Hey!
This is Kexin :D
My research envisions more collaborative and socially intelligent interactions between humans and AI, aiming to design systems that augment human sensemaking and agency by distributing cognition across people and intelligent agents in complex, information-rich environments. I study how humans and AI jointly construct, communicate, and act on knowledge through adaptive feedback, social framing, and multi-agent collaboration. Drawing on theories of distributed cognition and human-centered system design, I explore how AI can become a teammate that supports, not substitutes, human reasoning, creativity, and decision-making.
​
Before beginning my Ph.D. in Information Sciences at University of Illinois Urbana–Champaign, I received my B.S. in Cognitive Science and M.S. in Electrical and Computer Engineering from the University of California, San Diego. My early work in creativity support and design cognition examined how computational systems scaffold novice designers’ problem framing and exploration. These experiences shaped my ongoing interest in how intelligent systems mediate cognitive and social processes in complex environments. Now, I am a third year PhD student at UIUC iSchool, working with Prof. Jessie Chin, where I lead several projects submitted and accepted to ACM Creativity & Cognition, CHI, and journals like Computers in Human Behavior.
(Under Review) CHI' 25
MultiColleagues, a multi-agent conversational system that shows how AI agents can act as colleagues by conversing with each other, sharing new ideas, and actively involving users in collaborative ideation.
(Under Review) CHI' 25
Our project explores how personality influences people’s preferences and interactions with AI writing companions. Through two co-design studies, we examine how aligning AI behaviors and communication styles with users’ personalities can enhance engagement, satisfaction, and collaboration in human–AI writing teams.
(Under Review) Computers in Human Behavior
Optimizing decision-making with SERA: The role of gist and
verbatim summaries in information-overload scenarios.
SERA, an AI chatbot that provides gist- and verbatim-based feedback to help users navigate complex decisions, revealing how adaptive feedback can improve sampling efficiency, confidence, and decision quality under information overload settings.
ACM C&C 2025 (Poster)
Our study investigates how large language models generate and recognize humor in emotionally sensitive, support-oriented conversations, revealing persistent challenges in contextual nuance, emotional appropriateness, and role understanding.
