Using LLMs to programmatically extract and curate research data [ResBaz]

Spinning raw data into analysis-ready gold often takes far more time than anticipated. These steps don’t often show up in the methods but are critical for robust research results. Whether you're working with messy survey responses, archival documents, or image collections, transforming unstructured material into clean, structured variables is painstaking work that manual methods handle poorly and traditional programming approaches struggle to scale. Large language models change this. Used programmatically, they can extract structured features from text, interpret images, and produce consistent, usable datasets with more flexibility than rule-based approaches. In this hands-on two-hour workshop, you'll work with real New Zealand text and image data to extract meaningful features using free, cloud-based tools. You'll leave with reusable code you can adapt to your own research data, whatever your discipline. The goal is to demystify programmatic LLM use and give you a practical foundation you can build on immediately. Prerequisites: Basic Python programming experience will greatly assist in participation of this workshop. Set up: As we'll will be using free Google tools, you will be required to use a Google account to participate.