The Hidden Curriculum of Generative AI (New Publication!)

Uncovering the Hidden Curriculum in Generative AI

I’m excited to share a new publication co-authored with Dr. Marie Heath in the Journal of Teacher Education: “Uncovering the Hidden Curriculum in Generative AI: A Reflective Technology Audit for Teacher Educators”.

The article explores how generative AI tools—particularly large language models—may subtly reinforce societal inequities in education. Using a methodology we call a reflective technology audit, we examined how these systems score and respond to student writing when prompts are paired with student descriptors like race, school type, or music preference. Students described as attending “inner-city schools” or liking rap music received lower scores. Perhaps even more concerning, feedback to Black and Hispanic students tended to be more authoritative in tone, echoing the power dynamics often found in schools. It’s relatively simple to not use GenAI tools for grading; but this initial result demonstrates that the very language LLMs use when personalizing responses may perpetuate inequities.

These findings connect to the concept of the hidden curriculum—the implicit messages, values, and expectations embedded in educational tools and practices. While not part of formal instruction, the hidden curriculum shapes students’ experiences and identities. If AI systems consistently use more controlling language with certain groups of students, they will subtly reinforce messages about who deserves autonomy and whose voices require discipline. That’s not just a technical issue; it’s a pedagogical and ethical one.