AIEd Knowledge BaseAIEd KB
On this page

Research Article

Rethinking artificial-intelligence literacy through the lens of teacher educators: The adaptive AI model

added 2026-09-21 · updated 2026-09-21

Synthesis: Eyal (2025) argues that most frameworks for measuring AI Literacy impose fixed competency levels and overlook context, then builds an alternative with practitioners. Using Design-Based Research, 22 higher-education teacher educators examined five published assessment models, tested them against real dilemmas, and co-designed the Adaptive Artificial-Intelligence-Literacy Model (AALM). The model replaces linear ladders with three inter-related axes: context fit (infrastructure, socio-cultural factors, local needs, developmental stage), professional needs (discipline, pedagogy, leadership, support), and dynamic development (growth over time as technology and roles change). The axes overlap in named CF_PN, PN_DD, and DD_CF zones around a central AI-literacy area. It also produced a 20-item reflective self-assessment questionnaire rated 1 to 5. Eyal casts Teacher AI Competency as situational rather than uniform: a pre-service teacher in a resource-limited Bedouin region, a math exam coach, a social-science novice, and a school principal need different competencies. The work ties AI literacy to Metacognition and reflective practice, and reframes Workplace Learning around Teaching flexibility instead of standardized rubrics.

Key Findings

  • Twenty-two teacher educators (82% female; 64% held a Ph.D.) co-designed the AALM during a 180-hour course, across eight sessions (64 academic hours) in four iterative design cycles.
  • Participants analyzed five recently published frameworks (Filo et al., AI-TPACK, Five Big Ideas, ABCE, Use-Modify-Create) and faulted them for lacking practical assessment tools, socio-cultural attention, and implementation guidance.
  • The AALM reorganizes assessment around three axes (context fit, professional needs, dynamic development) with explicit overlap zones and a central integrated area rather than fixed predetermined stages.
  • The self-assessment questionnaire has 20 items on a 1 to 5 scale, five per axis plus five integrative questions, each paired with an example and improvement opportunities.
  • Three cases drawn from 22 shared dilemmas showed the model's reach: a third-year pre-service teacher working with the Bedouin population, two contrasting subject teachers, and a principal overseeing about 800 students and 70 teachers.

How the design-based research unfolded

The study ran inside a year-long 180-hour course, "Innovation and digital-learning design in academia." Data collection spanned four iterations within eight sessions (64 academic hours), conducted mostly online with one face-to-face session. The first iteration had teams examine existing assessment models and apply them to a case study of a pre-service teacher. The second asked for concrete alternatives and peer critique. The third presented the emerging model for agreement and correction, and the fourth translated its components into questionnaire indicators. Instruments were Zoom recordings and transcripts, individual written responses, and collaborative products such as model drafts and the final questionnaire. A methodological journal documented revisions and their rationale. Analysis combined content and thematic analysis with triangulation, peer debriefing, and a member check; participation was 100% across all study stages.

The adaptive model's three axes

The context-fit (CF) axis covers adaptation to available infrastructure, socio-cultural factors, local needs, and the educator's developmental stage in AI literacy. The professional-needs (PN) axis addresses discipline, pedagogical aspects, leadership, and support. The dynamic-development (DD) axis captures growth in response to technological advancements, evolving roles, changing needs, and expanding capabilities. The axes are inter-connected: CF_PN marks where local conditions meet professional demands, PN_DD where professional needs evolve over time, and DD_CF where shifts in the educational environment influence literacy development. At the core sits the central AI-literacy zone, where context, needs, and development converge. No perfect profile exists: educators develop the skills most relevant to their situation at a given time. Unlike prior frameworks, the AALM allows non-linear progression, addresses infrastructure limits, and ships with a practical self-assessment instrument.

What the reflective questionnaire showed

The fourth iteration produced a reflective diagnostic questionnaire, built through shared document editing and consensus-building discussion until participants reached agreement. Its items ask educators to rate themselves from 1 (to a limited extent) to 5 (to a great extent). Context-fit items cover adapting AI to school infrastructure, aligning it with students' cultural characteristics and abilities, spotting context-specific opportunities, and accounting for contextual limitations. Professional-need items cover using AI to enhance teaching quality, identifying pedagogical opportunities, achieving teaching goals, selecting subject-appropriate tools, and designing meaningful AI-integrated activities. Dynamic-development items cover staying current with AI in education, experimenting, reflecting on experience, identifying development areas, and growing confidence. Five integrative items ask about aligning professional development with school and student needs, developing creative solutions, evaluating integration effectiveness, and sharing insights with colleagues. Each item also leaves room for practical examples and improvement opportunities.

What this means for practice

  • Teacher educators and faculty developers. Build AI professional development around authentic practice, ongoing reflection, and supportive learning communities not lists of technical skills.
  • Instructional designers. Administer the 20-item questionnaire at program entry and again later, and read overlap zones rather than a single total to locate where support is needed.
  • Institutions and policymakers. Judge AI literacy against local infrastructure, cultural context, and role demands, so teachers in resource-limited settings are not misread as low-literacy.

Limitations

  • Validation was qualitative only. Content validity came from the 22 participating teacher educators during the third iteration, and the study included no quantitative reliability testing or large-scale empirical validation.
  • Participants were one cohesive Israeli group, selected for teacher-education and technology experience, a graduate degree, and strong digital and pedagogical skills, so the model's transferability rests on further testing.
  • The three case studies are illustrative dilemmas shared by participating educators rather than systematically sampled cases, and the questionnaire is self-report with no behavioral or outcome measure.

Connected Concepts

Connected Articles

Citation

Eyal, L. (2025). Rethinking artificial-intelligence literacy through the lens of teacher educators: The adaptive AI model. Computers and Education Open, 9, 100291.

Metadata

Study designAction Design Research
LevelHigher Ed, Teacher Education
Intended audienceInstructors, Faculty Developers, Instructional Designers
Learning and instructionMetacognition, Workplace Learning
Research methods and evaluationDesign-Based Research
Foundations of AI in educationAI Literacy, Teacher AI Competency, Design Thinking
SourceHistoryEditChat

Embed this page

Copy the code below to embed a chromeless version of this page in a learning management system or other website. The embedded view hides the site header, navigation, and footer.