Experimenting with LLM & Agentic AI for Social Science Research
EDU S590
Subject & Catalog Number
Course Information
Description
What might LLMs and agentic AI mean for social science research? This hands-on, exploratory module gives you a structured space to investigate that question together through practical experimentation.
As LLMs become increasingly embedded across the research process, students and researchers need more than general awareness of these technologies. They need opportunities to test what these tools can do, understand where they break down, and think carefully about what their use means for the quality, transparency, and integrity of scholarly work.
This course explores the use of LLMs and agentic AI across the research pipeline, including research design, instrument development, automated data collection, quantitative analysis, and qualitative coding. It is not a traditional methods course: students will not learn how to design an experiment or conduct interviews from scratch. Instead, the module focuses on how AI tools may reshape existing research practices, workflows, and decisions.
No prior experience with LLMs or AI is required, though students should have some familiarity with the research process. By the end of the module, students will have gained practical exposure to AI-supported research workflows, developed foundational skills for evaluating these tools, and formed a clearer view of how LLMs might be used responsibly in social science research.
Because these technologies are evolving quickly, the syllabus may be updated before the course begins.
Available for Harvard Cross Registration
By the end of the course, students should be able to:
1. Describe several ways LLMs and agentic AI are being used or proposed for use across the social science research process.
2. Experiment with AI-supported workflows for selected research tasks, such as literature exploration, instrument development, data collection, quantitative analysis, or qualitative coding.
3. Recognize common strengths, limitations, and risks of using LLMs and agentic AI in research contexts.
4. Practice basic techniques for designing, testing, documenting, and refining AI-supported research workflows, including the use of prompts, tools, structured outputs, and multi-step or agentic AI systems.
5. Reflect on the ethical, methodological, and scholarly implications of using LLMs in social science research, including issues such as transparency, validity, bias, accountability, and research integrity.
6. Develop an initial framework for deciding when and how AI tools may be useful, inappropriate, or in need of further scrutiny in their own academic or professional work.
As LLMs and agentic AI become increasingly common in knowledge work, students will benefit from learning how to engage with these tools thoughtfully, experimentally, and critically rather than treating them as either shortcuts or black boxes.
The course is especially relevant for students who expect to design research projects, collect or analyze data, conduct literature reviews, evaluate evidence, build research workflows, or communicate research findings in academic or professional settings.
think strategically
mindset
research methods
