LLMs and Reasoning
Subbarao expresses skepticism about the reasoning capabilities of large language models (LLMs), arguing that merely increasing model size won't enhance their reasoning skills. He contrasts LLMs with reinforcement learning systems, like AlphaGo, which are inherently designed for reasoning tasks. The discussion emphasizes the need for exploring alternative models rather than relying solely on the expansion of existing LLM architectures.In this clip
From this podcast
Machine Learning Street Talk (MLST)
Prof. Subbarao Kambhampati - LLMs don't reason, they memorize (ICML2024 2/13)
Related Questions