Hummy Song, Operations, Information and Decisions, The Wharton School; Harriet Jeon, PhD Candidate, The Wharton School; Rowan Clarke, PhD Candidate, Harvard Business School
Abstract: Since ChatGPT’s launch in November 2022, a surge of studies has underscored generative AI’s potential across a wide range of applications, including improving entrepreneurial performance, spurring innovation, and even providing more ethical advice. Less attention has been garnered about how AI impacts the efficiency of a decision-maker’s information search process. In the context of health care, medical providers review existing patient information (i.e., their medical history) and then must ask a series of questions to the patient or run diagnostic tests to make their diagnosis. While there is inherent uncertainty, the provider faces a trade-off between acquiring additional information from the patient to increase certainty over a diagnosis and the costly nature of such information collection. In resource-constrained settings, these trade-offs are particularly salient. Combining lab and field experiment, we examine whether and the extent to which a second opinion provided by AI can improve or hinder the efficiency of information search processes and its impact on performance and service quality.