🤖 AI Summary
This work addresses the critical challenge of high-cost errors arising from overconfident predictions of large language models in decision-making systems. To mitigate this risk, the authors propose the Risk-Aware Causal Gating (RACG) framework, which uniquely integrates causal effect estimation with distribution-free risk control. By evaluating the causal pathways from actions to outcomes, RACG dynamically determines whether to execute, defer, or reject a prediction, thereby enabling safe decision-making under minimal privilege. The framework incorporates counterfactual risk estimation, an adaptive gating policy, and a selective prediction mechanism. Empirical results across multiple simulated interventions and real-world decision benchmarks demonstrate that RACG significantly reduces high-cost errors—outperforming existing confidence-based and selective prediction approaches at equivalent abstention rates—while preserving the majority of decision utility.
📝 Abstract
Modern decision systems increasingly rely on learned components whose outputs may be confident yet wrong, exposing downstream actions to costly errors. We introduce Risk-Aware Causal Gating (RACG), a framework that decides whether to act on, defer, or abstain from a model's prediction by combining causal effect estimation with calibrated risk control. RACG models the causal pathway from candidate actions to outcomes and gates each decision according to an estimated counterfactual risk rather than raw predictive confidence. To make gating reliable, we derive distribution-free bounds on the probability of acting under high-risk conditions and show how these bounds translate into operating thresholds that satisfy user-specified safety constraints. We further propose an adaptive gating policy that adjusts to distribution shift by monitoring discrepancies between predicted and realized outcomes, tightening the gate when causal assumptions appear violated. Across simulated interventions and real-world decision benchmarks, RACG reduces high-cost errors substantially while preserving most of the utility of an ungated policy, and it outperforms confidence-based and selective-prediction baselines at matched abstention rates. Our results indicate that explicitly separating causal risk from predictive uncertainty yields decision systems that are both safer and more transparent, offering a principled mechanism for trustworthy automation in high-stakes settings.