Podcast Explores Intelligence Beyond Benchmarks And Limits Of Scaling

RSRavid Shwartz Ziv@ziv_ravid

New episode of The Information Bottleneck is out 🎙️ We talk with Mengye Ren (@mengyer, NYU) about what intelligence actually means once you step outside a benchmark, and why scaling a single centralized model isn't the whole story. We get into why intelligence has to be defined in open environments and what that does to how we measure progress. We talked about creativity and the problems there, and why we haven't understood creativity yet and we're already ready to give it away. We also talk about smaller models that actually accumulate experience instead of re-deriving everything from scratch, continual learning and embodied grounding, why frontier labs skip explicit latents, JEPA and world models, and whether ML education still matters when AI writes the experiments.