Recent Progress and Future Outlook of Reasoning in Large Language Models

Large language models (LLMs) have made remarkable progress in recent years, driven in large part by advances in their reasoning capabilities. In this seminar, I will introduce research on LLM reasoning, including work I have personally been involved in. Topics covered include test-time reasoning improvements stemming from Chain-of-Thought (CoT), graph-based analyses of reasoning-model trajectories, capability extensions through reinforcement learning and model-architecture refinement, and reasoning in the era of Agents. I will review progress to date and discuss future directions and open challenges.