Yann LeCun has repeatedly argued that today’s Large Language Models (LLMs) are a dead end as a path to human-level or general AI