UG Projects

Research Project Descriptions

This page contains descriptions of research projects for undergraduate researchers here. These projects are all related to generative AI, but they are all different. Some of them are more technical and some of them are more focused on social science questions. Some of them are more open-ended and some of them are more structured. I am happy to work with you to find a project that is a good fit for you. If you have any questions or interest in joining the group, please contact Dr. Katz.

Project 1: Think Aloud Design and Generative AI

Description

This project is about using generative AI to help analyze think aloud data. Think aloud is a method of user testing where a user is asked to verbalize their thoughts as they complete a task. This data is often analyzed by hand, which is time consuming and expensive. We are exploring the use of generative AI to help automate this process.

The specific data come from a study where participants were asked to complete a design task. They were given information from one of three perspectives and asked to design a solution. The data we are using is the think aloud data from this study.

Tasks

This project involves using generative AI to analyze think aloud data. The specific tasks are:

  1. Use generative AI to generate think aloud data. We are primarily interested in using pre-trained models to analyze the data. There is a lot of prompt engineering to test which prompts work best for the model.
  2. We are also interested in using speech-to-text models to generate transcripts of the audio data. This is a more traditional use of generative AI, but it is still an interesting problem.

Project 2: Public Perceptions of Generative AI + ______

Description

This project is about using social media data to understand public perceptions of generative AI. We are interested in understanding how the public perceives generative AI and how this perception changes over time. We are also interested in understanding how the public perceives generative AI in different contexts. For example, how do people perceive generative AI in the context of education? How do people perceive generative AI in the context of ethics?

Tasks

This project involves using social media data to understand public perceptions of generative AI. The specific tasks are:

  1. Use social media data to understand public perceptions of generative AI. We are primarily interested in using Twitter, YouTube, and Reddit data, but we are open to other social media platforms. It is best when there is an API that we can use to collect the data.

  2. We are also interested in using generative AI to analyze these social media data. This will involve some machine learning including embedding the data and using clustering algorithms to identify topics.

Project 3: Systems Thinking Assessment

Description

This project is about using generative AI to help assess systems thinking. Systems thinking is a way of thinking about the world that focuses on the relationships between parts of a system. It is a way of thinking that is important for solving complex problems. We are interested in using generative AI to help assess systems thinking. We have a dataset of student responses to a systems thinking assessment. We are interested in using generative AI to analyze this data based on a rubric in addition to simply identifying the topics students are discussing.

Tasks

This project involves using generative AI to help assess systems thinking. The specific tasks are:

  1. Use generative AI to analyze student responses to a systems thinking assessment. We are primarily interested in using pre-trained models to analyze the data. As with the other project, there is a lot of prompt engineering to test which prompts work best for the model.

  2. We are also interested in using the similar workflow as in project 2 to analyze the data. This will involve some machine learning including embedding the data and using clustering algorithms to identify topics.

Project 4: Applying Generative AI to Study Causal Beliefs

The problem motivating this project is that causal beliefs that people hold of sociotechnical systems inform their decision making and engagement with the world. If we want to understand how people make decisions, then it can be useful to understand their causal beliefs. However, causal beliefs are often difficult to identify and measure because common methods for identifying causal beliefs are expensive and time consuming.

This project aims to use generative AI to identify causal beliefs from text. The idea is that if we can get a generative AI to write a causal story about a sociotechnical system, then we can use the story to identify the causal beliefs of the author. Such soicotechnical systems can include the environment, education, healthcare,and AI.

Tasks

This project involves using generative AI to identify causal beliefs from text. The specific tasks are:

  1. Use generative AI to write a causal story about a sociotechnical system.
  2. Use the story to identify the causal beliefs of the author.

Project 4: Choose Your Own Adventure

Description

Possibly the most exciting project is the one that you choose! If you have an idea for a project that you would like to work on, please let me know. I am open to any ideas that you have. I am also happy to help you develop your idea into a project.