Woojung Song

I'm an integrated M.S.–Ph.D. student in Data Science at Seoul National University, advised by Yohan Jo in the HOLI Lab.

I study how language-model agents behave in practice: how they use and trust tools, interact with users, and express values and personas through their actions. I develop evaluations that connect model behavior to concrete tasks and changing contexts. My goal is to use these measurements to improve agent reliability and build more faithful simulations of people and fictional characters.

opusdeisong@snu.ac.kr
Woojung Song

Research Interests

  1. [Agent Reliability] How agents use tools and interact with users

    I study recurring agent behaviors, failures caused by unreliable tools, and the effects of user behavior on task completion.

  2. [Human Simulation] How values and personas translate into behavior

    I evaluate whether models' expressed values and personas predict their behavior, and whether simulated characters adapt to the right context and point in time.

My research asks: How can we measure and improve agent behavior across tasks and social contexts?

I'm currently seeking a summer research internship in agents, alignment, or social simulation, and I'm always open to collaboration. Feel free to reach out at opusdeisong@snu.ac.kr.

News

Earlier news