel human. The 2% missing. The unpredictability, the slight imperfections, the little reflections or off center cheese shreds that tell you someone was really there. Here is the description I gave. Then on the left the AI generated pic and on the right the actual picture of the food I described.
Okay, so this dish is called cilantro chicken. It's an Italian restaurant. The image is a close-up of a big plate with a bowl in the center and that is where the food is sitting at. We're talking about a spaghetti at the bottom coated with a cilantro pesto, not heavily, just lightly coated. And then on top is a thin sliced chicken breast with more of the light pesto sauce that looks very glistening. It looks coated. It looks like a vinaigrette dressing in terms of how it is coated on top. It's got some shredded parmesan cheese and these are very thin shreds of parmesan cheese and it's got a parsley leaf for decoration that is two very small branches of parsley. The piece of chicken is long and it covers the entire spaghetti. You could see the spaghetti coming out from the bottom but the chicken is thin and it's long so it doesn't cover it fully because its not the width of the pasta. The plate is white and you could see the reflections of a light. You could just see white around the plate which is a reflection of the light and you could also see small reflections of light on the coating of the chicken which like I said looks very much like a vinaigrette . You could see small pieces of cilantro from the pesto that you can tell were blended when they created the pesto so very very small tiny specks of cilantro. Create this dish in a way that is a close up shot to where you don't see a background. The big white plate covers the whole screen.
Here’s what AI gave me based on that description—and to be honest, it was pretty good. The plating was close. The overall vibe felt right. But the chicken? Plump. Glossy. Almost too perfect. And while it looked “real,” it didn’t feel real on the pasta. It was missing the messy, instinctual details... the subtle light reflections, the tiny cheese strands, the way the pesto clung to the chicken like vinaigrette. It understood the dish… but it didn’t experience it. The realness is gone.

For the second test, I gave AI the actual photo. A simple plate of spaghetti and meatballs. White plate, red sauce, two meatballs on top, a spoon on the side, and a small sprinkle of parsley. Nothing fancy. Just real food, plated by a human hand in a real restaurant.

This time, AI nearly nailed it. The colors were accurate, the spoon was in the right place, and the composition was nearly identical. AI in food photography is a very real thing. But even with all of that, something was still missing. The sauce looked too perfect, the noodles a little too neat. The cheese didn’t stray outside the lines. What was missing was the randomness, the little messiness that comes with a photo taken by a person, and cooked by a Chef in a moment, without overthinking it. Some glossiness that comes naturally through the oils and liquids of the sauce were missing. Random cheese specks that stuck to the pasta that AI either thinks are not needed or felt it was better to not include, or maybe just missed it overall, were missing. The human element was gone, despite having a near perfect, and some may argue a better interpretation of the picture.
There’s no doubt that AI is doing amazing things. In my line of work, it helps me tremendously. It’s saved me hours of time, pushed my creativity, and given me new tools to work with that didn’t exist just a few years ago. I believe AI plays a major role in our future. But as powerful as it is, I’ve also run into some frustrating loops. Sometimes it misunderstands a prompt no matter how clearly I word it. Sometimes it adds elements from past conversations I never asked for. Other times, it recreates something with such perfection that it forgets the unintentional beauty of imperfection.... Or life. That sauce isn’t always centered, cheese strands don’t always cooperate, and lighting doesn’t always hit the same way twice. And maybe that’s the one thing AI will never truly master:
The unpredictability of the human hand... Or humans in general.
So no, this isn’t an “AI is bad” post. This is a “here’s what it’s great at, here’s where it still falls short" AI is the future, and you may read or hear that "it is coming for you" or your jobs, but until we are in Terminator 2 where the machines are experiencing life and doing things we do daily, there will always be a missing element. The Experience. AI can do magical things and yes, it can even write beautifully. But it still depends on us to feed it experience, to tell it what to see. It can recreate, but it can’t feel. That’s why I’ll always use AI in my process… but I’ll never stop trusting my eye or my instinct.
Comments