Title: AI Meets Programmable Finance (video recording)

Abstract: Traditional financial markets are bound by rigid structures and centralized policies, while AI has revolutionized the strategies of some market participants without altering the underlying calcifications. This talk examines the designs of a flexible and adaptive market infrastructure that aligns with the agility of AI-driven participants; we conduct our study in the concrete context of decentralized finance (DeFi) which merges computer science principles with financial markets by natively integrating programmability, atomicity, and compositionality.

We explore how fundamental market functions (e.g., exchanges and lending platforms) can adapt to dynamic conditions using AI, emphasizing the need for adversarial resistance in an environment where all participants operate on equal terms ("the other side can do magic too"). We propose adaptive solutions to protect liquidity providers and respond to non-stationary markets in (near) optimal ways for a variety of natural performance metrics. Our results are of basic interest in theoretical economics by providing (optimal) computational solutions to classical economic models (e.g., Glosten-Milgrom model of market making) and of independent interest in the theory of reinforcement learning.

We conclude by advocating the establishment of resilient, equitable, and transparent financial systems powered by the synergy of AI and decentralized technologies. We showcase the integration of AI in blockchains to create "genuinely smart" contracts; conversely, we show how to design (via blockchain principles) community-built, owned, and governed AI marketplaces.

Speaker bio: Pramod Viswanath is the Forrest G. Hamrick Professor of Engineering at Princeton University. His research interests are in designing scalable decentralized platforms (blockchains) and using the underlying principles to design and build AI marketplaces with native ownership rights of participants.