You are not fundamentally wrong, but there are degrees. LLMs can do some symbolic manipulation and a few reasoning steps. Maybe they are just the result of patt

You are not fundamentally wrong, but there are degrees. LLMs can do some symbolic manipulation and a few reasoning steps. Maybe they are just the result of pattern matching, but I have a hunch that most of the time humans do the same. We fake our understanding as much as we can, we use all shortcuts.