Due to its huge success in industry, mechanism design has been one of the central research topics at the interface of economics and computer science. However, despite decades of effort, there are still numerous challenges, in terms of both theory and applications. These include the problem of how to design mechanisms for selling multiple items, dynamic auctions, and balancing multiple objectives, given the huge design space and buyer strategy space; and the fact that in practice, the most widely applied auction format (the generalized second price auction) is neither truthful nor optimal. Furthermore, many theoretical results are based upon unrealistic assumptions that do not hold in real applications.

This book presents the AI-driven mechanism design framework, which aims to provide an alternative way of dealing with these problems. The framework features two abstract models that interact with each other: the agent model and the mechanism model. By combining AI techniques with mechanism design theory, it solves problems that cannot be solved using tools from either domain alone. For example, it can reduce the mechanism space significantly, build more realistic buyer models, and better balance different objectives.

The book focuses on several aspects of mechanism design and demonstrates that the framework is useful in both theoretical analysis and practical applications.

Autorentext

Weiran Shen is an assistant professor at Gaoling School of Artificial Intelligence, Renmin University of China. His research interests lie mainly at the interface of computer science and economics, including but not limited to multi-agent systems, game theory, mechanism design, and the connection between these domains and AI techniques. His research in mechanism design has already been implemented by online advertising platforms, such as Baidu and ByteDance.

Pingzhong Tang is an associate professor at IIIS, Tsinghua University. His current research focuses on the interdisciplinary topics relating to AI, multiagent systems, and economics. He works on both theoretical and applied problems. Examples of his past work include simple and optimal auctions, dynamic ad auctions (published in Econometrica and used in Google ads), and water rights market design (used in Gansu province, China), as well as reinforcement mechanism design (used in Baidu advertising and Taobao search).

Song Zuo is a senior research scientist at Google Research. His primary research interests are in the area of auction and dynamic mechanism design for internet advertising and general real-world applications. He was awarded the 2017 Google PhD Fellowship for his research.