An example of the weakness of ChatGPT as an answer engine is that I tested it by asking two opposite questions.
1. Why has bitcoin outperformed ethereum this year?
2. Why has ethereum outperformed bitcoin this year?

For (2) instead of saying it hasn’t, it gave me a plausible set of answers. Plausible yet wrong. Trust but verify.

@carnage4life you do realize “trust but verify” means… “don’t trust”
@carnage4life this is what makes gen AI so bad - it *looks* right which means increased work to validate the information. This is why when it hallucinates symptoms for a patient in health care use case or precedent court cases for drafting a legal brief it is so much worse than using nothing
@carnage4life Did you ask this with search enabled? It has no recent knowledge otherwise