Intelligence in an Open World - with Mengye Ren (NYU)

Mengye Ren of NYU's Center for Data Science argues that intelligence must be defined and measured in open environments rather than closed benchmarks, and that scaling one centralised model is not the whole story. The conversation examines creativity in current models, which sample bottom-up without an internal loop of consideration, and what a next paradigm needs: continual learning, memory, embodied grounding and smaller models that accumulate experience instead of re-deriving it.