I use Gemini almost daily, and occasionally Grok, and I've noticed an increase of hallucinations, especially with Gemini.

For some reason, whenever AI makes a mistake (=often), people say it hallucinated. That's the standard term for this, somehow.

What I've noticed is an increase of factual inaccuracies.

I often ask AI to analyse what I wrote for grammar and accuracy and, more than once a day, AI corrects me on things I know for a fact to be correct.

Imagine writing "the capital of France is Paris", and then AI corrects you and says it's Lyon. Obviously, this is a made-up example, because the errors Gemini made are minor and related to YouTube content I've watched, but it's still a little worrying.

I'm not entirely sure why that's happening.

We know that AI is unprofitable. We know that it costs way too much to run but doesn't generate money. We know that people use it for all kinds of purposes, including as a virtual s*x partner and other things like that.

Unfortunately, the problem is that a lot of people treat AI as gospel. Even though there's a disclaimer at the bottom of every AI chatbot ("Gemini/Grok/ChatGPT can make mistakes" etc) no one seems to take that seriously enough.

The problem is scale. If AI can hallucinate for a small thing, it can do that for a big thing as well.

That's scary.