Future-of-Work AI

Cognitive AI

Project 1 Quantifying Ambiguity in Visualization with VLMs: A Tool for Real-Time Design Feedback

This project tackles the often-overlooked issue of perceptual ambiguity in data visualizations. Using vision-language models (VLMs), the team quantifies the likelihood of misinterpretation and provides designers with real-time feedback to improve visualization clarity. Integrated into platforms like Observable and ggplot2, this ambiguity-aware toolkit helps analysts proactively identify and address confusing chart elements. (PI: Jinwook Seo, SNU)

---

Project 2 Supporting Collaborative Learning in Programming Education

This project investigates student collaboration in programming tasks, where meaningful interaction occurs in CS education environments. We use LLMs and conversation analytics to systematically analyze patterns and quality of collaboration in CSCL environments. Our goal is to identify key pedagogical factors that foster effective teamwork and to design targeted strategies that support collaborative learning. To this end, we are developing AI-based tools to support and assess collaboration in CS education. (PI: Gahgene Gweon, SNU)

Publications & Presentations

• An, M., Kwon, C., Lee, Y., Hur, J., Lee, D., S., Gweon, G., Stamper, J. (under review). Deriving Instructional Insights from Human–LLM Co-Evaluation of Student Collaboration in Data-Centric Programming, Submitted to SIGCSE TS 2026
• Lee, Y., Kwon, C., Seoh, S., Gweon, G., Stamper, J., Rosé, C. (2025) Capturing Collaborative Competency with GPT-4o and ENA, CSCL 2025 (Learn more)
• Wu, Z., Shi, J., Murray, R. C., Rosé, C., & San Andres, M. (2025). LLM Bazaar: A Service Design for Supporting Collaborative Learning with an LLM-Powered Multi-Party Collaboration Infrastructure, CSCL 2025
• Naik, A., Yin, J. R., Kamath, A., Ma, Q., Wu, S. T., Murray, R. C., Bogart, C., Sakr, M., Rose, C. P. (2025). Providing tailored reflection instructions in collaborative learning using large language models, British Journal of Educational Technology
• Xiao, R., Hou, X., Stamper, J. (2024). Exploring How Multiple Levels of GPT-Generated Programming Hints Support or Disappoint Novices, CHI EA 2024

---

Project 3 Development of Cognitive AI for Multi-Agent Systems Using LLM

This project investigates the use of LLMs in multi-agent systems, focusing on how agents with distinct roles can coordinate effectively. The research combines natural language understanding and graph-structured data to improve communication and reasoning among agents. Collaborating with Carnegie Mellon University, the project also develops human-AI interfaces and customized LLMs for domain-specific tasks. (PI: Bongwon Suh, SNU)

Publications & Presentations

• J Ryu, K Kim, D Heo, H Song, C Oh, B Suh (2025). Cinema Multiverse Lounge: Enhancing Film Appreciation via Multi-Agent Conversations SIGCHI, ACM, (Learn more)
• K Kim, H Jeon, J Ryu, B Suh (2024). Will LLMs Sink or Swim? Exploring Decision-Making Under Pressure, EMNLP 2024, (Learn more)
• K Kim, H Song, B Suh (2024). Self-Referential Review, SIGIR, ACM (Learn more)
• J You, B Suh (2024). Evaluating and Improving Value Judgments in AI, ICWSM, AAAI (Learn more)
• K Park, H Lim, J Lee, B Suh (2024). Enhancing Auto-Generated Baseball Highlights, SIGCHI, ACM (Learn more)
• W Hyun, I Lee, B Suh (2024). LEX-GNN: Label-Exploring Graph Neural Network, CIKM, ACM (Learn more)
• K Kim, B Suh (2024). SymphoNEI: Symphony of Node and Edge Representations, DASFAA (Learn more)
• K Kim, B Suh (2024). HopLearn: Multi-Hop Neighbors for GNNs, DASFAA (Learn more)

Participating Researchers