At the University of Florida, a few thoughts on RAGs, research methods, and the inconvenient fact that choosing what goes into a knowledge system is already part of the argument.
So rather than asking students to trust the machine, we let them build one — sources, chunks, biases and all. They grow up so quickly.
Back home, or
into the footnotes?
The Event
- Conference
- Research in the Age of Artificial Intelligence
- Hosted by
- * University of Florida;
- Where & when
- University of Florida — 12 March, 2026
- Official Links
- * University of Florida
Research summary - Every Tool is an Argument: Teaching Research Methods through Bespoke RAG Curation
An AI system that answers questions using a selected collection of sources is often called retrieval-augmented generation, or RAG. Building one is more than a technical exercise. Like a museum curator, its creator chooses what to include, how to arrange it, and which paths through the material are easy to follow. A RAG system for classical mythology or civil rights history is only as useful as its fit with the research questions it is meant to answer.
In a university research methods assignment, students build their own RAG systems. They define a question, select sources, decide how to divide those sources into passages, document their choices, and test the answers their systems produce. Each decision shapes what the system can find and say. Source selection is never neutral; even dividing a text into searchable pieces is an act of interpretation.
Building the tool helps students see choices that commercial AI systems can make hard to spot: what counts as evidence, whose questions the system serves, and how information becomes accessible. The presentation offers implementation guidance, considers practical barriers, and draws on student examples to show how building a RAG system can deepen understanding of research methods.