Tools and Knowledge
Subbarao emphasizes that AI tools, like search engines, are built upon our collective knowledge rather than generating new insights independently. He argues that the effectiveness of these technologies relies on the vast data available on the web, which has been curated by humans. Additionally, he expresses concern about the increasing noise in information as AI-generated content proliferates, suggesting the need for a "seed vault" to preserve the integrity of our current web data.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
Can knowledge creation shape the world as discussed in the episode ChatGPT and the Dawn of Computerized Hyper-Intelligence | Brian Roemmele | EP 357 and the clip Preserving Knowledge?
Can knowledge creation shape the world as discussed in the episode "ChatGPT and the Dawn of Computerized Hyper-Intelligence | Brian Roemmele | EP 357" and the clip "Preserving Knowledge"?
What's your opinion on using large language models (LLMs) for scientific research, especially for generating new ideas for hypotheses as discussed in the episode "Neurosymbolic AI in Search with Professor Laura Dietz - Weaviate Podcast #49!" and the clip "Knowledge Graph Queries"?