Generative AI and Transness

Generative AI models are trained to identify patterns in existing data in order to generate new content, including text, images, and videos. Generative AI has taken off recently, particularly with the release of ChatGPT. ChatGPT, developed by OpenAI, is a large language model-based chatbot where users can enter a prompt and receive a formatted response. One of the major uses of LLM-based chatbots is information seeking. However, many have expressed concerns that, like other machine learning models, LLM-based chatbots return false, misleading, or biased information. Given these LLM-based chatbots are trained on vast amounts of historical web data, we are seeking to understand how the political landscape might affect the outcomes of chatbots around politicized identity issues. In this project, we are specifically examining how transgender and non-binary identities might be treated by chatbots as they are trained on politicized content over time.

Researchers

Morgan Klaus Scheuerman, Dipto Das, Katy Weathington, Adrian Petterson, Michael Ann DeVito, Jed Brubaker