Identifying Biases in VLMs' Zero-Shot Classifications of Race and Gender
An audit framework comparing how vision-language models label race and gender in face images against annotations from a demographically balanced sample of U.S. adults.
Silvia Téliz
I am a computational social scientist and PhD candidate at the Annenberg School for Communication, University of Pennsylvania. Drawing on computer vision and multimodal AI, I study how political images represent social identity, and how social biases are embedded in the technology we use to analyze visual data.
Education
- University of Pennsylvania PhD Communication, expected 2027
- University of Pennsylvania MA Communication, 2024
- Universidad Panamericana BA Communication, 2018
Affiliations
Contact
- Email silvia.teliz@asc.upenn.edu
- Elsewhere Google Scholar
- Bluesky
- GitHub
- ORCID
- CV (PDF) →
Featured publications
All publications →- 2026
Skin Tone as a Visual Cue in Political Campaigns: Evidence from Elections in Mexico
International Journal of Press/Politics (in press)
- 2026
Poder Prieto: Media, Colorism, and Cultural Activism in Mexico
Planet Digital (NYU Press)
Current research
All projects →Mechanisms of Social Bias in Large Language Models: Disentangling Linguistic Framing from Group Specification
An investigation of whether social bias in large language models is driven by the linguistic framing of a prompt or by the explicit specification of a social group.
Visual Cues as Political Communication: A Computational Look into Mexican Federal Electoral Campaigns
How skin tone and other visual cues in political advertising influence electoral outcomes in Mexico.
Recently
All news →- Sep 2026 “Skin Tone as a Visual Cue in Political Campaigns: Evidence from Elections in Mexico”, co-authored with my advisor, Sandra González-Bailón, is out now in the The International Journal of Press/Politics! You can read the open-access article here.
- Sep 2026 I had the honor of teaching a guest lecture on bias in AI systems to an amazing group of (400!) undergraduate students at Stony Brook University.
- Jul 2026 Our project “Identifying Biases in VLMs’ Zero-Shot Classifications of Race and Gender” was awarded $10,000 USD through the Annenberg School for Communication Internal Grant. I will be leading this project as lead researcher during the 2026-2027 academic year, collaborating with my colleagues from the Politics, Identities, and Communication Lab, Dr. Cristina Monzer and Dr. Deen Freelon.