ChatGPT’s AI-generated headshots are all over Twitter and Threads, starting fights and creating confusion. Nobody knows if the poster is engagement-baiting (uploading a headshot that looks exactly like them just to stir up comments) or if they genuinely can’t see the resemblance themselves.
And that’s what makes this trend so fascinating: everyone else seems to see it, while the person in the photo swears up and down it’s not them.
Here’s how it works: You upload a few casual selfies, ChatGPT does its “magic”, and suddenly you have a professional-looking business headshot.
Sounds great, right? Except… people keep staring at these headshots and saying things like:
- “These are NOT the same people. My bff said the pictures look like they could be my cousin.”
- “I’m very confused lol who is this??”
- “Nah, hold on, mine really does not look like me– I know everyone claims this, and they are wrong, but this one is simply noooot me.”
- “There is at least a decade age difference between these two people.”
- “Mine did me dirty. Where’s my butt chin?”
Meanwhile, everyone else is scrolling by and thinking:
- “I thought they was trolling like girl that’s literally YOU in that photo.”
- “What really blows my mind is that the ChatGPT headshot looks better than the vast majority of photos they uploaded.”
- “The photos look exactly like them!!! Just slightly better looking. It’s definitely weird.”
So, why the disconnect? Why does everyone else recognize us when we can’t recognize ourselves?
Let’s unpack this puzzle with science, psychology, and a sprinkle of internet chaos.
Disclaimer: We’re not here to gaslight you. There are a few exceptions to this article where the photos completely miss the mark, just not as many as people think.
There’s a massive difference between selfies and headshots.
Professional photographers are trained to understand lighting, posing, and their subject’s best angles. But a lot of technical gear also comes into play, like camera lenses.
If you take a photo of the same person with a 24mm, 50mm, and 100mm lens, their face will look dramatically different in each shot. Selfies are usually taken with wide-angle smartphone lenses (24-40mm), which can elongate or distort facial features.
Gif sourced from Photographer Dan Vojtech.
On the other hand, professional headshots are typically shot with an 85-100mm lens– a focal length known for producing the most true-to-life facial proportions.
That’s why so many people say their AI headshot “added 10-15 pounds” to their face. It didn’t. It just used realistic camera proportions instead of the slimming, exaggerated angle you’re used to in selfies. In fact, selfies make your nose look 30% bigger than IRL due to lens distortion. Crazy, huh?
Lighting is the silent game-changer.
Lighting can completely transform how we see someone.
Harsh, uneven lighting (think: front-facing bathroom selfies) can cast unflattering shadows or exaggerate wrinkles and blemishes. In contrast, professional headshots use soft, diffused light to highlight natural contours and smooth over distractions. Photographers use a soft-box that diffuses light as evenly as possible over the subject to make everything more balanced.
Here’s a visual example:
Hence, the AI headshot doesn’t always change your features; it’s just simulating a better-lit, unedited, flipped version of your face.
Your brain versus everyone else’s.
You don’t actually know what you look like.
That’s not an insult. It’s science. According to the mere exposure effect, we become attached to the version of ourselves we see most often—which is usually our mirror image. That reversed version becomes our “true self” in our mind, even if it’s not how others see us.
So much so that, in a classic experiment, people preferred images of their face reversed (mirror-style) over how they look to others.
This disconnect intensifies when the photo includes subtle adjustments like corrected symmetry, even lighting, or more accurate proportions, all common in professional headshots. These changes aren’t drastic to others, but they feel jarring to us. So, when we see a realistic headshot (not mirrored, not filtered), our brains go: “Nope. Not me.”
View on Threads
We like what we know, even to a fault.
It’s been psychologically proven that the more we’re exposed to something, the more we tend to like it. It’s why songs grow on us, why we end up liking certain brand logos, and yes, even how we feel about our face.
We’re used to seeing ourselves in low-quality selfies, often with specific angles or expressions that have become our “default face.”
Interestingly enough, studies show people rate more familiar-looking versions of themselves as more attractive—even if those aren’t the most accurate, which explains why onlookers say things like:
- “What really blows my mind is that the ChatGPT headshot looks better than the vast majority of photos they uploaded.”
- “The photos look exactly like them!!! Just slightly better looking. It’s definitely weird.”
Long story short, familiarity rewires how we perceive ourselves.
The closer someone is to you, the more they see what you see.
Remember the earlier comment, “My bff said the pictures look like they could be my cousin”? It has backing; it’s not just in your head.
Here’s the thing: your friends, family, or partner might agree with you that the AI headshot doesn’t quite nail your vibe. That’s because they know your micro-expressions, how you smile when you’re nervous, how your eyes squint just slightly when you’re amused, that micro-frown you don’t even notice you do.
That’s why your close circle might say the photo looks “off.” They compare it to a mental catalog of hundreds of live, dynamic moments. But strangers see the version of you that your AI headshot emulates—a well-lit, balanced, proportional version that may not include every one of your quirks but still is you.
So, who’s right?
The Photofeeler team tests proximity bias.
We tested this at Photofeeler using AI headshots of team members and strangers. Here’s a before and after headshot of our marketing lead:
What do you think? Does it look like her?
She says no, her husband says no, and the rest of the Photofeeler team (whom she sees frequently) say no. But if you think it’s spot on, that also checks out.
Next, we tried with this random woman:
Everyone agrees that she looks like her AI headshot, but none of us know her.
Now, back to a familiar face with another Photofeeler team member:
We don’t see it whatsoever. Maybe others do?
And finally, back to a complete stranger:
For us, it passes.
If the man in the first photo showed up for an interview after us only seeing the second photo, we’d still run with it. Still, this one is slightly less convincing because it lacks human-like details that would separate him from any other 30-year-old man with short hair, chisled facial features, brown eyes, and glasses.
So, where’s the defining yes/no line?
The ultimate litmus test is the criminal sketch test.
Let’s put it this way: if you committed a crime and this AI photo was shown to a witness, would they recognize you?
If the answer is yes, then sorry… that’s you.
View on Threads
AI got it close enough to real life that strangers don’t see an immediate difference unless they break down the images, characteristic-by-characteristic. And there’s a good reason for the missing minor details…
Final Thoughts
Even if a polished AI photo looks “off,” it doesn’t always mean it is. Others likely still recognize you, even if you don’t.
Still curious how others perceive you? That’s what Photofeeler is for. Get honest, unbiased feedback from people who only know your real image, not your mirror one.
