I gave a talk at the Korea & MSRA Societal AI Workshop, held on May 30 at Sungkyunkwan University (Seoul Campus). I shared our recent work using the Azure API, including Reasoning Capabilities of LLM, MC-LARC, and GIF-ARC—all of which were done with our students! I also enjoyed the panel discussion on the future of societal AI and AGI; it was great to hear such diverse perspectives. Lastly, I was pleasantly surprised to meet an audience member who came specifically to hear my talk on ARC—an unexpected encounter with a fellow ARC enthusiast!

  • 🗓️ Date: May 30 (Fri), 1:00–6:00 PM
  • đź•’ My Talk: 1:40–2:10 PM
  • 📍 Venue: Room 33B101, Business Hall, Sungkyunkwan University (Seoul Campus)
  • đź”— Workshop Info

Featured Projects:

  1. LLM-on-ARC
  2. MC-LARC
  3. GIF-ARC

Title: Diagnose, Debug, and Enhance: A Three-Stage Framework for AI Compositional Reasoning

Abstract: Large language models still lag behind humans on abstract reasoning tasks—especially in composing multiple operations—despite impressive pattern-matching abilities. In this talk, I introduce a three-stage framework to tackle this bottleneck. First, we diagnose core weaknesses via a process-centric evaluation of ARC tasks, measuring logical coherence, compositionality, and productivity. Second, we debug the compositionality failure by converting ARC problems into multiple-choice questions (MC-LARC), revealing precise failure modes in the “Understand” and “Apply” stages. Finally, we enhance compositional reasoning with GIFARC, which injects human-intuitive analogies extracted from GIFs to guide models through “analogy → grid composition” steps. Together, these steps—diagnose → debug → enhance—form a cohesive pipeline toward more human-like, explainable, and aligned AI reasoning.

Picture Picture