Researchers accuse an AI lab paper of overlapping their ARTS work without attribution
xwang_lk · x · 2026-10-15
Researcher xwanglk publicly flagged that a recent paper has substantial conceptual overlap with their team's June 2026 ARTS work ("Learning the ARTS to Search for Automated Discovery") without acknowledgment, writing that "proper attribution is fundamental to research integrity."
The quoted thread explains ARTS's core insight: separating the scientist agent into a hypothesis proposer and an executor, which (a) lets LLM reasoning focus on hypothesis generation, (b) allows isolated training of the generator without degrading execution, and (c) drastically reduces context — since long context degrades LLM reasoning.
The author also argues major AI labs' outsized narrative-shaping power comes with a responsibility to recognize prior contributions regardless of origin.
More from Companies & People
- Snorkel AI drops Frontier Data Summit keynote on 'the research era of AI data' — ajratner · 2026-10-15
- Sentdex: shockingly many people I ask privately turn out to have joined OpenAI — Sentdex · 2026-10-15
- Boston Dynamics shows Atlas factory: one robot a day today, one hour targeted next year — CyberRobooo · 2026-10-15
- Together AI x NVIDIA x Reflection pack a NYC builder meetup, Berlin and SF next — togethercompute · 2026-10-15
- AI Companies Play Cat-and-Mouse With Secret-Stealing Spies Amid FBI Arrest — pstAsiatech · 2026-10-15
- NDAs now ban uploading deal materials to non-enterprise Claude/ChatGPT over training concerns — sebpaquet · 2026-10-15