Goal: The purpose of this study is to evaluate how Large Language Models or LLMs (e.g. ChatGPT, Gemini) can impact through its answers asked by users, particularly how question framing can result into opinion shaping and polarized perception. For this research, we want to conduct a study in two stages; in stage one we want to generate queries for LLMs. The other stage will be to propose intervention to mitigate it. Since studies have found that polarization caused by human-AI interaction can have a negative impact on both an individual and societal level, it is very important to identify and resolve it specifically for AI based tools which are being used on daily basis. We want to propose intervention while interacting with LLMs to know if it helps mitigating the bias caused by LLM models while framing queries.
All eligible participants will receive $10 upon completion of the task. Please fill out the link to the form provided in you are interested in participating in the study. We will reach out to you personally with details for further action within 48 hours.
Location: Remote
Time: Around 40 minutes
Eligibility: UMBC student, at least 18 years old, fluent in English and resides in the United States
Md Nazmus Sakib
PhD Computer Science Student