• williams_482@startrek.website
    link
    fedilink
    English
    arrow-up
    27
    arrow-down
    1
    ·
    1 year ago

    You realize that ChatGPT has no concept of “true”, right? It produces output which looks coherent and reasonable and tends to stumble into truthful statements on accident, by virtue of drawing from a dataset of people saying mostly true things. Of course, the bot is equally capable of spouting off outright lies in an equally convincing manner.

    This is a very unreliable way to verify a surprising fact. I strongly recommend against it.