Portrait of Fan Yao

Fan Yao

Assistant ProfessorUniversity of North Carolina at Chapel Hill

I am an assistant professor in the Department of Statistics and Operations Research (STOR) at UNC Chapel Hill. I study the algorithmic interplay between human behavior and ML/AI systems. I received my PhD in Computer Science from the University of Virginia, advised by Haifeng Xu and Hongning Wang. Prior to obtaining my PhD, I spent two years at the University of Chicago and received both my BS and MS in Computational Mathematics from Peking University.

Research interests

  • Social and economic aspects of AI
  • Human-centered ML/AI
  • Strategic and multi-agent systems
  • Recommender systems
  • Digital platform economy

Share these interests? Email me. Prospective students, please read this page first.

Research

Three questions run through my work.

A feedback loop: people supply data and feedback to ML/AI systems, which return incentives such as exposure, rewards and answers; strategic interaction sits between them. Markers 1 to 3 match the three questions below. Now AI agents join the loop on both sides, raising two open questions: what changes for mechanisms built for humans, and what new feedback loops arise when models learn from agent-made data.Incentivesexposure · rewards · answersData & feedbackchoices · content · preference labelsStrategic interactionlearning dynamics · equilibria123Peopleusers · creatorsstrategic actorsML/AI systemsrecommenders · LLMslearning algorithmsNOWAI agents join the looptheir AI agentsagentic systemsMechanisms built for humans:what changes with agents?Learning from agent-made data:new feedback loops? A feedback loop between people and AI systems: incentives (1), strategic interaction (2), data and feedback (3), matching the three questions below. Now AI agents join on both sides, raising new questions for mechanisms and for learning from agent-made data.IncentivesData & feedbackStrategicinteraction123PeopleAI systemsAI agentsagentic AINOWAI agents join the loopMechanisms built for humans: what changes?Learning from agent-made data: new loops?

  1. Creators & AI

    How should platforms and AI systems reward the people who create content?

  2. Strategic learning

    What happens when learning algorithms meet strategic people and agents?

  3. Alignment & trust

    How do training choices shape LLM alignment, and how can deployed models be protected?

Drawing onAlgorithmic game theoryOnline & reinforcement learningDynamical systemsLearning theory

Find a reading route through my papers

Publications

Ask asks Faro about that paper.

Show 8 more publicationsShow fewer

Full list on Google Scholar

Explore with AI

Beta
Ask Faro Questions about my papers, courses or research ideas.

Teaching

I teach machine learning through its mathematics and its consequences for people and society, connecting theory with code, discussion, and projects.

Fall 2026 · UNC Chapel Hill

STOR 565 Machine Learning A mathematical introduction to learning from data: loss, optimization, generalization, and implementation. Advanced undergrad & MSPython experiments Course website STOR 323I Machine Learning: Ethics and Society Fairness, privacy, strategic behavior, and generative AI, through technical and philosophical lenses. UndergraduateDiscussion & case studies Course website

Courses & teaching philosophyTry a learning challenge

News

  1. Invited talk on decision-aware preference alignment in closed-loop LLM training, and a co-organized session on LLMs and Social Media Analytics with Ganghua Wang, at the INFORMS Annual Meeting, San Francisco.
  2. Invited talk on the adaptation–stability tradeoff in iterative preference optimization at SLDS 2026, New York.
  3. Three papers accepted at NeurIPS, ICML, and EC 2026.
  4. Received an award from the UNC AI Acceleration Program.
  5. Invited talk on LLM alignment dynamics at the LLMs and Strategic Agents for Machine Learning workshop, Northwestern University.
Earlier news
  1. Invited talk on LLM alignment dynamics at the FTEC Seminar, HKUST (Guangzhou), online.
  2. Co-hosted the INFORMS session Preference Learning in Large Language Models with Chong Liu and gave the talk “Rethinking the role of GenAI in content creation ecosystem,” Atlanta.
  3. Invited panelist at the Aligning AI with Society workshop, UNC.
  4. Joined STOR at UNC Chapel Hill as a tenure-track assistant professor.

Beyond Research

Tennis, cooking, music, two cats, and a few places I have called home.

More about meCuriosity notebook