Reliable AI Agents 분야의 전문가인 DGIST의 목지수 교수님을 모시고 아래와 같이 세미나를 개최하오니, 관심있는 많은 분들의 참여 부탁 드립니다.

*일시: 2월 4일(수) 오후 2시 00분

*장소: 반도체관 400118호

*연사: Prof. Jisoo Mok (DGIST)

*Talk Title: From Chatbots to Agents: Building Smarter, More Reliable AI Systems

Abstract

Large Language Models (LLMs) and their multimodal counterparts (MLLMs) are penetrating every facet of our daily lives, evolving from conversational chatbots to increasingly autonomous problem-solving agents. As these AI models take on more human-like roles and interact more deeply with individuals, questions of trustworthiness, reliability, and alignment with human values naturally emerge as central research challenges. This seminar explores research directions for systematically addressing these requirements. The first part introduces techniques for personalizing LLMs and MLLMs, highlighting how tailored AI agents can improve the usability and effectiveness while raising new challenges around robustness and evaluation. The second part focuses on making AI agents more factual and reliable: addressing the persistent problem of hallucination and improving retrieval-augmented generation (RAG) techniques that ground model outputs in external knowledge sources. The final part discusses open research problems in building AI systems that are not only more capable but also more transparent and controllable: generating human-like dialogues through sophisticated persona prompting, enabling open-ended creative and exploratory tasks, and designing secure and private multi-agentic systems that can reason, plan, and collaborate over long horizons.

Biography

Jisoo Mok is an Assistant Professor in the Department of Electrical Engineering & Computer Science at DGIST, where she leads the Language-driven Multimodal Intelligence (LaMI) Laboratory. Prior to joining DGIST, she received her Ph.D. from Seoul National University under the supervision of Prof. Sungroh Yoon and her B.S. from Caltech. During her Ph.D. studies, she collaborated extensively with global AI labs through internships at Google Research, Amazon Alexa AI, and NAVER AI Lab. Her current research focuses on building more reliable, trustworthy, and human-like AI agents.

*Host: 고종환 교수 (전자전기컴퓨터공학과)