Sunghee Yun (He/Him)

Co-Founder & CTO @ Erudio Bio, Inc., Co-Founder & Leader & Chair of Silicon Valley AI Nexus (K-PAI Nexus), Advisor @ Korean American Semiconductor Professional Alliance (KASPA), CGO / Global Managing Partner @ LULUMEDIC, KFAS-Salzburg Global Leadership Initiative Fellow @ Salzburg Global Seminar, AI-Korean Medicine Integration Initiative Task Force Member @ The Association of Korean Medicine, Director of AI Semiconductor @ K-BioX, Visiting Professor @ Sogang University, Advisory Professor @ DGIST, Global Advisory Board Member @ Innovative Future Brain-Inspired Intelligence System Semiconductor of Sogang University, and Network Expert @ Gerson Lehrman Group

Abstract

Artificial Intelligence (AI) has moved from a decades-long research arc into a period where capability that once took decades now arrives in months — from Deep Learning (DL) and the Transformer to Large Language Models (LLMs) and, most recently, goal-pursuing “agentic” systems backed by unprecedented capital and compute. This talk opens by mapping that trajectory and the agentic stack now reshaping how knowledge work is done, then turns to its real subject – what this acceleration means for researchers. Rather than asking the familiar binary — will AI replace researchers, or merely assist them? — I argue that the question itself flattens a high-dimensional shift onto a single axis, and offer a more useful frame – cognitive catalysis.

Under this view, AI now touches every stage of inquiry — hypothesis, design, data, analysis, and writing — not as a system that produces answers for humans to rubber-stamp, but as a catalyst that accelerates the passage from knowing to understanding while leaving the desire to understand irreducibly human. The talk examines how this reshapes research design, data analysis, and knowledge production, and confronts the risks the same power introduces – black-box results, p-hacking at scale, and a flood of plausible-but-unverified output. It gives particular attention to research ethics after generative AI — authorship and accountability, the collision of the reproducibility crisis with hard-to-trace results, and emerging responses such as AI-use disclosure and data-lineage tracking — pressing the distinction between work a researcher genuinely understood and text they merely generated.

The closing movement contrasts academia and industry as two games driven by the same catalyst but opposite incentives — novelty, peer recognition, and open inquiry versus deployment, product, and speed-to-market — and warns of a danger they share – optimizing the metric instead of the knowledge. Drawing on the “conference analogy” — why brilliant people still cross the world to convene when every paper is already online — it argues that AI can catalyze individual cognition at a speed no gathering can match, yet cannot reproduce the network effect of many expert minds in one room. The talk ends on an exquisite irony for a room full of future scholars – the more powerful the catalyst becomes, the more indispensable genuine human expertise, curiosity, and judgment become — because understanding cannot be automated, only catalyzed, and the answer is not found but created.