Spring 2026 IDE Lunch Seminar (Virtual Ticket): Vartan Shadarevian
Spring 2026 IDE Lunch Seminar (Virtual Ticket): Vartan Shadarevian
Vartan Shadarevian: "General Strategic Intelligence: AI Agents for the New Economy"
Biography: Vartan Shadarevian is a PhD candidate in Economics at Princeton University, where he works in economics and AI. His research spans multi-agent AI, game theory, financial markets, and regulatory policy, cited by FTC Commissioners. Prior to Princeton, he practiced law at Skadden (financial regulatory enforcement) and Kirkland & Ellis (investment funds), and holds a JD from Harvard Law School, where he was a John M. Olin Fellow in Law & Economics. He co-founded and led the Aleph Policy Initiative, a nonprofit that advised the U.S., U.K., and Canadian governments.
Abstract: What happens when AI agents operate as strategic actors in unfamiliar environments? This talk introduces GENSTRAT, a benchmark built on procedurally generated imperfect-information extensive-form games, with a view toward building general economic environments, and uses it to evaluate the capabilities and limitations of AI agents. The talk also discusses ongoing work on emergent coordination, generative mechanism design, and implications for competition, regulation, and market structure as AI agents are deployed with greater strategic autonomy.
Spring 2026 IDE Lunch Seminar (Virtual Ticket): Vartan Shadarevian
Vartan Shadarevian: "General Strategic Intelligence: AI Agents for the New Economy"
Biography: Vartan Shadarevian is a PhD candidate in Economics at Princeton University, where he works in economics and AI. His research spans multi-agent AI, game theory, financial markets, and regulatory policy, cited by FTC Commissioners. Prior to Princeton, he practiced law at Skadden (financial regulatory enforcement) and Kirkland & Ellis (investment funds), and holds a JD from Harvard Law School, where he was a John M. Olin Fellow in Law & Economics. He co-founded and led the Aleph Policy Initiative, a nonprofit that advised the U.S., U.K., and Canadian governments.
Abstract: What happens when AI agents operate as strategic actors in unfamiliar environments? This talk introduces GENSTRAT, a benchmark built on procedurally generated imperfect-information extensive-form games, with a view toward building general economic environments, and uses it to evaluate the capabilities and limitations of AI agents. The talk also discusses ongoing work on emergent coordination, generative mechanism design, and implications for competition, regulation, and market structure as AI agents are deployed with greater strategic autonomy.
Good to know
Highlights
- 1 hour
- Online