Weirui Peng
Hi! I'm an HCI researcher with a Master's degree from Columbia University, where I was advised by Lydia Chilton. My research focuses on Human-AI collaboration, specifically designing intelligent tools that bridge the gap between human cognition and AI systems for complex tasks like collaborative work, design, and coding.
Previously, I completed my undergraduate studies at The Chinese University of Hong Kong. I am also fortunate to be mentored by distinguished researchers, including Rachel Adler, Toby Jiajun Li, Sherry Tonghuang Wu, Niki Kittur, and Nikolas Martelaro.
News
- October 2024: Attending UIST 2024 ✈️ in Pittsburgh
- September 2024: One paper got accepted to CSCW 2024 🎉
Research
Presented below is a comprehensive list of my PUBLISHED works, as well as those available on ARXIV, with the exclusion of any unpublished manuscripts.
LADICA: A Large Shared Display Interface for Generative AI Cognitive Assistance in Co-Located Team Collaboration
Zheng Zhang , Weirui Peng , Xinyue Chen , Luke Cao , Toby Jia-jun Li
Large shared displays, like digital whiteboards, support co-located team collaborations by assisting with tasks such as brainstorming and organizing ideas, but most existing systems either limit functionality or reduce human control, neglecting natural group dynamics. To address cognitive challenges such as diverse ideation and synchronization with live discussions, we developed LADICA, an interface that enhances brainstorming and analysis while promoting mutual awareness of team ideas. A lab study demonstrated LADICA's effectiveness, showing its ability to extract key information from discussions and improve team collaboration.
Accepted to ACM CHI 2025
What Should We Engineer in Prompts? Training Humans in Requirement-Driven LLM Use
Qianou Ma , Weirui Peng , Chenyang Yang , Hua Shen , Kenneth Koedinger , Tongshuang Wu
Prompting ChatGPT for complex tasks, such as building a customer support chatbot, often requires meticulous prompt engineering, including fluent writing and chain-of-thought techniques, yet conveying customized requirements remains a human-centric challenge. We introduce Requirement-Oriented Prompt Engineering (ROPE), a paradigm that emphasizes creating clear, complete requirements during prompting, supported by an assessment and training suite offering LLM-generated feedback. A study with 30 novices demonstrated that requirement-focused training significantly improved prompting performance compared to traditional methods.
Accepted to ACM TOCHI
Exploring the Diversity of Music Experiences for Deaf and Hard of Hearing People
Kyrie Zhixuan Zhou , Weirui Peng , Yuhan Liu , Rachel F. Adler
Sensory substitution techniques have been suggested to help deaf or hard of hearing (DHH) people engage with music, but little is known about how these methods enhance their experience. To understand DHH people's interaction with music, we conducted interviews and analyzed Reddit communities, finding that they use sign language and visual/haptic cues to feel the music, and prefer familiar, non-lyrical, loud music. However, current hearing aids and enhancement techniques fall short, leading to suboptimal experiences, highlighting the need for better design to improve music accessibility for DHH individuals.
Accepted to CSCW 2024
Enhancing User Engagement in AI Auditing Through Gamification and Storytelling
Weirui Peng , Mengyi wei , Kyrie Zhixuan Zhou
Engaging end users in AI system auditing is challenging given their varying tech literacy and a lack of incentive. In this paper, we propose a collaborative approach combining gamification and storytelling to incentivize end-user participation in AI auditing and make the process more engaging. The user-centric pipeline incorporates game-like rewards and narrative frameworks. Our methodology aims to make AI auditing more interactive and inclusive, contributing to the development of more ethical AI systems.
Accepted to CSCW 2023 User AI Auditing Workshop