We consider an extension to classic position auctions in which sponsored creatives are embedded within AI-generated content rather than shown in predefined slots. Leveraging advanced LLM technologies, it becomes viable to seamlessly integrate sponsored creatives with AI content and accurately estimate the context-aware benefits of differing insertion positions. However, this approach introduces novel challenges; substitution effects require rigorous treatment compared to standard position auction settings, where slots are independent of each other.

We study the price of anarchy of the generalized second-price auction where bidders are value maximizers (i.e., autobidders). We show that in general the price of anarchy can be as bad as 0. For comparison, the price of anarchy of running VCG is 1/2 in the autobidding world. We further show a fined-grained price of anarchy with respect to the discount factors (i.e., the ratios of click probabilities between lower slots and the highest slot in each auction) in the generalized second-price auction, which highlights the qualitative relation between the smoothness of the discount factors and the efficiency of the generalized second-price auction.

In recent years, the growing adoption of autobidding has motivated the study of auction design with value-maximizing auto-bidders. It is known that under mild assumptions, uniform bid-scaling is an optimal bidding strategy in truthful auctions, e.g., Vickrey-Clarke-Groves auction (VCG), and the price of anarchy for VCG is 2. However, for other auction formats like First-Price Auction (FPA) and Generalized Second-Price auction (GSP), uniform bid-scaling may not be an optimal bidding strategy, and bidders have incentives to deviate to adopt strategies with non-uniform bid-scaling. Moreover, FPA can achieve optimal welfare if restricted to uniform bid-scaling, while its price of anarchy becomes 2 when non-uniform bid-scaling strategies are allowed.

We investigate auction mechanisms to support the emerging format of AI-generated content. We in particular study how to aggregate several LLMs in an incentive compatible manner. In this problem, the preferences of each agent over stochastically generated contents are described/encoded as an LLM. A key motivation is to design an auction format for AI-generated ad creatives to combine inputs from different advertisers. We argue that this problem, while generally falling under the umbrella of mechanism design, has several unique features. We propose a general formalism---the token auction model---for studying this problem. A key feature of this model is that it acts on a token-by-token basis and lets LLM agents influence generated contents through single dimensional bids.

We study autobidding ad auctions with user costs, where each bidder is value-maximizing subject to a return-over-investment (ROI) constraint, and the seller aims to maximize the social welfare taking into consideration the user’s cost of viewing an ad. We show that in the worst case, the approximation ratio of social welfare by running the vanilla VCG auctions with user costs could as bad as 0. To improve the performance of VCG, We propose a new variant of VCG based on properly chosen cost multipliers, and prove that there exist auction-dependent and bidder-dependent cost multipliers that guarantee approximation ratios of 1/2 and 1/4 respectively in terms of the social welfare.

We analyze the optimal information design in a click-through auction with stochastic click-through rates and known valuations per click. The auctioneer takes as given the auction rule of the click-through auction, namely the generalized second-price auction. Yet, the auctioneer can design the information flow regarding the click-through rates among the bidders. We require that the information structure to be calibrated in the learning sense. With this constraint, the auction needs to rank the ads by a product of the value and a calibrated prediction of the click-through rates. The task of designing an optimal information structure is thus reduced to the task of designing an optimal calibrated prediction.

Auto-bidding has become one of the main options for bidding in online advertisements, in which advertisers only need to specify high-level objectives and leave the complex task of bidding to auto-bidders. In this paper, we propose a family of auctions with boosts to improve welfare in auto-bidding environments with both return on ad spend constraints and budget constraints. Our empirical results validate our theoretical findings and show that both the welfare and revenue can be improved by selecting the weight of the boosts properly.

An incentive-compatible auction incentivizes buyers to truthfully reveal their private valuations. However, many ad auction mechanisms deployed in practice are not incentive-compatible, such as first-price auctions (for display advertising) and the generalized second-price auction (for search advertising). We introduce a new metric to quantify incentive compatibility in both static and dynamic environments. Our metric is data-driven and can be computed directly through black-box auction simulations without relying on reference mechanisms or complex optimizations. We provide interpretable characterizations of our metric and prove that it is monotone in auction parameters for several mechanisms used in practice, such as soft floors and dynamic reserve prices. We empirically evaluate our metric on ad auction data from a major ad exchange and a major search engine to demonstrate its broad applicability in practice.

We study the computation contests where players compete for searching a solution to a given problem with a winner-take-all reward. The search processes are independent across the players and the search speeds of players are proportional to their computational powers. One concrete application of this abstract model is the mining process of proof-of-work type blockchain systems, such as Bitcoin. Although one's winning probability is believed to be proportional to his computational power in previous studies on Bitcoin, we show that it is not the case in the strict sense.

In a typical learning problem, one key step is to use training data to pick one model from a collection of models that optimizes an objective function. In many multi-agent settings, the training data is generated through the actions of the agents, and the model is used to make a decision (e.g., how to sell an item) that affects the agents. An illustrative example of this is the problem of learning the reserve price in an auction. In such cases, the agents have an incentive to influence the training data (e.g., by manipulating their bids in the case of an auction) to game the system and achieve a more favorable outcome. In this paper, we study such incentive-aware learning problem in a general setting and show that it is possible to approximately optimize the objective function under two assumptions: (i) each individual agent is a "small" (part of the market); and (ii) there is a cost associated with manipulation. For our illustrative application, this nicely translates to a mechanism for setting approximately optimal reserve prices in auctions where no individual agent has significant market share. For this application, we also show that the second assumption (that manipulations are costly) is not necessary since we can "perturb" any auction to make it costly for the agents to manipulate.

Dynamic mechanisms are a powerful technique in designing revenue-maximizing repeated auctions. Despite their strength, these types of mechanisms have not been widely adopted in practice for several reasons, e.g., for their complexity, and for their sensitivity to the accuracy of predicting buyers» value distributions. In this paper, we aim to address these shortcomings and develop simple dynamic mechanisms that can be implemented efficiently, and provide theoretical guidelines for decreasing the sensitivity of dynamic mechanisms on prediction accuracy of buyers» value distributions. We prove that the dynamic mechanism we propose is provably dynamic incentive compatible, and introduce a notion of buyers» regret in dynamic mechanisms, and show that our mechanism achieves bounded regret while improving revenue and social welfare compared to a static reserve pricing policy. Finally, we confirm our theoretical analysis via an extensive empirical study of our dynamic auction on real data sets from online adverting. For example, we show our dynamic mechanisms can provide a +17% revenue lift with relative regret less than 0.2%.