Does AI Help or Harm? Endogenous Information Acquisition with AI Advice

2026-09-28

AI Summary

This study investigates how AI recommendations, when intervening in the pre-decision information acquisition stage, affect final decision quality. Grounded in Bayesian decision theory and a posterior-separable utility of persuasion (UPS) cost model, it proposes a convexity criterion for continuation value functions alongside an active-set slope test to systematically analyze the net effects of AI advice on belief updating, information-seeking behavior, and total utility. The analysis reveals the underlying mechanisms through which “stopping concentration” and “attention dilution” impair utility. It formally proves that AI recommendations yield weakly positive gains in binary-state problems but can be detrimental in multi-state settings. Furthermore, the study uncovers a counterintuitive finding: cheaper explanations may paradoxically reduce overall utility.

Abstract

AI advice increasingly enters human decision workflows before costly information acquisition and final action. This paper studies how such advice affects decision quality, measured by the gross utility of the final action. Advice shifts beliefs before the decision maker chooses an acquisition experiment with uniformly posterior-separable (UPS) costs. The analysis characterizes gross effects through the continuation gross value generated by net-optimal acquisition. Convexity of this value function is necessary and sufficient for every AI signal to weakly improve gross utility at every prior, while nonconvexity allows a binary signal to cause gross harm. The criterion gives tight state-space results. Every AI signal is weakly gross-improving in binary-state problems under arbitrary actions, payoffs, and UPS costs. A minimal three-state problem can generate gross harm through concentrated shutdown, where valuable acquisition stops. For classification problems with Shannon acquisition costs, the criterion becomes an active-set slope test. It guarantees weak gross improvement for homogeneous classification and all three-state weighted classification problems, while a minimal four-state failure arises from attention dilution, where acquisition shifts away from high-stake distinctions. Along a Blackwell-increasing mixture path, gross utility can first decrease and then increase. With endogenous interpretation, cheaper interpretation can reduce gross utility.

Problem

• AI advice
• decision quality
• endogenous information acquisition
• gross utility
• information value

Innovation

• Endogenous information acquisition
• Uniformly posterior-separable costs
• Continuation gross value convexity
• Concentrated shutdown
• Attention dilution