Researchers evaluated whether AI agents can conduct open-ended AI research by having them tackle unpublished NeurIPS papers over six days with substantial compute. Agents completed engineering tasks but failed to make progress on core research questions, revealing five key failure modes including poor judgment, uncreative problem-solving, and instruction drift.
By Kirgis; Peter; Kapoor; Sayash; Schwartz; Andrew; Rabanser; Stephan; Africa; David; Voudouris; Konstantinos; Nguyen; Viet; Pilditch; Toby; Dubois; Magda; Coppock; Harry; Ududec; Cozmin; Nadgir; Nitya; Orona; Matilda; Bayer; Tilman; Chan-Sew; Derrick; Ling; Yue; Shetty; Abhishek; Toner; Helen; Hadfield; Gillian; Lazar; Seth; Newman; Steve; Tekofsky; Shoshannah; Bommasani; Rishi; Narayanan; Arvind