Pioneering Teacher Educators Navigating AI Integration in Pre-Service Teacher Preparation: Strategies and Challenges

Synthesis: This qualitative interpretive phenomenological study interviewed 13 pioneering teacher educators (seven pedagogical advisors and six lecturers) in seven Israeli teacher training institutions about how they prepare pre-service teachers for AI literacy and AI-integrated teaching, using semi-structured interviews analyzed through human thematic analysis combined with AI-assisted dialogic analysis. Participants saw themselves as agents of change who must model critical AI use, reported a crisis of trust when students submitted unprocessed AI outputs, and redesigned assessment around process and classroom evidence. They also raised epistemic concerns about cognitive atrophy and the erosion of expertise, plus policy gaps, ethics and privacy worries, and cultural and economic barriers. The authors read these findings through post-digital theory and argue that AI-era Professional Development requires redefining literacy, pedagogy, and human agency rather than adding a technical module.

Key Findings

1. Participants defined their role as preparing pre-service teachers for informed AI use, describing themselves as agents of change who model pedagogical innovation rather than transmit knowledge; Orna argued that teacher educators who do not integrate AI cannot expect students to, and Yair that lecturers must demonstrate responsible AI use in class.
2. Epistemic concerns were central. Galit described a shift in conceptions in which AI forces a redefinition of what learning, knowledge acquisition, and inquiry mean, and warned that over-reliance diminishes intellectual curiosity and the joy of learning.
3. Iris warned that using AI before professional mastery erodes expertise, and insisted students learn the classical approach to text adaptation first; Yaakov reported watching seminar students let AI do the work and "simply did not develop," naming this his fear of cognitive atrophy, since "AI delivers products but not processes."
4. Rivka proposed reorienting from knowledge acquisition to skills acquisition, since knowledge is now universally accessible and the value of education lies in the critical, accurate, dialogic skills used to engage it.
5. A crisis of trust emerged when students submitted raw AI-generated outputs, sometimes without reading or editing them. Riham described losing trust in students displaying skills they did not expect, and Iris called the discovery that thinking tasks had not been attempted "really frustrating."
6. Orna recounted a case study submitted in the name of a pupil from an entirely different cultural background, with the AI tool's address in the first line, which she called one of the hardest confrontations she has had with students.
7. Participants named an "authenticity crisis": take-home written work could no longer reliably evidence learning, prompting the question of how to assess students authentically in the age of AI.
8. Strategies included moving work into class (flipped classroom, presentations, peer teaching, elevator pitches, in-class visual models) and process-based assessment such as examining prompts, Google Docs version history, and DocWiz monitoring of group contributions.
9. Some returned to non-digital evidence (reflective "journey journals" in physical notebooks, handwritten reflective paragraphs) and, as a last resort, to supervised examinations and anticipated oral thesis defenses; Orna described returning to exams as personally painful but unavoidable.
10. In field experience, pre-service teachers used AI for differentiated lesson planning: five interviewees directed students to build teaching units with AI and two guided adaptation to specific pupil profiles including special needs, and students produced digital games, podcasts, visual aids, theorist "bots" they had to train, and kindergarten learning-corner simulations.
11. Student experiences ranged from heightened Self-Efficacy to frustration when tools failed complex pedagogic demands; Sarit reported students concluding "I am the teacher, and AI helps me," and Rivka noted AI's value in easing the isolation of student teachers alone in the field.
12. Institutional support varied widely: workshops sequenced by study year, techno-pedagogical units working inside classes, and self-directed modules such as the six-unit Prizma module, but no formal AI literacy course in most institutions, which Mira attributed to rapid change and a scarcity of qualified lecturers.
13. Persistent challenges included the "tsunami" pace of technological change, absent unified policy (Talia noted that what one lecturer allows another forbids), student deficits in critical evaluation of outputs, cultural and religious variation in technology exposure, and economics: nine of thirteen participants funded advanced AI tools from their own pockets, warning this could deepen the Digital Divide among students.

Study Design & Method

The study adopts a qualitative interpretive phenomenological approach in the tradition of Smith et al. (2021), examining seven Israeli teacher education colleges through the lived experience of 13 participants: seven pedagogical advisors and six lecturers (11 females, two males). Five institutions belong to the state-Jewish stream, one to the state-religious stream, and one to the Arab stream, and participants were purposively selected for high proficiency in AI integration on the recommendation of institutional leaders. Data came from semi-structured in-depth interviews of 45–60 minutes, audio-recorded and transcribed verbatim. Analysis combined human thematic analysis in ATLAS.ti (Version 26) following Grounded Theory coding, with the first six interviews generating the coding system, and an independent AI-assisted analysis in NotebookLM, followed by systematic comparison of themes. Trustworthiness followed Lincoln and Guba (1985) and Tracy (2010): triangulation of method and investigators, a reflexive journal, an audit trail, and member checking with five interviewees across educational streams. Ethics approval came from the Ethics Committee of Levinsky-Wingate Academic College (protocol 08-081225, 8 December 2025), with informed consent covering AI-assisted analysis.

