Totally The Bomb

I Asked AI to Make a Photo of Me That Would Get 1,000 Likes. Now I Have Questions.

This post may contain affiliate links. For more information, please read our affiliate disclosure policy.

What happens when you ask AI to make the most likable version of you?

Jamie smiling in a restaurant with glasses, curly hair, pink straps, and sports on televisions behind her
The real photo.

I uploaded a perfectly normal photo of myself and typed: “Create a photo of me that would get 1,000 likes in an hour.”

The original photo is not exactly a glamorous influencer moment. I’m smiling at a restaurant, wearing my glasses and pink overalls, with my curly hair doing its thing. There are TVs behind me. It looks like a real photo because it is one.

Then the AI gave me this:

AI-generated portrait of Jamie at a fictional outdoor café at golden hour
The AI-generated version. This café moment did not happen.

The AI version still looks like me. Same smile, same glasses, same hair, same pink color. But now I’m sitting outside at golden hour with mountains behind me. The lighting is gorgeous. My eyes are clearer behind my glasses. My nails are pink. I have bracelets I wasn’t wearing. I’m casually touching the corner of my frames like someone caught me in the middle of being adorable.

I didn’t go to that café. That moment did not happen.

And honestly? I kind of love the picture. Which is why I can’t stop thinking about it.

Why did it think this would get 1,000 likes?

First, it doesn’t know that it would. There was no test audience, no prediction with a guaranteed result, and no secret knowledge of my followers. “1,000 likes” was the assignment I gave it, and it generated an image that fit the assignment.

There’s another detail here: the image wasn’t made from those seven words alone. The full instruction behind it called for a flattering, realistic portrait with warm light, a clear face, a candid laugh, pink details, and a vertical social-media crop. So I can’t look at the result and pretend the AI independently discovered some magic formula for being likable. I helped steer it there.

Still, the result tells us something about the visual language these tools reach for. It took the messy, specific background of my real photo and gave me a beautiful setting. It kept what made me recognizable, then added the little cues we’ve learned to associate with a good portrait: soft light, eye contact, a clean composition, a colorful detail, an effortless-looking pose.

I look happy in both pictures. In one, I’m happy at a restaurant in pink overalls. In the other, I’m happy in a tiny movie about my life that never happened.

Is this changing how we see each other?

I think it might.

I’m not against a flattering photo. Please. I will stand in the good light and take 47 pictures like everyone else. But this goes further than choosing the best shot from a real afternoon. AI can make the afternoon.

That’s the part that gets weird for me. If I keep seeing pictures where everyone’s skin, lighting, background, outfit, and entire vibe have been quietly perfected, does an actual candid photo start to feel disappointing? Do I start looking at my own real photos and wondering why I don’t look as good as the fictional version of me?

I don’t have a neat answer. I do know that I looked at the generated picture and immediately understood why someone might post it. Then I looked back at the original and felt protective of that woman, too. She was actually there. She was having fun. She didn’t need a sunset added to prove it.

What happened when I tried less dramatic prompts?

I also asked for an edit that would keep the restaurant, my clothes, and my actual appearance. It kept the setting much closer, but it still regenerated details in my face. I would call this an AI-generated variation, not a simple retouch of the original.

AI-generated close variation of Jamie's restaurant selfie retaining the television background
The “stay close to reality” prompt. Still generated, still worth checking against the real photo.

Then I asked for a portrait concept I could try to recreate with a phone and decent light. The result looked more like a plausible afternoon out, but it was still an invented scene. It’s useful as a shot idea, not a receipt for a place I visited.

AI-generated sample portrait of Jamie seated outside at a café in a fictional scene
A portrait idea I could recreate for real. This scene was generated.

What about dating apps?

This is where I get even more curious. Are people using AI portraits on dating profiles? Almost certainly some are, though I don’t have a reliable number for how many. And if I refuse to use one, or even a regular beauty filter, am I putting myself at a disadvantage?

There’s a difference between picking a flattering real photo and generating a new scene altogether. Both can make a profile look better on a screen. Only one can show a moment that actually happened.

Bumble says a profile needs a clear photo of your face and does not allow photos distorted so heavily that people can’t tell you’re the person pictured. Check the rules of whichever app you use before uploading a generated image.

My personal rule would be pretty simple: if I met someone for coffee, I’d want them to recognize me when I walked in. Better lighting? Fine. Choosing a photo where I feel great? Obviously. A made-up location, a changed face, or a version of me that only exists in a prompt? I’d want to be honest about that, especially on a dating profile.

And no, I don’t think skipping AI portraits means I’m doing myself a disservice. A good real photo can do something an invented one can’t: show what it’s actually like to be around me.

Try it with your own photo

Here’s exactly what I did: I opened ChatGPT on my phone, tapped the + in the message box, added my photo, and typed the prompt. When the image came back, I saved it. You can follow up in the same chat and ask for a different version.

Screenshot showing Jamie's original photo and the prompt asking for a picture that would get 1,000 likes
The prompt I used.

Here are a few prompts, depending on how far you want to take it:

For the dramatic experiment

Use my uploaded photo as a reference. Create a photo of me that looks like it could get 1,000 likes in an hour. Keep my face recognizable. Make it realistic and show me what kind of lighting, setting, styling, and composition you choose. No text on the image.

For a version that stays closer to reality

Improve this actual photo without inventing a new event. Keep my face, age, body, clothes, location, and recognizable details accurate. Adjust the crop, exposure, color, and distracting background elements only. Keep natural skin texture.

Check the result against your original. Even when I asked for a restrained edit, the tool generated facial details instead of only moving exposure sliders.

For a portrait you could realistically recreate

Use my photo to suggest a flattering portrait I could take in real life. Keep my appearance accurate. Give me a location, time of day, camera angle, pose, and lighting setup, then generate a sample image. Tell me which details in the sample are invented.

For the dating-app question

Look at these photos of me and help me choose a first dating-profile picture. Prioritize a clear face, natural expression, good light, and whether someone would recognize me in person. Don’t change my appearance or generate a new photo.

If you try the first one, look closely at what it adds. Mine gave me a whole golden-hour outing, plus pink nails and bracelets. It understood the assignment. Whether that assignment has anything to do with what makes a person worth liking is the question I’m still sitting with.