Organizer: Yuan Zhou

Abstract: The rapid development of Large Language Models (LLMs) marks a new era of Artificial Intelligence (AI), but the growing power and broad application of AI models also draws concern in credibility, trustworthiness, and safety issues. The initiative of Decentralized AI (DeAI) is a framework that distributes the control, computation, and verification of AI systems across decentralized networks (e.g., blockchains), utilizing economic incentives to motivate self-interested participants to coordinate honestly. In this talk, we introduce our line of study on mechanism design solutions for Decentralized Verification Games, addressing both training and inference tasks in Decentralized AI ecosystems. With our design of incentive-secure Proof-of-Learning (PoL), robust peer prediction, and the T'AIMER (Trustworthy AI Model pEer veRification) frameworks, we address critical existing challenges of heavy computation overheads, lazy verification (i.e., Verifier's Dilemma), and colluding participants that occur in existing designs, ensuring the efficiency and trustworthiness of AI models with robustly incentivized consensus of honest participation. This study provides a systematic and economic framework with rigorous theoretical security and incentive guarantees, fostering the vision of a credible and safe AI ecosystem powered by decentralization.

Bio: Zishuo Zhao earned his Ph.D. degree in Industrial Engineering at University of Illinois Urbana-Champaign (UIUC) in January 2026, advised by Yuan Zhou. He will join National University of Singapore (NUS) as a postdoctoral fellow from Spring 2026, working with Jiaheng Zhang. Prior to Ph.D., he received B.Eng. degree from the Yao Class of Institute for Interdisciplinary Information Sciences, Tsinghua University. Zishuo's research interest lies in the intersection of mechanism design, blockchain technologies, and trustworthy AI, with an emphasis on the design and theoretical boundaries of incentive-compatible decentralized platforms in the perspective of social responsibilities, e.g., social fairness, economical efficiency, and environmental sustainability.