Yann LeCun now has enough attention and funding to show why he believes that large language models (LLMs) are not a good foundation for AI.