Can you tell which images are AI-generated?

Can you tell which images are AI-generated?slop-sense.labtoagi.com

Trust your eyes and build a streak in Sprint or Survival play.

112 points•hckr78•11 days ago•85 comments•

#generated #images #ai #ai generated #eyes #streak #trust #beat

85 comments

Jowsey11 days ago

I hadn't been exposed to the new GPT-Image-2.5 model before this, so didn't have any "tells" to go off of, and found it almost impossible to begin with, but it seems the brain is pretty good at adapting for visual stuff! After 2 rounds, I was able to build up a 16-image combo, answering most within a second or two.

I think the "tell" I ended up with is that almost all of the AI images tend to center around a very obvious main "subject" (a flower, a bike, a wrench, etc) – presumably an inherent artefact of generating from a prompt – with everything else around the subject having a very strange depth to it. The depth, focus, and bokeh around the main subject just never look quite right. If you look at an image and find it has a very strong central subject with a little too much depth separation than expected, there's a good chance it's AI.

I recognise that this method of finding a consistent "tell" will probably not hold for future models, unfortunately. Everything else looks almost perfect at this point.

Retr0id11 days ago

It's hard to put into words but the "texture" of GPT-Image-2.5 is just a bit off, like the difference between blue noise and white noise. It's like someone turned up the local contrast just a little too high.

SoylentGreenGPT10 days ago

Exactly! The generated images taste different. Also…if you see a perfectly folded towel, it’s probably AI. At least in this set of images.

thayne11 days ago

There are occasionally more obvious tells as well, if you look closely. Some examples that I saw:

- A boat cast a shadow, but the mooring rope didn't

- A bench leg was out of perspective, and didn't touch the ground

- In a pile of logs,the ends were sharpened in an unnatural way[0]

- In several there seems to be an inconsistent contrast between something that looks more dirty, worn, weathered, etc. while the surroundings are more clean and smooth.

Some also feel off in a way that I can't really articulate.

[0]: I think the relationship to the prompt was interesting here, the prompt actually asked for a stack of wooden fence posts, but they were really thick for fence posts, and looked more like firewood with a sharpened point. Interestingly, the prompt included the phrase "without excessive sharpening" which I think meant increasing the sharpness of the image, but I wonder if the AI interpreted as referring to the sharpness of the fence posts.

wildzzz10 days ago

The fonts it uses for signs and posters. They often don't match the scene. The posters on a corkboard next to school gymnasium don't look like they were made by students or teachers, most likely amateurs at graphic design, they look professionally made. As for signs, a rustic looking sign isn't going to use helvetica-like typeface, it looks entirely out of place. Again, it looks like a professional made it in 2026, not an amateur merchant 50 years ago.

SoylentGreenGPT10 days ago

Exactly, if the words are too easy to read, and clearly not part of a photo used to advertise - then it’s probably AI.

variety867511 days ago

I realized this is only somewhat difficult because of the 10s timer. Most people probably don’t spend more than 10s looking at a picture on social media, excellent project.

hn_throwaway_9911 days ago

I thought that at first, but a comment below enlightened me. These types of games are a great way to get free training data.

variety867511 days ago

Great point, in the privacy disclosure they confirm they're collecting the training data.

> Reality Check records visits, round starts and finishes, image views, answers, response times....records are stored in Cloudflare D1

SmasherEpilepti11 days ago

> 60 seconds to find out.

> 90s round, 10s per image

> Play 60 seconds

This is confusing. What does "90s round" mean if it's a 60 second game? Is it about the 1990s?

muzani11 days ago

"Made with ♥ by Codex"

sph11 days ago

Another pro-AI psyop to show that AI generated content is perfectly fine, created by cunning machines and spread through mindless meat proxies glad to do their bidding.

/s but also not really.

EA-316711 days ago

Slop all the way down, I guess at least if we’re being charitable. If we’re being less charitable this is trying to gamify getting free training data.

fwlr11 days ago

Every generation prompt has this suffixed:

    The scene should feel incidentally noticed during daily life, not arranged for an advertisement. Natural perspective and subtle everyday wear. No cinematic grading, stock-photo staging, deliberate blur, artificial grain, fake low resolution, extreme background blur, exaggerated HDR or oversharpening. No collage, border, caption or added watermark.

At some point during the process of building up this collection of turd-polishing techniques, the thought should’ve occurred that this is actually a lot of effort to get generations confused for reality. But I suspect this process was being carried out by something that couldn’t have an original thought.

tgsovlerkhgsel11 days ago

A troll farm will happily go through that effort then (if it's a high effort part of the campaign rather than just 'flooding the zone') pick the images that are most likely to pass.

sitzkrieg11 days ago

freethink is inefficient, schedule this biomass to the remediation chamber

charlie9011 days ago

Highest I got was 26 in a row. AI seems to love generating "close up" images of a subject. Its like every image has to have an obvious subject (that was in the prompt), and it struggles to make scenes without a clear subject and more chaotic vibe. I notice this a lot with these "AI movies", every single camera shot has an obvious subject that makes them unbearable to watch, they can't just chill out and have a messy scene.

Read the full thread on Hacker News →

85 comments

Related stories