Adapting Higher Education Validity Models to Meet the Realities of Secondary Classrooms
When Eindhoven University of Technology (TU/e) overhauled its assessment strategy around generative AI, it demonstrated that governance is not about catching cheaters—it is about validity protection: ensuring that assessments measure real student understanding (Sadowski & Oliveira, 2026)
However, secondary schools operate under vastly different constraints than universities. While higher education institutions enjoy institutional autonomy and research budgets, secondary schools deal with rigid exam board syllabi, strict safeguarding duties for minors, severe teacher workload constraints, and an environment that a recent report described as "chaotic" and "crying out" for national guidance (Clarke, 2026; Seldon & Bunting, 2026)
Without waiting for central government directives, school leaders can adapt higher education validity principles into a streamlined, high-impact framework tailored for the 11–18 classroom (Sadowski & Oliveira, 2026; Seldon & Bunting, 2026)
The Unique Constraints of Secondary Education
Translating AI governance from higher education to secondary schools requires navigating four distinct structural realities:
Safeguarding and Minor Data Privacy: Unlike university students, secondary pupils are legal minors, making data privacy and compliance under Regulation (EU) 2024/1689 critical (European Union, 2024)
. Entering student work or personal data into unvetted commercial AI models poses severe GDPR and safeguarding risks, alongside deployer obligations under the EU AI Act (European Union, 2024) . High-Stakes External Assessment: Secondary schools must prepare students for standardized, invigilated external examinations, such as GCSEs, A-Levels, or the International Baccalaureate (Seldon & Bunting, 2026)
. Unregulated AI usage during home study creates a dangerous illusion of mastery that shatters under exam conditions (Seldon & Bunting, 2026) . Teacher Time Deficits: University task forces can dedicate hundreds of research hours to analyzing interaction logs (Sadowski & Oliveira, 2026)
. Secondary teachers, managing up to 200 students across multiple year groups, require low-friction, immediate solutions (Clarke, 2026) . Eroding Perceived Value of Schooling: Recent surveys of young people show that unguided AI exposure leads some students to question the value of traditional schooling, mistakenly assuming AI makes learning and teachers obsolete (Clarke, 2026; Seldon & Bunting, 2026)
.
The "Two-Lane" Assessment Strategy
To reconcile AI integration with external exam requirements, secondary schools should adopt a Two-Lane Assessment Model proposed for education reform (Seldon & Bunting, 2026)
Lane 1: Secure Summative Assessment (Proof of Individual Mastery)
Purpose: Ensures absolute assessment validity for grades and exam preparation (Seldon & Bunting, 2026)
. Execution: Closed-book in-class writing, invigilated assessments, and brief oral check-ins or vivas (Seldon & Bunting, 2026)
. Governance: Pure human performance without digital aids (Seldon & Bunting, 2026)
.
Lane 2: Formative AI-Augmented Learning (Process & Literacy)
Purpose: Builds critical thinking, prompt literacy, and subject mastery (Clarke, 2026; Seldon & Bunting, 2026)
. Execution: Students use vetted AI tools for brainstorming, proofreading, code debugging, or robotics projects—provided they document their process (Clarke, 2026; Seldon & Bunting, 2026)
. Governance: Process-visibility requirements, where students submit their interaction steps alongside final work (Sadowski & Oliveira, 2026; Seldon & Bunting, 2026)
.
KEY TAKEAWAY FOR SCHOOL LEADERS
Do not attempt to catch every instance of AI homework assistance (Clarke, 2026)
. Instead, split your strategy: use Lane 1 to guarantee grade validity under controlled conditions, and use Lane 2 to teach responsible, transparent AI collaboration (Clarke, 2026; Seldon & Bunting, 2026) .
Aligning Secondary AI Rules with European Standards
Secondary schools do not need to invent policy from scratch. They can anchor their rules in established European governance models:
Council of Europe Compass (2026): Emphasizes Education about AI (building AI literacy and critical understanding) and Education with AI (pedagogical integration) while guaranteeing democratic human oversight (Council of Europe, 2026)
. EU AI Act (2024): Classifies AI systems used in educational grading, admissions, and monitoring as high-risk applications under Annex III (European Union, 2024)
. Schools must ensure transparency, ban emotion recognition technologies, and train teaching staff in AI literacy under Article 4 (European Union, 2024) .
A Secondary School "Starter Menu" for Assessment Validity
By simplifying Samuel Messick’s validity facets—as adapted by Sadowski and Oliveira (2026) for higher education—school leaders can give department heads a clear menu of practical options (Sadowski & Oliveira, 2026)
| Validity Facet | School Governance Goal | Practical Classroom Tool |
| 01 CONTENT | Clear assignment boundaries | Traffic Light Labelling: Explicitly state on homework whether AI is Prohibited (Red), Assistive for editing (Yellow), or Fully Integrated (Green) (Sadowski & Oliveira, 2026) |
| 02 SUBSTANTIVE | Visible student thinking | Draft History & Prompts: Require students to attach their main prompts and two draft iterations for major home assignments (Sadowski & Oliveira, 2026) |
| 03 STRUCTURAL | Evaluating AI collaboration | Reflective Rubrics: Grade not just the final essay, but a 150-word student reflection analyzing where the AI was helpful versus where it hallucinated (Sadowski & Oliveira, 2026) |
| 04 GENERALISABILITY | Consistent department standards | Shared Department Templates: Standardize approved AI prompt structures for subjects like Science, Humanities, and Modern Foreign Languages (Sadowski & Oliveira, 2026) |
| 05 EXTERNAL | Real-world application | Critique the Bot: Have students edit and correct an intentionally flawed AI-generated answer to an exam-style question (Sadowski & Oliveira, 2026; Seldon & Bunting, 2026) |
| 06 CONSEQUENTIAL | Ethical & safe usage | Data Protection Policy: Enforce strict rules prohibiting students from entering personal details or school identifiers into public AI systems (European Union, 2024; Sadowski & Oliveira, 2026) |
Moving Forward: Immediate Steps for Headteachers
Establish Clear Traffic Lights: Replace vague AI policies with explicit assignment-level labels (Red, Yellow, Green) (Sadowski & Oliveira, 2026)
. Protect Staff Time: Focus teacher professional development on evaluating process and critical reflection rather than attempting to act as "AI detectors" (Clarke, 2026; Sadowski & Oliveira, 2026)
. Audit Approved Tools: Ensure all digital tools used in class comply with student data privacy regulations and the EU AI Act (European Union, 2024)
. Maintain Open Dialogue: Emphasize to students that AI is a tool to enhance human thinking, not a substitute for developing personal cognitive capability (Clarke, 2026; Seldon & Bunting, 2026)
.
References
Clarke, H. (2026, September 22). Schools not ready for AI and 'crying out' for government guidance, report warns. BBC News.
Council of Europe. (2024). Our member states.
Council of Europe. (2026, September 24). Compass for governance of AI & digital transformation in education.
European Union. (2024). Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act). Official Journal of the European Union, L 2024/1689.
Sadowski, B., & Oliveira, M. (2026). Governing AI in assessment: A starter menu. Assessment in Education: Principles, Policy & Practice.
Seldon, A., & Bunting, T. (2026). AI in education: A 10-point national framework for schools. AI in Education.[cite: 18, 19, 20]
No comments:
Post a Comment
Note: only a member of this blog may post a comment.