Getting honest feedback on your photos is harder than it sounds. Posting to Instagram gets you hearts, not critique. Photography forums can be brutal in ways that aren’t useful. And unless you’re in a camera club or paying for a workshop, real constructive feedback is surprisingly hard to come by on a regular basis. I’ve felt that gap myself. I shoot constantly, I review gear constantly, and somewhere in the middle of all that I realized I was getting very little outside perspective on whether the images I was making were actually any good.

So when I came across this Sean Tucker video where he feeds his own street photography into an AI critique tool and documents the results, I watched the whole thing twice.

What Tucker Is Actually Testing Here

The tool he’s using is called Critiqly, built by a developer named Karl. It’s a web app designed specifically to give structured photographic feedback using AI. Tucker isn’t using it as a parlor trick or a “look how weird AI is” experiment. He’s genuinely trying to answer a practical question: can this replace or supplement the feedback loop that most photographers are missing?

He runs two of his street photos through it. The first is a strong image he’s fairly confident about. The second is one he’s less sure of, more compositionally complex, where even he isn’t certain the photo is working. That second choice is the smart one, because it’s where the test gets interesting.

What the AI Got Right (and How It Said It)

For the first image, the AI’s critique was detailed and largely accurate. It identified compositional strengths, noted how the light was functioning in the frame, and flagged a specific area of the image that risked pulling attention away from the subject. Tucker’s reaction wasn’t “wow, AI is magic.” It was closer to “yeah, that’s roughly what a thoughtful human reviewer would say.” The feedback was structured around things like subject isolation, tonal balance, and visual hierarchy, the same framework you’d get from a decent photo editor or workshop instructor.

The more revealing test was the second image. The AI gave feedback that was technically coherent but somewhat generic. It assessed the frame using the same structural lens, but Tucker pushed back in the video on whether the AI understood what he was trying to do emotionally or narratively with the image. That’s a real limitation. The tool can evaluate whether a photo follows compositional logic. It struggles to evaluate whether a photo is doing something intentional that breaks that logic for a reason.

The Warnings Tucker Gives That Are Worth Taking Seriously

Tucker spends a significant portion of the video, the section starting around the 16-minute mark, on caveats. This is the part that most recap articles skip, and it’s actually the most useful section.

His core warning is this: AI feedback is calibrated toward consensus aesthetics. It knows what tends to work because it’s been trained on a lot of photography, and it will steer you toward the center of that distribution. If your goal is to make photos that look like good photos, that’s fine. If your goal is to develop a personal vision that deviates from convention in specific ways, the AI might flag your most interesting choices as weaknesses.

He also makes a point about dependency. Using AI critique as a regular feedback loop could gradually nudge your work toward whatever the model rewards, without you realizing it’s happening. That’s not a reason to avoid the tool. It’s a reason to use it with your eyes open.

Where I’d Push This Further

Here’s my honest extension of what Tucker demonstrates: this kind of AI critique tool is probably most valuable at the early and middle stages of your development, not at the advanced stage. When you’re still building your eye and don’t yet have a clear sense of why a photo isn’t working, having a structured breakdown of compositional problems is genuinely useful. It’s better than silence, and it’s better than vague encouragement.

But I’d use it the way I use gear tests. Run your images through it, read the output, then ask yourself whether the feedback is pointing to something you already sensed was off. If the AI flags something and your gut reaction is “yes, I knew that,” that’s useful confirmation. If the AI flags something and your gut says “no, that’s intentional,” that’s useful resistance. The tool is most powerful as a prompt for your own thinking, not as a verdict.

Tucker’s video also reminded me of something I see in gear discussions constantly: people want an external authority to validate decisions they’re already making. The AI critique tool can become the photography equivalent of spec-sheet chasing. The specs look great. Whether the image actually moves someone is a different question.

The One Thing Worth Keeping

The real value of a tool like Critiqly isn’t that the AI has better taste than you. It’s that getting any structured feedback regularly is better than getting none. Most photographers, including serious ones, are working in a vacuum most of the time. A flawed feedback loop beats no feedback loop.

Watch Tucker’s full video for the actual image breakdowns and his side-by-side commentary on the AI outputs. Seeing how he responds in real time to the critique, where he agrees, where he pushes back, is the part that makes this more than a product demo.