This was a great read, thanks! Trying not to get too metaphysical here, but the effectiveness of multiple personas makes me wonder how _we_ actually reason through problems "under the hood". Do you think studying LLMs can be or will be fruitful for better understanding humans?