Research and policy agenda

The authors read the trust and authenticity crisis through Mezirow's transformative learning theory as a disorienting dilemma, and the most successful participants as moving from excluding AI toward redesigning practice around human–AI entanglement, in Fawns's (2022) sense of entangled pedagogy. They argue for redefining AI literacy from tool proficiency to critical, ethical, and epistemological competence, for making human–AI processes visible and evaluable, and for cultivating critical AI agency: the capacity to evaluate outputs, resist the colonization of professional judgment, and keep a clear professional identity. Because AI literacy demands differ across disciplines and communities, they hold that curricular responses must be situated rather than universal. Practically, they call for formalizing AI literacy as a curricular requirement comparable to academic writing literacy, delivered by specialists combining pedagogical and technological expertise, plus institutional funding for professional tools and coherent cross-college policy. They also cite OECD (2026) evidence that generative AI supports learning under clear pedagogical principles but otherwise yields a "mirage of false mastery," including a finding that students using AI were 48% more successful at tasks while performance dropped by 17% when assistance was withdrawn.

What this means for practice

• Teacher educators. Move at least one high-stakes assignment onto in-class or process evidence — presentations, peer teaching, elevator pitches, in-class visual models, prompt records, document version history, group-contribution logs — because these 13 participants found take-home written work could no longer evidence learning.
• Instructors. Teach the classical method before AI assistance and say so in the brief: participants insisted students learn the traditional approach to text adaptation first, warning that AI use before professional mastery erodes expertise.
• Teacher educators. Sequence AI literacy across the study years and pair it with one stated course policy, since what one lecturer allowed another forbade and no institution in the study ran a formal AI literacy course.
• Institutions. Fund the tools centrally: 9 of 13 participants paid for advanced AI tools from their own pockets and warned that quality access would otherwise track ability to pay, deepening the digital divide among pre-service teachers.
• Instructors. Use AI where it buys differentiation rather than substitution — five participants had pre-service teachers build whole teaching units with AI and two guided adaptation to a specific pupil profile including special needs — and check the output against that profile.

Limitations

The authors acknowledge five limitations. The 13 participants were selected precisely for their pioneering engagement, so findings reflect early adopters and cannot be generalized to teacher educators in Israel or beyond; skeptical faculty are not represented. The data rest entirely on self-reported accounts, with the student voice absent. The study captures a specific moment in a rapidly evolving technological and policy landscape, limiting temporal transferability. Although the sample spans seven institutions across three educational streams, the small number of participants per institution limits systematic institutional comparison. Finally, the hybrid human plus AI-assisted analysis is still an emerging practice in qualitative research, and its interpretive complexity needs further methodological attention.

Connected Concepts

• AI Literacy — the core construct participants reframe from tool proficiency to critical and ethical competence
• Professional Development — the institutional setting whose AI-era reimagining the paper argues for
• Teacher AI Competency — teacher educators developing their own AI literacy while fostering students'
• Generative AI — the tool set reshaping lesson planning, materials, and assessment
• Academic Integrity — the trust and authenticity crisis behind assessment redesign
• Authentic Assessment — in-class, process-based, and non-digital evidence of learning
• Cognitive Offloading — participants' fear of cognitive atrophy and lost expertise
• Critical Thinking — the deficit participants saw in students' uncritical use of outputs
• Digital Divide — economic barriers and unequal access to quality AI tools
• Educational AI Policy — absent unified institutional policy and national frameworks discussed
• Assessment — shifts toward process documentation, examinations, and oral defenses
• Teaching — pedagogical advisors and lecturers as agents of change and epistemic mediators

Connected Articles

• Human-centered AI for teacher educators: Designing professional learning for critical AI literacy — teacher educators' perspectives on human-centered AI integration
• Conceptualizing pre-service teachers' readiness for AI integration into teaching practices: An intelligent-TPACK approach — conceptual work on preparing pre-service teachers for AI
• Assessing AI-TPACK readiness in mathematics teacher education: The role of self-efficacy and teaching beliefs — AI-TPACK in teacher education programs
• Assessing faculty self-perceived knowledge in using generative AI to teach 21st-century skills — faculty GenAI knowledge through a TPACK lens
• When faculty ask, 'what's the point of teaching?': GenAI as identity crisis, not skills gap — faculty identity tensions around generative AI
• Beyond Detection: Redesigning Authentic Assessment in an AI-Mediated World — redesigning assessment when detection fails
• The Impact of Generative AI on Academic Integrity of Authentic Assessments Within a Higher Education Context — authenticity and integrity in AI-era assessment
• Navigating uncertainty: university teachers' experiences and perceptions of generative artificial intelligence — teachers navigating uncertainty with generative AI
• Science educators' AI literacy and AI usage in teaching: Implications for post-qualification programs — AI literacy development among educators in practice

Citation

Goldstein, O., Marae-Haj, N., & Zidan, W. (2026). Pioneering Teacher Educators Navigating AI Integration in Pre-Service Teacher Preparation: Strategies and Challenges. AI in Education, 2(3), 29.