Prompt Engineering for Sociology & Anthropology
Sociologists and anthropologists run into LLMs at exactly the point where their work is most tedious: coding thousands of interview transcripts, classifying open-ended survey responses, and searching through fieldnotes. The research methods literature has already begun absorbing these tools into its standards of practice.
Where this is showing up in Sociology & Anthropology
- ATLAS.ti Intentional AI Coding (GPT-backed, with an AI Privacy Mode) lets researchers state a research goal upfront and generate codes tailored to it rather than generic descriptive tags.
- MAXQDA AI Assist adds AI coding, code-recommendation, chat-with-data, and multilingual summarization, with GDPR-compliant zero-retention processing.
- "Leveraging large language models for thematic analysis" (AI & Society, 2025) found GPT-4o hit κ=0.61–0.65 on multi-label thematic categorization and κ=0.91–0.95 on sentiment, proposing a dual-role human-LLM framework with reusable prompt templates.
- NVivo, Dedoose, and Delve have all shipped AI features in the last year — meaning qualitative software used across sociology, anthropology, and public health now assumes LLMs are part of the pipeline.
Projects you could build in this course
- An interview-transcript coding assistant with a human-in-the-loop validator and inter-rater reliability scoring
- An open-ended survey-response classifier that reports confidence and cites example passages
- A RAG assistant over fieldnotes or an ethnographic archive for longitudinal research