people saying LLMs are just fancy chatbots are severely underestimating the complexity of human language understanding