The piece profiles computer scientist Yann LeCun, a pioneer in artificial intelligence who recently left Meta to start his own venture, reflecting on his career, the state of AI research, and what intelligence really means. LeCun argues that intelligence should be understood as the ability to learn and adapt, not just excel on benchmarks, and he’s critical of current large language models for lacking genuine understanding despite impressive performance on tests. His departure from Meta and launch of a new startup are framed around his desire to pursue “world models” and more fundamental research rather than focus on commercial metric chasing, which he suggests has led some industry projects astray. The article offers insight into shifting priorities in AI research and debate over the limits of today’s popular models
