Time/时间： 2022.9.23（周五） 10:00-11:30
Venue/地点：Room 501, Jiageng Building 2/嘉庚二号楼501教室
Topic/题目：The Effects of Monetary Incentives and Accountability on Artificial Intelligence-assisted Decision-making
Online Conference ID/腾讯会议号：：357-549-637
Abstract/摘要：Artificial intelligence (AI) has improved performance across many domains, and is increasingly being used to support decision-makers in the decision-making process. However, research and anecdotal evidence show that people supported by AI in the decision-making process frequently overrely on the AI, which sometimes leads to undesirable consequences. In this study, we examine whether holding decision-makers accountable and providing monetary incentives reduce their reliance on AI. We conduct an experiment wherein credit staff from commercial banks perform online personal credit assessments using an AI-assisted decision-making system. We find that both accountability and monetary incentives reduce participants’ reliance on AI, increase participants’ efforts, and improve their decision-making accuracy when they are assisted by AI in the decision-making process. More importantly, our results show that for participants who overrely on AI, the more effort they exert, the less accurate their decision-making becomes. These results speak to the complexity of human–AI interaction in the decision-making process, indicating that when using AI to assist humans in decision-making, it is important to consider how much humans rely on AI, in addition to the amount of time they spend on the decision-making process. Our findings are important for firms establishing management control systems to monitor their employees’ behavior when making decisions using machine-assisted systems.
About the Speaker/报告人简介：James is a PhD candidate in accounting at Institute for Financial & Accounting Studies (2017 grade). His research activities focus on management accounting (e.g., performance measurement, incentives design, and corporate governance); application of AI in accounting decision-making; and neuroaccounting. He uses neuroscience, psychology, and economic theory, and apply lab and field experiments, archival data analysis, and natural language processing to test his empirical predictions.