Retrieval vs. Reasoning
Subbarao discusses the ongoing debate around the capabilities of large language models (LLMs), emphasizing that they do not function as traditional databases. He argues that while LLMs may not be retrieving information, labeling their output as reasoning is misleading, as they often fail to meet the criteria for sound reasoning. The conversation highlights the complexity of understanding autoregressive models and their unique role in AI development.In this clip
From this podcast
Machine Learning Street Talk (MLST)
Subbarao Kambhampati - Do o1 models search?
Related Questions