#artificial intelligence
Sep 2026
The Marathon of Scientific Reasoning: Robustness of Scientific Agents to Perturbations in Multi-Turn Interactions
It is shown that current scientific agents remain insufficiently robust to scientifically plausible perturbations, with failures often remaining undetected, propagating, and resisting recovery.
Xiaoting Lyu, Xin-Bo Ma, Yu-Fei Han et al.
· arXiv.org · 0 citations