Analysis and Implementation of the UNESCO AI Competency Framework: A Case Study of a Multimodal Teacher Training Program

Abstract

Abstract— The UNESCO AI Competency Framework for Teachers (2024) defines five competency areas (Human-centred mindset, Ethics of AI, AI foundations and applications, AI pedagogy, AI for professional development) across three progression levels (Acquire-Deepen-Create). Despite its recognition as a global standard, AI competency frameworks for teachers remain rare globally (only seven countries by 2022), while documented implementation models of the UNESCO framework are absent. Key challenges include translating theoretical descriptions into concrete educational activities, integrating multiple educational theories without cognitive overload, and ensuring progressive competency development. This work analyzes these challenges and presents an implementation case study through a design-based research approach. The research question explores how the UNESCO framework can be translated into a functional educational program that integrates educational theory without cognitive burden. Through an iterative design process spanning 7 + 2 cycles, a program was developed in two versions (Greek and English) with 110 participants in total. The methodology was based on systematic alignment of each module with the five areas and three levels of the framework, applying TPACK for the holistic approach to technology-pedagogy-content, UDL for accessibility, and Bloom's Taxonomy for cognitive complexity, through an "invisible theory" strategy that embeds pedagogical principles without academic terminology. The implementation analysis reveals specific design decisions for each area: For Human-centred mindset, five transversal pedagogical pillars were developed (critical thinking, ethical use, personalization, collaboration, assessment). For Ethics of AI, a dedicated module was combined with embedded ethical dilemmas presented through audio scenarios featuring virtual educators expressing concerns before expert solutions. For AI foundations, progressive complexity was applied from basic concepts (Acquire) to LLMs and prompt engineering (Deepen) to content creation (Create), with the development of EduPrompt Studio as a tool that implements automatic integration of educational theories into prompt generation. For AI pedagogy, UDL applications and differentiation were incorporated. For Professional development, an ecosystem was created with Teacher Toolbox, forums, and self-reflection. The implementation leveraged multimodal means—AI-generated podcasts with dialogical approach, virtual exhibitions on the ArtSteps platform for visualization of ethical dilemmas, H5P interactive exercises—to support different learning styles. Strategic partnerships with six international platforms (Brisk, Genially, Twee, Redmenta, Eduaide, Almanack) provided premium access (3-12 months) for immediate practical application of professional tools, ensuring the transition from theory to practice. The case study identifies three critical factors for successful implementation: explicit alignment with progression levels in each activity, strategic partnerships for practical tool application, and reusable architecture adaptable to different cultural contexts. The contribution is twofold: an analytical model for implementing the UNESCO framework and design principles for replicability in other educational systems. Future research will investigate effectiveness through systematic evaluation of participants.