Social Commonsense Reasoning with Yejin Choi - #518
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
00:00
Rethinking AI: Beyond Scaling
This chapter discusses insights from a keynote at the Stanford HAI Foundational Models Workshop, challenging the belief that scaling neural networks is the only path to achieving Artificial General Intelligence (AGI). It advocates for improved algorithms and knowledge integration to enhance smaller models and delves into the complexities of reasoning in AI, highlighting the need for generative reasoning over traditional benchmarks. The conversation outlines the limitations of current approaches and the importance of developing new paradigms for meaningful AI learning.
Transcript
Play full episode
