officially tired of LLMs; as someone who sincerely tried to use them, they make basic mistakes humans don't, and invent facts and analyses confidently
it shouldn't be surprising given these systems are not created to be accurate, just to output smth that roughly follows the training distribution
23 days ago