Berlin, 25 June 2026 — Competency frameworks for the use of AI in schools reduce pedagogical action to measurable skills, thereby overlooking central dimensions of professional responsibility. This is the key finding of an analysis of international AI literacy frameworks conducted by the research team of the AI2PI Teacher Academy at Humboldt University of Berlin.

Background: AI2PI Teacher Academy

The project "From Artificial Intelligence to Pedagogical Innovation" (AI2PI) is an Erasmus+ Teacher Academy comprising nine partner institutions from seven countries (duration: 2025–2028). Its objective is to empower teachers to critically reflect on AI tools in schools and classrooms.

Key Findings of the Analysis

The researchers examined three major international frameworks – DigCompEdu AI (EU), UNESCO AICFT, and OECD AI Literacy – from a post-digital perspective. Three dominant patterns emerged:

• Instrumental framing of AI: AI is primarily understood as a tool for optimizing existing practices.
• Individualization of responsibility: Responsibility appears as an individual teacher competency rather than an institutional or systemic issue.
• Underrepresentation of structural dimensions: Epistemic, infrastructural, as well as social and ecological aspects remain marginal.

Central Discussion: Limits of Competency Logic

The presentation focused on the theoretical discussion of professional action beyond operationalizable competencies. Three reference points structure this debate:

Responsibility as Response (Brady 2020)

Competency frameworks define responsibility as a demonstrable skill, thereby altering the concept itself. Operationalization produces a truncated form in which only that which is visible, measurable, and verifiable is considered valid. The problem lies not in incomplete accountability, but in the systematic narrowing of what is recognized as responsibility.

Agency as Situational Enactment (Biesta & Priestley 2015)

Agency is not a possessible competency, but emerges only in concrete situations of professional judgment. Competency frameworks describe skills, not the genesis of pedagogical judgments under situational conditions.

Identity as a Core Question (Korthagen & Vasalos 2005)

AI frameworks operate almost exclusively on the outer layers of the onion model (behavior, competencies). The crucial question remains unasked: Who do I want to be as an educator in a world shaped by AI? This is not a question of competency, but of identity.

Central Discussion: Limits of Competency Logic

The authors follow Macgilchrist, Flury & Roß (2025): The institutional pressure to impart immediately actionable AI competencies ("urgency of practice") displaces reflective and educational-theoretical questions. Instead, the following are required:

• Learning about AI, not just learning with AI
• Spaces for reflection on professional self-understanding under AI conditions
• Institutionally anchored formats for theory-guided self-inquiry
• Limitation, rather than abolition, of competency frameworks The gaps in these frameworks are not merely a deficit, but arise structurally from the fact that certain aspects of professional action—responsibility as a response process, situational judgment, professional identity—are in principle not translatable into competency models.

Contact

Dr. Ole Engel, Dr. Michael Schlauch, Prof. Dr. Ulrike Stadler-Altmann Humboldt-Universität zu Berlin | Institute of Education