Yejin Son

Hi! I’m Yejin Son, an incoming PhD student in Computer Science at the University of British Columbia, where I will be advised by Vered Shwartz, starting in Fall 2026. I will also be affiliated with the Vector Institute. I am currently a Master’s student in Artificial Intelligence at Yonsei University advised by Prof. Youngjae Yu. I am also a visiting researcher at the Jinesis AI Lab at the University of Toronto, an external collaborator at the Vector Institute, working with Prof. Zhijing Jin. I received my B.A. in Economics and Applied Statistics from Yonsei University.

Research Interests

I build socially aware AI that behave like humans, enabling two key applications: simulating human society for social science research and creating human simulators that help agents serve people from diverse cultural and situational backgrounds. Ultimately, I strive to contribute to AI that not only understands us more fully but also inspires us to connect more deeply with one another.

News

May 2026

I will be attending ICML 2026 Workshop with two workshop papers, presenting a poster on our work on latent cultural boundaries in LLMs at the Pluralistic Alignment Workshop, and our K-pop concert audience agent paper accepted as an oral at the Culture x AI Workshop. See you in Seoul!

Apr 2026

I will be attending ACL 2026 to present our paper on counterfactual unfairness in LLMs through humor. See you in San Diego!

Apr 2026

I am serving as a Program Chair of the ICML 2026 Workshop on Trustworthy AI4GOOD.

Mar 2026

I am excited to share that I will be joining the University of British Columbia as a PhD student in Computer Science, advised by Prof. Vered Shwartz, starting Fall 2026, supported by the Four Year Fellowship (4YF)!

Publications

Pluralistic Alignment @ ICML 2026 Workshop

The Wedge Questions: Latent Cultural Boundaries in LLMs via Persona Projection Divergence
Yejin Son, Yongjin Yang, Ryan Faulkner, Matt Ratto, Seungwon Lim, Youngjae Yu, Zhijing Jin
ICML Pluralistic Alignment Workshop 2026
[Paper]

Culture x AI @ ICML 2026 Workshop

Does Persona Make LLMs K-pop Fans? A Pilot Study of LLM-Based Online Concert Audience Agents
Kirak Kim, Hyojin Kim, Yejin Son, Sungyoung Kim, Kyung Myun Lee
Arxiv
[Paper]

ACL 2026 Main

Investigating Counterfactual Unfairness in LLMs towards Identities through Humor
Yejin Son*, Shubin Kim*, Junyeong Park, Keummin Ka, Seungbeen Lee, Jaeyoung Lee, Hyeju Jang, Alice Oh, Youngjae Yu
Arxiv
[Paper]

Arxiv

Mind the Motions: Benchmarking Theory‑of‑Mind in Everyday Body Language
Seungbeen Lee, Jinhong Jeong, Donghyun Kim, Yejin Son, Youngjae Yu
Arxiv
[Paper]

EMNLP 2025 Main

Subtle Risks, Critical Failures: A Framework for Diagnosing Physical Safety of LLMs for Embodied Decision Making
Yejin Son*, Minseo Kim*, Sungwoong Kim, Seungju Han, Jian Kim, Dongju Jang, Youngjae Yu, Chan Young Park
Published in EMNLP 2025
[Paper]

COLM 2025

G1yphD3c0de: Towards Safer Language Models on Visually Perturbed Texts
Yejin Choi, Yejin Yeo, Yejin Son, Seungju Han, Youngjae Yu
Published in COLM 2025
[Paper]

EMNLP 2023

Reading Books is Great, But Not if You Are Driving! Visually Grounded Reasoning about Defeasible Commonsense Norms
Seungju Han, Junhyeok Kim, Jack Hessel, Liwei Jiang, Jiwan Chung, Yejin Son, Yejin Choi, Youngjae Yu
Published in EMNLP 2023
[Paper]