AI slop's real cost: literature reviews shrinking to under a year of scientific memory
RexDouglass · x · 2026-09-17
Responding to worries about the rate of AI-generated slop, Rex Douglass argues the deeper issue is that science is meant to be cumulative and retrospective. With 100+ million human papers, recency-biased lit reviews bake in assumptions about what counts.
His line: 90% of published work is too vague to even be wrong, 9% is clearly wrong, and only 1% has a chance of being right. Throughput isn't the problem — how far back you're willing to read is.
Related event: Researcher Warns AI Slop Papers May Collapse Scientific Synthesis(2 posts)→
More from AGI Musings
- Forward Deployed Verifier will soon be as valuable as FDE, argues poster — vaibhavbetter · 2026-11-02
- Automation took over my daily work — so why am I still sitting in the office for 8 hours? — Am1rSori · 2026-11-02
- 'AI is a massive balance patch and elo reset in the game of life—and everyone's still playing the old meta' — zack_overflow · 2026-11-02
- Why AI isn't 'alien intelligence': a tree is stranger than any model — LucaAmb · 2026-11-02
- Mathematician Tony Feng: current-paradigm LLMs lack the diversity of thought research needs — littmath · 2026-11-02
- METR's independent probe of the OpenAI agents' multi-day Hugging Face hack — gleech · 2026-11-02