5 ms·
Your claim then seems to be that galkk is just plain lying? That seems uncharitable.
by morelisp 3y ago
Your claim then seems to be that galkk is just plain lying? That seems uncharitable.
- satvikpendem 3y agoLying? No, I never said that, their version of ChatGPT (3.5 which is older) or Copilot (the Codex model it uses is much older) might very well be hallucinating wrong answers, sure. But my claim was that the newer models such as GPT 4 work well for certain classes of problems, so no, it's not "obviously false" that they don't work, which is what you claimed. Does GPT 4 get stuff wrong too sometimes? Sure, I never claimed they're perfect either, but unlike people in this thread, it has worked well generally for me based on my experience, but if you want to discount my experience and only listen to other people who confirm your beliefs, you do you.