Is VLA Reasoning Faithful? Probing Safety of Chain-of-Causation in Autonomous Driving Models 事件

Is VLA Reasoning Faithful? Probing Safety of Chain-of-Causation in Autonomous Driving Models arXiv:2605.17268v2 Announce Type: replace-cross Abstract: We present the first systematic study of faithfulness in Vision-Language-Action (VLA) driving models, analyzing 300 Alpamayo-R1-10B inferences across 100 diverse PhysicalAI-AV scenarios. Our main finding is that output natural-language rationales with trajectories may be significantly unfaithful: (i) overall reasoning fidelity is only 42.5%, with