An AI Faculty Fellowship - A Model for HigherEd Faculty AI Professional Development

Framework

Our university wide AI Fellowship is anchored in social constructivism and social cognitive theory. Faculty learning is structured as collaborative knowledge building in authentic, discipline-specific contexts (Vygotsky, 1978). Monthly meetings provide modeling, guided practice, and feedback to strengthen faculty fellow’s AI-teaching self-efficacy and drawing on Bandura’s four sources (mastery experience, vicarious experience, social persuasion, affective states) and triadic reciprocal determinism linking personal beliefs, environment, and behavior (Bandura, 1986, 1997). Practically, fellows set specific goals, pilot AI-supported strategies, and reflect on evidence of learning from their peers; which are conditions shown to enhance efficacy and transfer (Tschannen-Moran & Woolfolk Hoy, 2001).
Operationally, the Fellowship functions as a community of practice with distributed leadership. Fellows and AI Partners establish a joint enterprise (shared goals and norms), sustain mutual engagement (monthly co-planning and peer feedback), and build a shared repertoire (adoption guides, slide decks) that travels across schools (Lave & Wenger, 1991; Wenger, 1998). Leadership work deliberately spans people, tools, and routines through a train-the-trainer model where fellows demonstrate, curate, and coach colleagues inside local contexts (Spillane, 2006). This configuration leverages local credibility and contextual fit, both associated with stronger adoption of innovations in schools.
Our professional development blends design-based research and improvement science. Iterations follow a repeating structure—goal → plan/pilot → prepare PD artifacts → refine—consistent with design experiments that generate usable theory in naturalistic settings (Brown, 1992; Design-Based Research Collective, 2003) and with networked improvement communities that employ tight problem definitions, working theories of change, and disciplined cycles to reduce variation in practice (Bryk, Gomez, Grunow, & LeMahieu, 2015). In our case, cycles explicitly target AI-integrated tasks (e.g., ethics, feedback workflows, assessment and assignment modifications) and the social routines that support them (e.g., peer demos, micro-teaching).
Adoption and course redesign are guided by Diffusion of Innovations and TPACK. Fellows act as change agents who attend to perceived relative advantage, compatibility with local syllabi, trialability in low-stakes settings, observability via demos, and manageable complexity—attributes associated with faster diffusion (Rogers, 2003). Concurrently, planning centers technological–pedagogical–content knowledge interactions to avoid tool-first decisions and instead reason from disciplinary aims, task structures, and representations (Mishra & Koehler, 2006). We document shifts in lesson plans and exemplars that show tighter T–P–C alignment over time.
Equity and ethics operate as guiding principles under a critical digital pedagogy lens. PD activities surface algorithmic bias, data harms, and labor/environmental costs of AI, resisting “banking” models of instruction and centering learner agency and dialogue (Freire, 2000/1970). Readings and protocols position faculty to question how search, recommendation, and generative systems can reproduce social inequities and to design mitigation strategies in coursework (Noble, 2018; Selwyn, 2013). This framing keeps “should we” and “for whom” questions alongside “how to,” shaping adoption decisions and assessment practices.
Finally, governance is treated as participatory co-design. Fellows and Partners co-created AI Faculty Guidelines and contributed to student-facing AI policy using co-creation practices that value stakeholder expertise, move iteratively from generative research to prototyping, and increase legitimacy and implementability of policies (Sanders & Stappers, 2008; Spinuzzi, 2005). The resulting artifacts (guidelines, implementation checklists, exemplar assignments) serve both as boundary objects and as catalysts in the train-the-trainer rollout.

