Fast-and-frugal AI cues reshape clinical judgment and cognitive effort in comparative field experiments in the US and India

Abstract

Uncertainty, time pressure, and information overload are common characteristics of clinical decision-making, prompting clinicians to rely on fast-and-frugal heuristics. The theory of fast-and-frugal decision-making posits such recourse does not require integration of all cues but can rely instead on take-the-best cue. The introduction of artificial intelligence (AI) has added a new class of informational cues that aggregate multiple clinical indicators into a single output with the potential to function as a dominant heuristic that relieves clinicians from having to process and integrate multiple cues. However, the evidence suggests that routine usage of AI among clinicians is low. Little research has addressed whether cultural, technological, and institutional factors may influence the usage, evaluation, and adoption of AI cues across healthcare markets. This study addresses these gaps through a between-subjects field experiment with clinicians in the U.S. and India. Participants assessed 15 patient scenarios with or without an AI cue. The results show that the AI cue becomes a take-the-best heuristic in the mature AI market (U.S.), but in the emerging market (India) usage of traditional cues remains dominant. Consequently, AI-generated output should be presented as complements to the support of clinical decision-making to allow clinicians to use familiar approaches to validate AI assessments.

Ähnliche Arbeiten