Research that tests the ideas before recommending them.

All research

1. 01

   Working paper

   A curriculum-based assessment framework: the research plan

   Five steps to test a framework that gives every teacher a default progression and assessment plan, linked to the official curriculum: prototype, classroom trials, impact evaluation, generalisation, governance.

2. 02

   Study

   Measuring how well AI models suit digital assessment

   Choosing a model for assessment tasks should rest on measured criteria, not on reputation. An indicator of how well open-source models suit e-assessment.

3. 03

   Workshop · IAEA 2026, 51st Annual Conference, Toronto

   Exploring AI — Solutions for Correction and Feedback: Could AI Be on Teachers’ Side?

   Students write constantly, yet individual feedback is rare and often arrives too late. What would change if meaningful feedback could be given far more often, and what should remain the teacher’s?

4. 04

   Talk · IAEA 2026, 51st Annual Conference, Toronto

   Trusting AI in Digital Assessment: Toward Agentic Architectures for Test Design, Administration and Scoring

   How far could digital assessment be delegated to AI? The Big Switch in four stages, the ways models drift, the six priorities to settle before delegating, and what institutions can do today.

5. 05

   Reflection

   Governing a formative assessment resource bank: five questions

   How can a decentralised education system build a shared bank of formative assessment resources that is coherent nationally and owned regionally? A governance model, roles, and a process for sourcing and sharing resources.