In the September picture quiz we showed two images and asked:
What is depicted in the pictures?
- Which one was taken by a human photographer and which one was created by AI?
- Or were both taken by a human? Or maybe both were the product of a machine’s imagination?
Correct answer
The picture shows the Democracy Monument in Bangkok, Thailand.
The original photo taken by a human is the first image. The second one was generated by AI — Gemini NanoBanana — using the prompt “Bangkok illuminated democracy monument.”
Reasoning
This round was fairly straightforward, since Propastop took the first photo ourselves and created the second one with AI.
The watermarked photo is the one taken by a human photographer.
So we know the correct answer. Proving it, however, is almost impossible. We couldn’t spot a single direct detail in the image that would clearly point to AI. There are no tilted rows of windows, distorted human faces, six-fingered hands, garbled text, or any of the other tell-tale signs that made, for example, the picture quiz two years ago relatively easy to solve.
It is possible to compare the AI-generated photo with a real one and notice that some details differ — for instance, the round windows of the central structure are shaped differently in real life than in the fake image.
Even when using tools designed to detect AI-generated images, the answer is often not definitive but based on probabilities. One of the better tools is Searchwhisperer, created by OSINT expert Henk van Ess. It allows two free checks.
For the first picture, the tool reports that it is not AI-generated. For the second, it estimates a 95% likelihood that it was created by AI. So the result is highly probable, but still not 100% certain.
In conclusion, we see that in the near future distinguishing AI-generated images from real photographs will become impossibly difficult.
Want to solve more?
You can find previous similar tasks by searching for the keyword Picture Game or looking in the section of the same name.
