SOURCE-LINKED INTELLIGENCE

ObserverBench: Testing Mechanistic Estimates for Intervention and Control

arXiv · AI, language, vision and robotics · article · Sep 2, 2026 · UTC

Mechanistic interpretability is increasingly used to guide interventions such as activation steering, circuit removal, and safety monitoring. Yet an internal estimate that is accurate on average can still choose a poor action. We present ObserverBench, a benchmark framework for testing whether an internal estimator---an observer---is adequate for the intervention, control, or safety task it directs. Each task fixes the model, information boundary, allowed actions, decision rule, held-out cases, and loss. The benchmark reports estimation accuracy separately from the loss caused by the chosen ac

Read original source ↗ Open in workspace

recordType

paper

region

Global