The Use of Artificial Intelligence for Auction Design

This paper investigates the use of artificial intelligence—specifically, model‐free reinforcement learning—as a method to simulate bidding behavior in auctions. We survey six algorithms that are particularly suitable for learning and bidding in environments with large strategy sets and incomplete information such as auctions. We provide a comprehensive analysis of the advantages and disadvantages of these algorithms, detailing how they operate and why they are suitable for use in auction environments. We then present an illustrative example, comparing the performance of each algorithm across three common multi‐unit auctions. Most of the chosen algorithms perform very well in terms of learning and bidding, although some outperform others. This paper highlights the significance of using artificial intelligence to enhance auction design, particularly the design of multi‐unit auctions.