Yann LeCun LLM superficial understanding highlights the limits of large language models, questioning whether fluency and scale translate into real intelligence.