LeCun's recent criticism of LLMs is quite harsh; are world models the future?

In an unsupervised learning podcast, Yang Leikun puts forward an unconventional view on the limitations of large language models, discusses a robotics technology roadmap, explains why he left Meta, and differs from Hinton and Bengio in his perspective on the Turing Award. He also predicts 2027, introduces the new company AMI and its bet on world models, compares OpenAI and Anthropic to Sun Microsystems, and advises PhD students to stop researching LLMs, sharing insights on AI safety, breakthrough research mechanisms, and Meta FAIR’s gains and losses.