"Human-laundering": Math researchers quietly rely on LLMs but never credit them
cloneofsimo · x · 2026-08-24
cloneofsimo coins the term "human-laundering" — researchers pretending solutions and ideas are their own rather than the machine's. He speculates that over 50% of current math research somehow involves or heavily relies on LLMs, yet authors avoid crediting them since it would make them look stupid and bias the paper toward reading as AI slop.
The quoted tweet highlights a striking contrast: this generational run of math results comes from one Anthropic guy (Levent) and his friend, not from MIT or Princeton math departments.
More from AGI Musings
- AI's impact on work: Beyond job losses to transforming operations — GregCook2011 · 2026-08-24
- After LLM counterexamples to major conjectures, a mathematician's spiritual crisis — Dr_Atoosa · 2026-08-24
- Opinion: Speed matters more than smarts for AI agents — maddie-lovelace · 2026-08-24
- Google CTO explores the boundaries of life's origins — danfaggella · 2026-08-24
- AI leads to massive devaluation of knowledge, shifting human values — SebastianNehrd2 · 2026-08-24
- From Agent to 'Butler': Redefining the role of AI assistants — yihui_indie · 2026-08-24