The challenge and danger with widespread use of these tools is that they, with some frequency, continue to just confidently state answers that contain completel

The challenge and danger with widespread use of these tools is that they, with some frequency, continue to just confidently state answers that contain completely wrong information. These algorithms are still terrible at just saying I don’t know or here’s a possible answer but I have low confidence this is correct. They give every answer with a tone of authority despite being completely wrong with non-trivial frequency. A basic search engine search, while much more labor intensive, at least quickly reveals when there’s some inconsistency around a potential topic that allows an inquisitive individual to probe further. These new tools could be super dangerous in that they just spread false information and many folks don’t know or care enough to check if the answers are even right. We’re going to see a lot more train-wrecks like that lawyer that used ChatGPT to do research on his brief and ended up citing a bunch of cases that were made up and didn’t exist.