Date of Award

Spring 5-1-2026

Document Type

Honors Project

Degree Name

Bachelor of Science

Department

Computer Science

Department Chair or Program Director

Marshall, Andrew

First Advisor

Davies, Stephen

Second Advisor

Anewalt, Karen

Third Advisor

Goldthorp, Jeffrey

Major or Concentration

Computer Science

Abstract

The quality of political discussions occurring on online platforms or social media sites has been deemed quite poor. To address this issue, I investigated whether a Large Language Model (LLM) can be used to promote civil and productive political discussions by identifying and responding to unproductive dialogue. I fine-tuned an existing LLM to detect elements of problematic dialogue, namely misinformation, misrepresentation of sources, logical fallacies, bias, and toxic language, and then respond in a corrective yet non-confrontational manner. The resulting model is referred to as FroBot and was evaluated through an experiment in which a human participant was placed in a chatroom with one FroBot and two other bots I developed. The two additional bots were CoolBot, which mimics a typical social media user, and HotBot, which intentionally produces unproductive dialogue. The results of this experiment, in terms of chat transcripts and post-surveys, indicate that FroBot was generally effective at identifying and moderating unproductive dialogue. However, CoolBot was also found to exhibit similar corrective behaviors, without being explicitly designed to do so. I speculate as to possible reasons for this unexpected behavior, including that CoolBot is possibly being influenced by FroBot. Overall, this project’s findings demonstrate the viability of using LLMs for improving political discussions in online settings.

Download

Share

COinS