Jaesik Choi

Jaesik Choi is a faculty member in the Department of Computer Science at KAIST. His research focuses on artificial intelligence and machine learning, with an emphasis on areas such as feature attribution, text-to-image generation, and robust sequential conformal prediction. He is affiliated with the Kim Jaechul Graduate School of AI and leads the Statistical Artificial Intelligence Lab@KAIST. Choi has contributed to several accepted papers at major conferences such as ICML 2026, ACL 2026, and NeurIPS 2025. His work includes topics like LLM jailbreaks, Korean meteorology benchmarks, and adversarial robustness in deep neural networks.

Jaesik Choi is a Director of the Explainable Artificial Intelligence Center at KAIST, focusing on Explainable AI, Interpretability, Prediction, Time Series Analysis, and Relational Learning. His research explores methods to enhance the transparency and interpretability of artificial intelligence systems, particularly in complex domains such as time series forecasting and relational modeling. Choi's work includes developing techniques for attributing contributions of neural network units, improving anomaly detection in embedded systems, and advancing deep reinforcement learning for real-time applications. His publications span topics like secure intrusion detection, multi-robot task allocation, and spatio-temporal analysis for sports videos, reflecting a broad interest in AI with practical and theoretical implications.