This is an interesting article on ARS Technica which summarises some of the problems of the AI GPT model. It's making up information and presenting it as facts. The WSJ reported this previously [webmasterworld.com...] and it called it out for making up lies.

Clearly, this is a problem for the model and it certainly cannot be relied upon at the moment. Should there be a fact checker for AI GPT models, and what's the efficacy of that fact-checking data model?

I'm sure, like most here, we've tried and tested the systems out there, and have found them to be a useful additional tool. I had it generate some information for me, keeping it simple, and avoiding "people-retated" information, and it was reasonably helpful. I had it generate short scripts, and it seemed fine. That's just my experience. Many others are reporting misinformation about themselves, which is interesting in its own right.

[arstechnica.com...]