Using Artificial Intelligence To Detect Teacher Candidates’ Funds Of Identity In Socioeconomically Diverse Urban Contexts

• In Event: AERA Poster Session Sunday 9:45 am
  In Poster Session: Inquiry in the Social Contexts of Education

Abstract

This study explores how artificial intelligence (AI) can automate detection of teacher candidates’ (n=372) funds of identity (FoI) across a multimodal dataset of essays, concept maps, and meme-oriented online discussions. Guided by the five-category FoI framework (Esteban-Guitart, 2016), we used a SentenceTransformer model for text and a vision-language model (CLIP) for images to calculate semantic similarity to a researcher-developed exemplar codebook. Maximum similarity scores per category were analyzed using ANOVA and Tukey HSD. Results reveal distinct modality effects: essays surfaced the richest identity signals, while meme discussions revealed subtle or latent FoI often missed in qualitative analysis. Findings demonstrate that AI complements human coding by scaling analysis and uncovering diverse identity expressions, advancing equity-centered teacher preparation in urban educational contexts.

Authors

• Yan Chen, University of Nevada - Las Vegas

• Nick V. Flor, University of New Mexico

• Moises Esteban Guitart, University of Girona