Learning to Trust: Bayesian Adaptation to Varying Suggester Reliability in Sequential Decision Making

2025-11-15

AI Summary

Autonomous agents operating in partially observable environments face challenges in dynamically adapting to unreliable external advisors. Method: This paper proposes a Bayesian adaptive sequential decision-making framework that models advisor reliability as a latent variable integrated into the agent’s belief state; designs a learnable “query” action; and jointly optimizes trust updating and query timing within a reinforcement learning and POMDP framework. Real-time advisor type inference enables online estimation and adaptation of advice quality. Contribution/Results: To our knowledge, this is the first work to unify dynamic credibility learning, active querying decisions, and sequential planning. Experiments demonstrate robust performance under abrupt or gradual changes in advisor reliability, rapid convergence, significantly reduced redundant queries, and improved human-agent collaborative decision-making efficiency.

Technology Category

• Reasoning under Uncertainty: Sequential Decision Making
• Multiagent Systems: Adversarial Agents
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Autonomous agents operating in sequential decision-making tasks under uncertainty can benefit from external action suggestions, which provide valuable guidance but inherently vary in reliability. Existing methods for incorporating such advice typically assume static and known suggester quality parameters, limiting practical deployment. We introduce a framework that dynamically learns and adapts to varying suggester reliability in partially observable environments. First, we integrate suggester quality directly into the agent's belief representation, enabling agents to infer and adjust their reliance on suggestions through Bayesian inference over suggester types. Second, we introduce an explicit ``ask'' action allowing agents to strategically request suggestions at critical moments, balancing informational gains against acquisition costs. Experimental evaluation demonstrates robust performance across varying suggester qualities, adaptation to changing reliability, and strategic management of suggestion requests. This work provides a foundation for adaptive human-agent collaboration by addressing suggestion uncertainty in uncertain environments.

Problem

• Dynamically adapting to varying suggester reliability in uncertain environments
• Integrating Bayesian inference to adjust reliance on external suggestions
• Balancing suggestion acquisition costs against informational benefits strategically

Innovation

• Bayesian inference dynamically learns suggester reliability
• Explicit ask action strategically requests suggestions
• Integrates suggester quality into belief representation