This repository is a practical guide to using large language models (LLMs) in social science research. It includes hands-on examples and implementation strategies across three key applications:
Text Annotation: Use LLMs to label unstructured data—such as search queries, AI-user conversations, or social media postsÂ
Synthetic Data Generation: Generate realistic interaction data to evaluate model safety or simulate rare scenarios that are difficult to observe in real-world datasets
Human Simulation: Simulate human responses for tasks such as survey completion or conjoint analysis