Examining Trust in AI: Analyzing Teacher Candidates’ Discourse in Response to Automated Feedback

• In Event: AERA Roundtable Session Saturday 3:45 pm - Gold 3
  In Roundtable Session: Perspectives and Practices of AI Use Across Educational and Global Contexts

Abstract

As AI becomes more prevalent in education, understanding how teachers construct and communicate trust is essential for meaningful adoption. This study explores how teacher candidates interacted with the CERMI dashboard, an AI-generated feedback tool, and expressed trust through verbal and non-verbal cues. Grounded in Speech Act Theory and research on human–machine trust, the analysis centers on transparency, reliability, and fairness. Transparency—especially explainability and verifiability—was key to trust-building, as participants tried to understand how feedback was generated. They also questioned the tool’s
reliability, particularly its accuracy and ability to reflect classroom complexity. Fairness concerns were mentioned less often. The study offers design implications for AI tools that foster trust by supporting teachers’ judgment, reflection, and instructional decision-making.

Authors

• Kayla Michelle Reist, University of Virginia

• Peter A. Youngs, University of Virginia