市场营销系学术讲座系列2023年第3讲

来源：市场营销系

主 题：1 + 1 > 2? Information, Humans, and Machines

主讲人：张颖婕（北京大学助理教授）

时 间：2023年4月12日（周三）上午10:00-11:30

地 点：明商1008会议室

讲座语言：中文

讲座摘要：

With the explosive growth of data and the rapid rise of artificial intelligence (AI) and automated working processes, humans (either as employees or consumers) inevitably fall into increasingly close collaboration with machines. This incurs, as a consequence, problems in human-machine interaction, not to mention the dilemmas posed by the need to manage information on ever-larger scales. Considering the general superiority of machines in this latter respect, as compared with humans’ own performance, it is essential to explore whether and why human-machine collaboration is valuable. Indeed, due to various constraints such as machine-learning algorithms' inherent “black boxes”, human agents lack an efficient way to work with AI or even show resistance. This fact urges us to further investigate how we could motivate humans to effectively contribute in human-machine collaboration. We aimed to tackle the above issues in the present study. Specifically, we cooperated with a large Asian microloan company to conduct a two-stage field experiment. Drawing upon studies in psychology and information processing that propose different conditions in order to arouse humans’ epistemic motivations, we tuned the treatments by the level of information volume, the presence of collaboration, and the availability of machine interpretations. We observed that with large information volumes or machine interpretations alone, human evaluators could not add extra value to the final collaborative outcomes. However, with the help of large-scale information and the presence of machine interpretations, human involvement significantly reduced the default rate compared with machine-only decisions. We disentangled the underlying mechanisms with three-step empirical analyses. We revealed that the co-existence of large-scale information and machine interpretations can invoke humans’ active rethinking, which in turn, shrinks gender gaps and increases prediction accuracy simultaneously. In particular, we demonstrated that humans could not only double-check the information that had already been considered by themselves and the machine but also could spontaneously associate newly emerging features with others that have the potential to correct the machine’s mistakes but which had been ignored. Such capability both explains the necessity of human-machine collaboration and offers insights into system designs. Our experiments and empirical findings provide non-trivial implications that are both theoretical and practical, respectively.

主讲人简介：

Yingjie Zhang is an assistant professor of Marketing at Guanghua School of Management, Peking University. Before joining PKU, she was an assistant professor of Information Systems at the University of Texas at Dallas. She received her B.S. degrees in Computer Science and Economics from Tsinghua University and her Ph.D. degree in Information Systems and Management from Heinz College, Carnegie Mellon University. Her research interests center on FinTech, mobile and sensor technologies, big data and smart city, user-generated content, and sharing economy, using interdisciplinary methodologies including structural modeling, machine learning, field experiment, econometrics, and analytical models in IS and marketing.  Her research work has been published in the Information Systems Research, Transportation Research Part C, and ACM Transactions on Intelligent Systems and Technology. She is the winner of the Best Paper Award at the International Conference on Information Systems (ICIS 2019) and the winner of the INFORMS ISS Nunamaker-Chen Dissertation Award (2019). She received the Best Published Paper Award Runner-up of Information Systems Research in 2020.

• 上一篇：贸易经济系学术讲座系列2023年第3讲
• 下一篇：组织与人力资源系学术讲座系列2023年第4讲