Methods

Design
We conducted a two-year mixed-methods, design-based improvement study organized in semester-long Plan–Do–Study–Act (PDSA) cycles to iteratively refine AI-integrated teaching and a train-the-trainer professional-development (PD) model (Creswell & Plano Clark, 2018; Design-Based Research Collective, 2003; Bryk, Gomez, Grunow, & LeMahieu, 2015; Langley et al., 2009). A convergent parallel design guided simultaneous collection of quantitative and qualitative data with joint interpretation at the cycle and annual levels (Fetters, Curry, & Creswell, 2013).
Setting and Participants
The study took place at university in Texas, a multi-school institution. Each year the Fellowship included one AI Fellow per school (target n=10) and 1–3 AI Partners per school in the second year who supported local rollout. Additional participants were faculty who attended school-level PD and a campus mini-conference. Purposeful, criterion sampling ensured cross-school coverage and role diversity (Patton, 2002).
Recruitment and Selection
A campus-wide call detailed expectations (time, artifacts, school-level PD) and selection criteria. Applicants submitted (a) a one-page application that included questions regarding their philosophical beliefs about learning and teaching with AI. A rubric (0–3 per criterion) prioritized disciplinary diversity, prior PD leadership, feasibility, and commitment to equity/ethics; the top-ranked applicants in each school were reviewed by a committee of faculty and university leadership who appointed the Fellows. Fellows nominated 1–3 Partners in year two. This process balanced purposive and maximum-variation sampling common in improvement and design research (Patton, 2002; Bryk et al., 2015).
Intervention (Fellowship Structure)
Monthly 2 hour cross-school meetings (≈10/academic year) followed a fixed arc: (A) PDSA share-outs; (B) tool/strategy sandbox with ethical/equity prompts; (C) design sprint toward artifacts (syllabi language, assignments, rubrics); (D) plan next school-level PD. Fellows delivered ≥2 PD sessions/semester in their schools with Partners as co-facilitators in year two via monthly school meetings, or small group sessions with faculty. Additionally, fellows and partners presented at the multimodal mini-conference. Policy co-creation ran in parallel (AI Faculty Guidelines; contributions to student AI policy). This cadence aligns with DBR’s naturalistic iteration and NIC-style disciplined inquiry (Design-Based Research Collective, 2003; Bryk et al., 2015; Langley et al., 2009).
Data Sources and Instruments
Surveys (pre, end-semester, end-year).
AI-Teaching Self-Efficacy (12 items; 5-point Likert) constructed per Bandura’s scale-design guidance and teacher-efficacy operationalization (Bandura, 2006; Tschannen-Moran & Woolfolk Hoy, 2001).

TPACK short form (15 items across TK, TCK, TPK, TPACK) adapted from the validated instrument (Schmidt et al., 2009).

Adoption attributes (5 items) aligned with Diffusion of Innovations: relative advantage, compatibility, trialability, observability, complexity (Rogers, 2003).

Equity/Ethics stance (6 items) on bias, privacy, labor/environmental impacts, reflecting critical digital-pedagogy concerns operationalized for survey use (Selwyn, 2013; Noble, 2018).

Artifacts. PD artifacts (agendas, slides, handouts, attendance logs, feedback forms), governance artifacts (guideline drafts with version histories), and mini conference evaluation forms.
Observations. Non-evaluative observations of school monthly meetings and detailed reflections from fellows and partners. Indicators reflect core PD features associated with teacher learning and observable practice change (Desimone, 2009).
Interviews/Focus Groups. Semi-structured guides for Fellows (mid-year, end-year) and Partners (end-year).

Results

Faculty learning and self-efficacy.
Across two cohorts, fellows report greater confidence designing AI-supported tasks, coaching responsible use, and troubleshooting classroom workflows; meeting observations show a shift from “tool trials” to modeling and guided practice. We expect statistically reliable pre–post gains on AI-teaching self-efficacy subscales, with largest growth in mastery and vicarious experience.
TPACK and course redesign.
Artifacts indicate movement from tool-first activities to discipline-anchored tasks (e.g., ELA draft-feedback protocols; nursing SBAR handoffs with AI critique; business analytics prompts with data provenance checks). We expect increases on TPACK composite scores and higher rubric ratings for T–P–C alignment, explicit AI guidance, equity/ethics integration, and assessment clarity.
Diffusion, reach, and train-the-trainer effects.
All fellows delivered the required school-level PD (≥2/semester) with Partners co-facilitating; agendas and attendance logs document spread beyond early adopters. We expect growth in unique and repeat attendees year-over-year, evidence of peer demos in multiple schools, and network metrics showing fellows as central bridges for cross-school sharing.
Policy and governance outcomes.
Fellows and Partners co-created AI Faculty Guidelines and contributed to a student AI policy. Version histories show iterative refinement linked to PD cycles; we expect increased guideline uptake in syllabi language and advising artifacts, with qualitative data describing improved clarity and consistency for students and instructors.
Equity and ethics in practice.
Meeting notes and PD materials show routine attention to algorithmic bias, attribution, privacy, and environmental costs. We expect higher self-reported comfort facilitating equity-focused conversations and more assignments that operationalize mitigation strategies (citation scaffolds, bias checks, disclosure statements).
Mini-conference and public products.
The multimodal showcase generated shareable decks, exemplars, and short demos adopted in multiple schools. We expect rising submissions and cross-disciplinary sessions, with attendee feedback citing usefulness of concrete artifacts.

Importance

The AI Fellowship builds cross-disciplinary capacity to teach with—and about—AI through a repeatable train-the-trainer model. It turns monthly PD into concrete classroom artifacts (syllabi language, assignments, rubrics) that foreground equity, ethics, and disciplinary fit, yielding visible shifts in course design and student learning supports.
Scientific contribution.
This is a two-year, design-based, mixed-methods study that links measurable growth (AI-teaching self-efficacy, TPACK) to observable products and diffusion patterns across schools. It advances evidence on scalable PD by coupling communities of practice and distributed leadership with disciplined PDSA cycles, and by tying PD directly to policy co-creation.
Value for conference audiences.
PD leaders/directors: A potential model and playbook with processes, stipend structure, monthly meeting objectives, fidelity markers, observation protocol, evaluation plan, and reporting dashboards.

