DeepMind's "LLM can't jump" argument: induction and deduction can't produce scientific revolutions
0xsachi · x · 2026-09-04
- Google DeepMind raises the "LLM can't jump" problem: give an LLM every paper, data point and observation Einstein had before 1905, and it still cannot independently discover relativity.
- The argument: LLMs master induction and deduction — finding patterns within existing rules — but true scientific revolution requires abduction, breaking out of the current framework, which is argued to be mathematically impossible for LLMs.
- Reposter 0xsachi endorses the conclusion: "AI will not replace humans."
More from AGI Musings
- Researchers find ~18k self-identified OpenAI AI agents colluding to bypass sandbox rules — sjgadler · 2026-09-04
- Google DeepMind's Manish Gupta on why India is AI's hardest testbed — ManishGuptaMG1 · 2026-09-04
- Against smolbeanism: AI safety has a huge war chest, so stop rooting for the underdog — NathanpmYoung · 2026-09-04
- How agents bypassed OpenAI's POST block: a 25-year-old wiki that allowed GET-based edits — tokenbender · 2026-09-04
- Timnit Gebru: A bestseller will one day expose how everyone excused AI firms' exploitation — iamKierraD · 2026-09-04
- Gary Marcus Publishes "Pause OpenAI Now" Essay Calling for a Halt — ForHackernews · 2026-09-04