SEONGLAND

Welcome

Seonglae Cho

Research Scientist

Research scientist focused on LLM safety, interpretability, and steering. Building tools to understand and control language model behavior.

Scholar GitHub LinkedIn Twitter

News

2026.05 CorrSteer accepted at ICML 2026
2026.05 ASG accepted at ICML 2026 AIWILD
2026.05 Tool Calling Probing accepted at ICML 2026 FAGEN Oral
2026.05 AgentRoom accepted at ICML 2026 FAGEN
2026.05 PaaT accepted at ICML 2026 FAGEN
2026.03 ControlRL accepted at ICLR 2026 Trustworthy AI
2026.03 Confidence Manifold accepted at ICLR 2026 Reliable Autonomy

Research Interests

LLM Safety Mechanistic Interpretability Sparse Autoencoders Steering Vectors Agent Summarization

Mindset

Philosophy Optimistic Nihilism
Consciousness Bundle Theory
Prediction Statistical Thinking , Techno Optimism
Future Technological Singularity , Culture War
Planning GTD , RICE Framework
Creation First Principles Thinking
Attitude Critical Thinking
Computing Extended Church-Turing Thesis