• mriswith@lemmy.world
        link
        fedilink
        English
        arrow-up
        5
        arrow-down
        13
        ·
        edit-2
        2 days ago

        That’s because it’s a programmed response.

        There literally is no concept of a joke in their algorithms, it’s an advanced chat bot basically going through a list of responses and comparing them similar questions it as analyzed, and spits out the answer that is likely to match.

        Take a giant spreadsheet with two columns. First one is questions, and the second is responses. Compare queries against first column. When a close match is found, prints out what is in the second column. That’s an extreme simplification of what ChatGPT does. It was never programmer to know or understand what is in either column.

        • jsomae@lemmy.ml
          link
          fedilink
          arrow-up
          23
          arrow-down
          2
          ·
          2 days ago

          That is absolutely not how LLMs work. “Literally no concept of a joke in their algorithms” is debatably true, but it’s a rather useless remark, because you could equally well say that there is no concept of code in their algorithms, and yet they are able to code. (inb4 somebody comes in and says “not that well!” to which I say: well enough for some uses.)