my biggest pet peeve around LLMs is when people (usually those invested in its success) call it βintelligentβ. it definitionally, how it functions on a base level, is not intelligent. the way LLMs are built, it can never hit real intelligence. itβs just predictive