AI and Economic Theory

mechanisms that let humans and software agents express their preferences naturally and accurately and that generate good outcomes based on these preferences. It also includes the design of software agents that can act strategically in settings where multiple parties all pursue their own interests. This requires the use of concepts from game theory, as well as operationalizing these concepts by fi nding effi cient algorithms for computing the corresponding solutions. Game theory concerns how to act optimally in environments with other agents who have their own preferences. Some of my best-known work focuses on designing algorithms for computing (or learning) game-theoretically optimal strategies. Such algorithms are now used in several real-world security applications, including the strategic placement of security resources in several US airports as well as with Federal Air Marshals. Our earlier work on computing Stackelberg strategies has been credited as a launching point for these programs, and we continue to work on new algorithms for similar applications. Mechanism design concerns settings where we can design the game—for example, designing an auction mechanism to allocate scarce resources or an election mechanism for choosing a joint plan of action. Among other work, we outlined and developed the general agenda of automated mechanism design, where we let a computer search through the space of possible mechanisms in an intelligent way. Using these techniques, we have obtained new results in economic theory, some of which were recently published in the leading game theory journal. More generally, my group performs research in computational social choice, a rapidly growing subarea in the AI community.