openreview.net
Position: LLMs can't jump
A position paper on LLMs, abduction, and scientific invention.
About Position: LLMs can't jump
Tom Zahavy's position paper argues that generative AI excels at induction and deduction but lacks the abductive leap needed to create novel scientific axioms. Using Einstein's formulation of general relativity as a case study, it proposes physically consistent, multimodal world models as a path toward artificial scientific invention.
Categories
Topics
Discussion highlights
AI-generated summary based on Hacker News comments
🔗 Full discussion: https://news.ycombinator.com/item?id=49181083
No comments yet. Have you tried it?