This seminar explores how to integrate large language models (LLMs) into research on human attitudes and behavior. It provides an introduction to LLMs like ChatGPT, Claude, Gemini, and Llama, discusses the critical concept of prompt engineering, and teaches you how to use LLMs in three common use cases:

1. Coding open-ended survey responses
2. Generating simulated or synthetic samples, and
3. As treatments in randomized experiments.

The course is designed to give you all the tools you would need to start using these LLMs in these ways in your own work. You’ll gain a foundational understanding of what these generative AI tools are, the most efficient ways to interact with them, and detailed knowledge of how to apply LLMs in these three common cases.

This seminar includes a set of exercises and supplemental tutorials to help you apply the skills from the seminar yourself. By working through these applications, you will gain the ability to really use LLMs in your own projects.

As a foundation for the course, you will first receive an accessible introduction to the history, structure, and nature of large language models (LLMs). We will then turn to the differences between LLMs of different types (open and closed source LLMs, LLMs provided by different companies, etc.).

A significant amount of seminar time will be devoted to “prompt engineering” or principles of interacting with LLMs. We will contrast this with fine-tuning (which we will only discuss at a conceptual level) and cover different approaches and methods to prompting. You will spend time exploring methods of prompting with LLMs.

The course will provide detailed demonstrations of how to interact with LLMs in public-facing interfaces (like ChatGPT), online developer platforms, and through APIs in R and RStudio. You will be given time to work through exercises and gain experience working with these interfaces as a part of the seminar.

The final part of the course covers three common use cases: coding open-ended survey responses, generating simulated or synthetic samples, and as treatments in randomized experiments. Of these three, we will devote more time to synthetic samples and treatments than on coding of open-ended texts.