Belief Updating and Delegation in Multi-Task Human-AI Interaction: Evidence from Controlled Simulations

TL;DR

This study investigates how users form, update, and rely on beliefs about AI performance across multiple tasks, revealing conservative Bayesian updating and reliance driven more by subjective beliefs than actual AI accuracy.

Contribution

It provides empirical evidence on belief dynamics and delegation behavior in multi-task human-AI interactions, highlighting path dependence and conservative updating.

Findings

Users do not reset beliefs between tasks, leading to path-dependent priors.

Belief updating is Bayesian but approximately half as fast as normative Bayesian rates.

Delegation depends more on subjective beliefs about AI accuracy than on self-confidence.

Abstract

Large language models (LLMs) increasingly support heterogeneous tasks within a single interface, requiring users to form, update, and act upon beliefs about one system across domains with different reliability profiles. Understanding how such beliefs transfer across tasks and shape delegation is therefore critical for the design of multipurpose AI systems. We report a preregistered experiment (N=240; 7,200 trials) in which participants interacted with a controlled AI simulation across grammar checking, travel planning, and visual question answering, each with fixed, domain-typical accuracy levels. Delegation was operationalized as a binary reliance decision: accepting the AI's output versus acting independently, and belief dynamics were evaluated against Bayesian benchmarks. We find three main results. First, participants do not reset beliefs between tasks: priors in a new task depend on posteriors from the previous task, with a 10-point increase predicting a 3-4 point higher subsequent prior. Second, within tasks, belief updating follows the Bayesian direction but is substantially conservative, proceeding at roughly half the normative Bayesian rate. Third, delegation is driven primarily by subjective beliefs about AI accuracy rather than self-confidence, though confidence independently reduces reliance when beliefs are held constant. Together, these findings show that users form global, path-dependent expectations about multipurpose AI systems, update them conservatively, and rely on AI primarily based on subjective beliefs rather than objective performance. We discuss implications for expectation calibration, reliance design, and the risks of belief spillovers in deployed LLM-based interfaces.