My Research Vision

Core Philosophy

My research is driven by a fundamental belief: technology should amplify human capabilities rather than replace human judgment. I design interactive systems that leverage the complementary strengths of humans and AI, creating experiences where both can contribute their unique strengths.

Research Trajectory

Learnersourcing (2010-2015)

My PhD thesis introduced "learnersourcing"—the idea that learners themselves can generate valuable educational content through their collective activity. This work established my interest in systems that scale through user participation rather than just automation.

Interaction at Scale (2015-2022)

Building on learnersourcing, I expanded my research to "Interaction at Scale"—designing systems that support interaction at massive scale, powered by crowdsourcing, data-driven techniques, and AI. This era bridged my early work on learnersourcing with my current focus on interaction-centric AI, exploring how to leverage collective human activity and data to create more powerful interactive systems.

Interaction-Centric AI (2022-present)

I've developed a broader vision of "Interaction-Centric AI"—designing AI systems where the interaction itself is the primary design consideration, not just the AI model. This means creating interfaces that enable users to understand, control, and collaborate with AI effectively, building on the foundation of scalable interaction systems.

Key Research Themes

1. Human-AI Interaction

How can we make AI systems that truly understand and augment human creativity, learning, and decision-making?

Impact: Moving beyond "chat with AI" to designing purposeful, controllable, and creative human-AI partnerships.

• Intent Alignment

  • EvalLM: Interactive evaluation of LLM prompts on user-defined criteria (CHI 2024)
  • CUPID: Evaluating personalized alignment from interactions (COLM 2025)

• AI-Powered Creative Tools

  • CreativeConnect: Supporting graphic design ideation with generative AI (CHI 2024)
  • GenQuery: Visual search with generative models (CHI 2024)
  • Demystifying tacit knowledge in creative domains (CHI 2024, Honorable Mention)

• Multi-Agent Conversational Systems

  • ChoiceMates: Multi-agent conversations for unfamiliar decision-making (in submission)
  • Proxona: Supporting Creators' Sensemaking and Ideation with LLM-Powered Audience Personas (CHI 2025)

• Structured Interaction Paradigms

  • Cells, Generators, and Lenses: Object-oriented interaction with LLMs (UIST 2023)
  • IdeaBlocks: Expressing exploratory intents with generative AI (UIST 2025 Demo)

2. Designing AI-powered Learning Systems

Transforming how people learn with intelligent, adaptive, and scalable educational technologies.

Impact: Democratizing education through AI that adapts to individual needs and scales expert knowledge.

• Intelligent Tutoring & Assessment

  • TeachTune: Reviewing pedagogical agents with simulated students (CHI 2025)
  • Diagnosing students' cognitive skills in math problem-solving with LLMs (in submission)
  • Teach AI How to Code: LLMs as teachable agents for programming education (CHI 2024, Honorable Mention)

• Enhanced Learning from Videos

  • VideoMix: Aggregating how-to videos for task-oriented learning (IUI 2025)
  • VIVID: Collaborative authoring of dialogues from lecture videos (CHI 2024)
  • Beyond Instructions: Taxonomy of information in how-to videos (CHI 2023)

• Learnersourcing & Crowdsourced Education

  • CodeTree: Learnersourcing subgoal hierarchies in code examples (CSCW 2024)
  • KUIZ: Encouraging modular learnersourcing of MCQs through LLM interventions (L@S 2024 Workshop)

• Diverse Learner Needs

  • LearnerVoice: Dataset of non-native English learners' speech (Interspeech 2024, ISCA Best Student Paper Shortlist)
  • How older adults use online videos for learning (CHI 2023)
  • DAPIE: Explanatory dialogues for children's questions (CHI 2023)

3. Evaluating & Benchmarking Human Values in AI

Building rigorous methods to understand what AI can and cannot do.

Impact: Ensuring AI systems work reliably across diverse contexts and users.

• Novel Evaluation Frameworks

  • FLASK: Fine-grained LLM evaluation based on alignment skill sets (ICLR 2024 Spotlight)
  • Mind the Blind Spots: Focus-level evaluation framework for LLM reviews (EMNLP 2025 Oral)
  • BloomIntent: Automating search evaluation with LLM-generated user intents (UIST 2025)

• New Datasets & Benchmarks

  • PANORAMA: Patent examination evaluation trails (NeurIPS 2025)
  • LearnerVoice: Non-native English speech dataset (Interspeech 2024, ISCA Best Student Paper Shortlist)

My Approach

I believe in:

  • Building real systems: My research involves building and deploying systems that real people use
  • Rigorous evaluation: Using both quantitative and qualitative methods to understand impact
  • Interdisciplinary collaboration: Working across HCI, AI, education, and social computing
  • Practical impact: Ensuring research addresses real-world problems

Awards & Recognition

  • Dec 2025: KAIST College of Engineering Impact Research Award
  • Nov 2025: Grand Prize for the KAIST Global Research Cooperation Award
  • Mar 2024: Outstanding Ethics Professor Award, KAIST Graduate Students Association
  • Dec 2023: Excellence Prize for the KAIST Global Research Cooperation Award
  • Dec 2022: KIISE/IEEE-CS Young Computer Researcher Award (젋은정보과학자상)
  • Dec 2022: Creative Education Award, KAIST
  • Nov 2022: NeurIPS 2022 Keynote: "Interaction-Centric AI"
  • Nov 2021: HCOMP 2021 Keynote: "Crowd-Powered Interactive Systems"
  • Jun 2020: Kyu-Young Whang School of Computing Career Award
  • Mar 2020: Grand Prize for Creative Teaching, KAIST
  • Feb 2020: Songam Distinguished Research Award, KAIST
  • Dec 2019: ACM Excellence in Service Award
  • Feb 2019: Excellence in Teaching Award, KAIST
  • Aug 2015: Brown Fellowship - Brown Institute, Stanford University
  • May 2012: 2nd Place, Student Research Competition - CHI 2013

Research Impact

Publications

  • Published 100+ papers in top-tier venues (CHI, CSCW, UIST, IUI, Learning at Scale, etc.)
  • 20+ paper awards and recognitions
  • Research spans multiple domains: HCI, AI, education technology, and social computing

Real-World Applications

  • SkillBench: Co-founded a startup optimizing AI deployment for productivity and workforce development
  • Ringle: Led the product development of AI-powered English proficiency diagnosis
  • Research translated to industry collaborations and technology transfer
  • Systems deployed and used by real users in educational and professional contexts