This wasn't good at all. This 4 o was a model I was preparing 3 posts for every day, and I couldn't get the same performance with GPT 5. It's disappointing.
If the business model had evolved together with artificial intelligence,
we wouldn’t be talking about a 75% layoff today
we might be talking about a 75% hiring spree instead.
Two photos both generated using Nano Banana or Nano Banana Pro.
One of them (left) kind of looks synthetic. The other one (right)? I’d swear it’s a real photo.
It’s scary how good these are getting.
Honestly, I don’t want to believe the right one is AI-generated.
But it raises an important question:
How do we detect images like this?
What tools, techniques, or even open-source projects can help?
I have no solution this time.
But I’m convinced we need open-source AI detectors now more than ever.
If anyone knows any projects, research, or even homegrown heuristics — please share.
On 29 October 1923, Turkey officially became a republic. The shift from empire to democracy wasn't just political, it redefined identity, sovereignty, and the modern Middle East.
Anthropic has shared their documentation in a `.txt` format that large language models (LLMs) can understand, such as [https://docs.anthropic.com/llms-full.txt](https://docs.anthropic.com/llms-full.txt).
Is there an existing standard directory structure or format for `llm.txt` files?
Are there any known resources or guidelines about what rules or structure such files should follow and how they should be prepared?