Document Type
Conference Proceeding
Publication Date
5-1-2026
Abstract
AI can summarize how people feel.
But who decides what counts as “feeling”?
The lecture series brings in scholars to engage the wider university community in meaningful conversations. I am using that space to ask a question at the center of my work:
When AI tells us what society thinks, is it actually telling us the truth, or just what it has been trained to recognize as truth?
My talk, “What AI Can Tell Us About Society: Insights from Social Media Data,” leans directly into that tension.
I will break down how sentiment analysis works in practice, share insights from my undergraduate research team at DePauw University, and walk through our Juneteenth tweet project. We manually annotated thousands of tweets to teach the model what nuance actually looks like.
Because it turns out:
Joy, grief, resistance, and critique can all exist in the same sentence.
And most models are not built to hold that.
We will not just talk about AI. We will do it together.
Recommended Citation
Eremionkhale, Amy Ehinomen, "What AI Can Tell Us About Society: Insights from Social Media Data" (2026). Business Analytics Faculty Publications. 10, Scholarly and Creative Work from DePauw University.
https://scholarship.depauw.edu/bus_facpubs/10
Comments
Invited presentation for the Creason Lecture at Indiana State University's College of Arts and Sciences.