Samuel Wong | Department of Statistics and Actuarial Science, University of Waterloo
https://swong.ca/index.html
Pierre-Nicholas Roy | Department of Chemistry, University of Waterloo
http://www.science.uwaterloo.ca/~pnroy/
Giang Tran | Department of Applied Mathematics, University of Waterloo
https://uwaterloo.ca/scholar/g6tran
Alex Kontorovich | Department of Mathematics, Rutgers University
sites.math.rutgers.edu/~alexk/
Many online platforms have associated algorithmic tools that can optimize on a user's behalf, such as algorithmic bidders or price optimizers. Recent advances in agentic AI enable even more use cases for LLM-powered tools and advice. But as algorithmic assistance becomes increasingly common, how will users strategically maneuver their usage of these tools, and what will be the system-wide impact?
In this talk we will address these questions through a game-theoretic lens. We will explore applications of strategic AI agent usage, from advertising auctions to persuasion and sales. Viewing algorithmic agents as components of a broader system, we will discuss ways to mathematically model agent interaction and predict equilibrium outcomes. Along the way, we will study the strategic choices of the users who control the agents (and who anticipate their use by others). We will also discuss implications for platform and AI agent design.
Based on joint work with Yiding Feng, Keegan Harris, Nicole Immorlica, and Alex Slivkins