AI for Survey Researchers: A Three-Workshop Online Series

Large language models are now part of everyday research, but most of us meet them through chatbot interfaces that hide how they work, where our data goes and what decisions are being made on our behalf. This three-part online series from the National Centre for Research Methods (NCRM), led by Dr Paulo Serôdio, takes survey researchers from how LLMs generate text right through to agentic AI and autonomous research workflows.

Each half-day workshop runs online from 9:30am to 1:30pm (UK time) and combines explanation with live demonstrations and practical exercises built around survey research scenarios. If you have been experimenting with tools like ChatGPT in your own work (see our blog on ChatGPT in academia), this is a chance to understand what is going on under the bonnet.

The three workshops

Workshop 1: How large language models work (Monday 23 November 2026)

• Tokens, training, alignment and inference
• Open-weights versus closed-weights models, and data security
• Reproducibility challenges and the limits of chatbot interfaces for research

Workshop 2: Context engineering (Monday 30 November 2026)

• Prompt design and optimisation
• Retrieval-augmented generation (RAG), API-based workflows and batch processing
• Memory, tool-calling, MCP servers and evaluation engineering

Workshop 3: Agentic AI and harness engineering (Monday 7 December 2026)

• The agentic AI ecosystem, from IDE-native agents to orchestration tools
• Harness engineering, SDKs, memory and token economics
• Oversight, auditability and research transparency

For a taste of where agentic AI is heading in our own field, listen to our podcast on Agentic AI and the Future of Dementia Research.

What you will learn

By the end of the series participants will be able to:

• Explain how LLMs generate text, and what model architecture, training and alignment mean for research practice
• Tell open-weights and closed-weights models apart and weigh up their data governance implications
• Optimise prompts and build evaluation pipelines to check LLM outputs
• Make structured API calls, manage parameters and use retrieval-augmented generation where it fits
• Map the agentic AI ecosystem and explain how platforms orchestrate memory, tools and context
• Design human-in-the-loop safeguards and audit protocols for agentic research workflows

Who should attend

The series is for survey researchers, methodologists and quantitative social scientists in academia and government who use, or are thinking about using, large language models in their research. It is pitched at entry level and open to every career stage, from doctoral students to senior investigators. If you are newer to survey work, our Methods Matter podcast on surveys and questionnaires is a good primer.

No programming experience or software installation is needed for Workshop 1. Workshops 2 and 3 build on the concepts from Workshop 1, and Workshop 3 is easier if you are comfortable reading code, though you will not need to write any. Having API access to a commercial LLM provider will help for Workshops 2 and 3, and setup guidance is sent in advance.

NCRM recommends attending all three, but you are welcome to join individual workshops.

Cost

The fee for the series is £150, and it stays the same however many of the workshops you attend. A full refund is available if you cancel up to two weeks before the course; no refunds are given after that.

Register

Organised by the National Centre for Research Methods, University of Southampton. Full details on the NCRM course page.