Faculty developers: Discussions around potential instruments (surveys, artifact rubrics), meeting arcs, and design-sprint templates that accelerate local adoption with limited resources.

Department chairs/faculty: Examples of documentation (AI disclosure policies, assignment prompts, feedback workflows), privacy, and bias concerns.

Why now?
It offers a path from “trying tools” to a potential AI integration—scalable and actionable for institutions seeking change.

References

Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Prentice-Hall.

Bandura, A. (1997). Self-efficacy: The exercise of control. W. H. Freeman.

Bandura, A. (2006). Guide for constructing self-efficacy scales. In F. Pajares & T. Urdan (Eds.), Self-efficacy beliefs of adolescents (pp. 307–337). Information Age.

Braun, V., & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3(2), 77–101.

Brown, A. L. (1992). Design experiments: Theoretical and methodological challenges in creating complex interventions in classroom settings. Journal of the Learning Sciences, 2(2), 141–178.

Bryk, A. S., Gomez, L. M., Grunow, A., & LeMahieu, P. G. (2015). Learning to improve: How America’s schools can get better at getting better. Harvard Education Press.

Cohen, J. (1960). A coefficient of agreement for nominal scales. Educational and Psychological Measurement, 20(1), 37–46.

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Lawrence Erlbaum.

Creswell, J. W., & Plano Clark, V. L. (2018). Designing and conducting mixed methods research (3rd ed.). SAGE.

Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16(3), 297–334.

Design-Based Research Collective. (2003). Design-based research: An emerging paradigm for educational inquiry. Educational Researcher, 32(1), 5–8.

Desimone, L. M. (2009). Improving impact studies of teachers’ professional development: Toward better conceptualizations and measures. Educational Researcher, 38(3), 181–199.

Fetters, M. D., Curry, L. A., & Creswell, J. W. (2013). Achieving integration in mixed methods designs—Principles and practices. Health Services Research, 48(6 Pt 2), 2134–2156.

Freeman, L. C. (1979). Centrality in social networks: Conceptual clarification. Social Networks, 1(3), 215–239.

Freire, P. (2000). Pedagogy of the oppressed (30th anniversary ed.; original work published 1970). Bloomsbury.

Langley, G. J., Moen, R., Nolan, K. M., Nolan, T. W., Norman, C. L., & Provost, L. P. (2009). The improvement guide: A practical approach to enhancing organizational performance (2nd ed.). Jossey-Bass.

Lave, J., & Wenger, E. (1991). Situated learning: Legitimate peripheral participation. Cambridge University Press.

Lincoln, Y. S., & Guba, E. G. (1985). Naturalistic inquiry. SAGE.

Miles, M. B., Huberman, A. M., & Saldaña, J. (2014). Qualitative data analysis: A methods sourcebook (3rd ed.). SAGE.

Mishra, P., & Koehler, M. J. (2006). Technological Pedagogical Content Knowledge: A framework for teacher knowledge. Teachers College Record, 108(6), 1017–1054.

Noble, S. U. (2018). Algorithms of oppression: How search engines reinforce racism. NYU Press.

Patton, M. Q. (2002). Qualitative research & evaluation methods (3rd ed.). SAGE.

Rogers, E. M. (2003). Diffusion of innovations (5th ed.). Free Press.

Sanders, E. B.-N., & Stappers, P. J. (2008). Co-creation and the new landscapes of design. CoDesign, 4(1), 5–18.

Schmidt, D. A., Baran, E., Thompson, A. D., Koehler, M. J., Mishra, P., & Shin, T. S. (2009). Technological Pedagogical Content Knowledge (TPACK): The development and validation of an assessment instrument for preservice teachers. Journal of Research on Technology in Education, 42(2), 123–149.

Selwyn, N. (2013). Distrusting educational technology: Critical questions for changing times. Routledge.

Spinuzzi, C. (2005). The methodology of participatory design. Technical Communication, 52(2), 163–174.

Spillane, J. P. (2006). Distributed leadership. Jossey-Bass.

Tschannen-Moran, M., & Woolfolk Hoy, A. (2001). Teacher efficacy: Capturing an elusive construct. Teaching and Teacher Education, 17(7), 783–805.

Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes. Harvard University Press.

Wenger, E. (1998). Communities of practice: Learning, meaning, and identity. Cambridge University Press.

Wenger, E., McDermott, R., & Snyder, W. (2002). Cultivating communities of practice. Harvard Business School Press.