csspfs@cuhk.edu.hk
csspfs@cuhk.edu.hk
Note:This seminar has been cancelled. (Updated on 14 September 2026)
csspfs@cuhk.edu.hk
Agentic AI for Social Science Research is a four-day course on doing empirical social science by directing coding agents — building the system, not just chatting with a model. It moves from how a language model actually works, to the agent loop, to a hardened setup in Claude Code, to a full research pipeline. Read it here as a book: walk it section by section, search it, and copy any command straight out of the page.
Workshop outline
Day 1 (Aug 31): The Model — how an LLM works, and the prompt as a measurement instrument you validate.
Day 2 (Sep 1): Agents — the agent loop, tools, context, and the harness around a model — and how it breaks.
Day 3 (Sep 2): Claude Code & Codex — the hardened agent configured for real research, shown for both toolchains.
Day 4 (Sep 3): The Pipeline — Open Scholar Skills end to end, from idea to a verified, submission-ready paper.
Remarks
Maximum number of participants will be30.
Priority will be given to PhD students and early-career faculty.
About the Speaker
Dr. Yongjun Zhang is an associate professor in the Department of Sociology and the Institute for Advanced Computational Science at Stony Brook University, with affiliations in the AI Innovation Institute, the Center for the Study of Changing Systems of Power, and the Department of Asian and Asian American Studies. His work sits at the intersection of computational social science, AI, and sociology, with recent projects focused on how AI agents reshape the social organization of knowledge production. He is the developer of Open Scholar Skills, an open-source Claude Code plugin of specialist AI skills covering the end-toend social science research pipeline.
csspfs@cuhk.edu.hk
csspfs@cuhk.edu.hk
sociology@cuhk.edu.hk