Apologies for cross-posting.
Dear Colleagues and Students,
INFORMS Service Science Online Forum Series - Episode 12 features our next speaker Prof. Ricky Roet-Green (Simon Business School, University of Rochester), who will present her research on how to design AI-human service systems for strategic customers: whether firms should mandate AI as the first point of contact, or grant customers the right to choose human service directly, with important implications for service operations, AI deployment, and customer welfare.
See you online for Episode 12 via this Zoom link, where AI, game theory, and service operations meet!
Speaker: Prof. Ricky Roet-Green (Simon Business School, University of Rochester)
Moderator: Prof. Guangwen Kong, Temple University
Topic: On the Design of an AI-Human Service System with Strategic Customers: Should AI be mandated?
Abstract: AI agents are increasingly deployed across industries, yet their limited proficiency remains a key concern. In many services, customers are first interacting with AI agents and are routed to a human agent only if the AI agent fails to serve them. However, a growing debate about mandating AI services has recently emerged, bringing the U.S. Congress to float bills aiming to grant consumers the right to choose human service directly. Motivated by this realistic setting, we develop a game-theoretical queueing framework to compare customer equilibrium behavior and system performance under two operational designs: an AI-mandate model and a customer-choice model. In both models, customers who attend the AI service are routed to a human service if their case is beyond the capability of the AI agent. Our comparison provides conditions under which each model performs better in terms of throughput and social welfare. Specifically, we consider the impact of the AI proficiency level and demand level on the optimal model choice. Our analysis reveals several non-intuitive results. We find that for a given proficiency level, selecting the model that maximizes the throughput does not depend on the demand level, signaling operational convenience. In addition, when demand is relatively high, we show that the choice model requires a higher AI proficiency level than the AI-mandate model to achieve maximal throughput. In terms of social welfare, our result reveals that granting customer choice can backfire when the AI proficiency level is relatively high. We further show the robustness of our results in the case where some of the customers are reluctant toward the AI service.
Time: June 5, 2026, Friday, 10:00 AM–11:00 AM Eastern Time
Zoom link: https://temple.zoom.us/j/93338101792
Add to Google Calendar
For more information, see our website https://sites.google.com/view/service-science-online-forum/, and our YouTube Channel https://www.youtube.com/playlist?list=PLCn8oCTLj5JEeIiA3_ATZp8gtlkWJCRpO. Join our mail list to get the information about new episodes of Service Science Online Forum: https://forms.gle/k5L52JZbW8kpLYrf7.
------------------------------
Renyu Zhang
Associate Professor
The Chinese University of Hong Kong
Hong Kong
------------------------------