Human-centered AI for teacher educators: Designing professional learning for critical AI literacy

Synthesis: Baran, Dilek, Ziba, and Xiao (2026) use a design-based research (DBR) approach, funded across 2023–2025 at a large Midwestern US research university, to examine how principles of Human-Centered AI (HCAI) and critical AI Literacy can inform the design of professional learning resources for teacher educators. Across three DBR cycles of needs analysis, prototyping, and refinement with seven teacher educators, they identified three convergent design needs: (a) improving instructional efficiency and effectiveness while preserving professional judgment, (b) modeling responsible and ethical AI integration, and (c) reforming Professional Development pedagogies with AI. These findings were translated into a five-module curriculum operationalizing critical AI Literacy competencies (foundational AI knowledge; Ethics and algorithmic bias; pedagogical integration; implementation, guidelines, and policy; and human-centered AI in education) through seven HCAI-informed activities, including educator-in-the-loop tasks that strengthen professional judgment, transparency, and Equity-oriented decision-making. The study reframes AI literacy as a design practice and positions teacher educators as designers and ethical stewards of AI integration.

Key Findings

Design-based research across three cycles. Cycle 1 conducted ~30-minute semi-structured needs-analysis interviews with five teacher educators (Max, Claire, Ella, Sandy, Vera); Cycle 2 prototyped online modules through an in-person co-discovery workshop with teacher educators and K-12 practitioners (including a school district running a district-wide AI initiative) plus six co-discovery sessions with two teacher educators (Kate, Iris); Cycle 3 refined the curriculum based on ~60-minute walkthrough interviews with the same two educators and an alignment matrix linking each activity to HCAI principles and critical AI literacy competencies.

Sample and context. Seven teacher educators spanning science, music, Edtech Platform, literacy, and social-studies education — tenured/tenure-track faculty, teaching professors, and a teaching assistant — participated across the cycles at a teacher-preparation program that had no formal, program-wide AI integration. Prolonged engagement spanned 13 months (over one academic year).

Three convergent design needs. Analysis yielded: (a) improving instructional efficiency and effectiveness while preserving professional judgment, (b) modeling responsible and ethical AI integration, and (c) reforming teacher-education pedagogies with AI — treated as the empirical anchor for all Curriculum Design.

Efficiency as expanding choices, not just timesaving. Teacher educators used AI as a thinking partner and tutor (generating examples, scenarios, and drafts), but the shared logic was that AI expanded the range of options while the decision to adopt, adapt, or discard remained with the educator — pointing to the HCAI principles of human-in-the-loop decision-making and educator Learner Agency.

Ethics as enacted practice. Educators treated ethics not as a topic but as a practice modeled in real time — transparent AI use, Scaffolding ethical decisions, creating classroom norms, and fostering reflective engagement — distributed across many small, repeated instructional decisions rather than a single curricular moment.

Reforming Assessment and pedagogy. Educators called for AI-resistant, process-based assessments prioritizing justification and reflection over AI-generated output, replacing summative essays with smaller scaffolded tasks that invite critical engagement with AI at multiple stages.

Five-module curriculum and seven activities. Findings were operationalized into five online modules (foundational AI knowledge; ethics and algorithmic bias; pedagogical integration; implementation, guidelines, and policy; human-centered AI in education) delivered through seven HCAI-informed activities, including educator-in-the-loop and student-in-the-loop tasks that foreground professional judgment, transparency, explainability, value-sensitive design, and equity-oriented decision-making.

Teacher educators as pivotal nexus. Teacher educators shape how preservice teachers first understand, evaluate, and integrate AI, yet rarely receive systematic preparation to lead this work; the study reframes AI literacy as a design practice and positions teacher educators as designers and ethical stewards of AI integration — directly supporting the knowledge base's Professional Development and Teaching concepts.

What this means for practice

• Instructors. Position AI as a thinking partner that expands options while leaving the adopt–adapt–discard decision with the educator, rather than a source of finished lesson plans or answers.
• Faculty developers. Model ethics as enacted practice — transparent AI use and real-time Scaffolding of ethical decisions — and design educator-in-the-loop activities that foreground professional judgment, transparency, explainability and equity-oriented decision-making.
• Instructors. Redesign Assessment toward process-based, AI-resistant tasks that ask students to justify reasoning and reflect, replacing summative essays with smaller scaffolded tasks that engage AI at multiple stages.
• Administrators. Commit program-wide resources to AI professional learning; the study site had no formal program-wide AI integration, leaving even experienced teacher educators without systematic preparation.

Limitations

• Design-based research with only seven teacher educators at a single Midwestern US teacher-preparation program; DBR is intended to develop design knowledge, not to establish generalizable effects.
• Participation was purposive and required "some experience with or interest in AI integration," so the sample skews toward already-motivated educators rather than the full faculty population.
• Evidence comes from interviews and design activities across three cycles over 13 months; there was no comparison group, no pre/post outcome measure and no student-level learning data.
• The program had no formal, program-wide AI integration at the time, so the three identified design needs are anchored to one institutional context and may not transfer to programs with existing AI policies.

Citation

Baran, E., Dilek, M., Ziba, M., & Xiao, X. (2026). Human-centered AI for teacher educators: Designing professional learning for critical AI literacy. Computers and Education Open, 100399. https://doi.org/10.1016/j.caeo.2026.100399