Research Symposium

BIO

Camden Crum is a Freshman Presidential Scholar at FSU majoring in Business Management. He is in the honors program and conducted his research on an AI project which he hopes will help him with integrating AI into business settings in future careers.

Open – Ended Survey Response Analysis Using LLMs

Abstract

This project pursued the goal of using a large language model (LLM) to effectively sort a representative sample of survey responses. We aimed to analyze the following questions: Under what conditions could using a LLM to analyze qualitative data be effective? How can we best leverage these models in qualitative research? How can we improve upon existing work at FSU using LLMs to classify open-ended survey data? This project is relevant in many different areas of work because AI is useful for making long and tedious processes more efficient. The goal of this is to lessen the number of hours humans spend analyzing data or performing tasks that AI can effectively accomplish within a matter of seconds. By using a Snowflake database and recently introduced features, we were able to create a pipeline that could handle over 2000 survey responses and sort them into categories that fit each response more consistently than if done by a disjoint group of individuals. In doing this, we made it easier to analyze large amounts of data in a shorter amount of time. The next steps in terms of using these results would be to present our work to other organizations on campus and fully integrate our pipeline into survey analysis at FSU. The results and findings of this research suggest that AI can be used to expedite qualitative coding work, giving organizations greater consistency with data analysis while saving time.

Keywords: AI, Computer Science, Surveys