Fascinating interview with Yann LeCun — one of the godfathers of modern AI — arguing that today’s obsession with LLMs/chatbots may ultimately be a dead end for achieving true human-level intelligence.
His core argument: language models are useful, but they don’t actually understand the physical world, reason about consequences, or build real “common sense.” Instead, he believes the future belongs to “world models” trained on video, sensor, and real-world data that can learn how reality itself …
