@DrewKadel I saw an article lamenting that the LLM couldn't reliably explain its "reasoning," and it's just such a fundamentally flawed idea about what ChatGPT is doing :(