Originally Posted by
Quossum
Be careful, because sometimes AI “hallucinates” and gives incorrect information or just makes things up.
I can't agree with this more.
I've looked at its results for basic financial information, Medicare plan information, and current sewing machine information. None of the results are correct, and several of them (particularly related to finance/health insurance) are so wrong they would cause very, very bad decisions to be made.
"AI" is not intelligence. It's learned language modeling - basically scraping existing content (whether correct or not) that its algorithm determines most closely fits the query, in some cases simply by the frequency something was posted. I had to laugh when one of my own posts from one forum was used as a ChatGPT result in another forum as a supposed "answer", but if you actually read it, the quote wasn't actually an answer but my original question in the other forum.