Closing: Rigor at AI Speed

Social science was built on patient, careful work. AI changes the speed without (necessarily) changing the rigor. You leave this course knowing how to use both. The pattern: the methodologically rigorous use of AI in research is not blanket automation. It is delegating the mechanical work (initial coding, transcription, literature scanning) to AI under careful human validation, then doing the interpretation yourself. The researchers using AI well are running studies at unprecedented scale. The ones using it badly are publishing papers with fabricated quotes and miscoded themes. The weXare thesis applies cleanly: AI handles volume, humans handle meaning. AI codes 5,000 interviews into themes. You read the themes and interpret what they mean for your theory. AI scans 500 papers for relevance. You read the 30 that matter. The judgment is yours. The volume is no longer the bottleneck. **Five takeaways to keep:** 1. Validate AI coding against human coding on a sample. Inter-coder reliability is the new gold standard. 2. Hallucination is a methodology threat. Verify every AI-generated citation and quote. 3. AI as interviewer changes the data. Decide deliberately whether to use it. 4. The ethnographer's contribution does not scale through AI. Some work stays human. 5. Disclose AI use to participants and in publications. Methodological transparency is non-negotiable. **What is next:** Take [Human-in-the-Loop Design](/en/learn/human-in-the-loop) for the design discipline. Take [Prompt Engineering](/en/learn/prompt-engineering) to get cleaner AI outputs for your research. Take [AI Governance and Compliance](/en/learn/ai-governance-and-compliance) for IRB and ethics frameworks. Now go run the study you could not run before, without compromising the study you would have run.