Research · Status: Research

Evidence before scale.

We want educational AI to earn trust through evaluation, clear limitations and work that others can inspect. Our research programme is at an early stage.

No papers published yet.

Open Leer Foundation does not currently have papers, preprints, datasets or benchmarks to list. Our first publications will appear here when they are ready.

When work is ready, this page will include authors, dates, methods, limitations, safety considerations and links to available artefacts.

Questions worth answering carefully.

These themes describe the direction of inquiry, not completed findings.

Grounded educational answers

How source quality, retrieval and citations can support useful explanations.

African language learning

How quality and safety can be evaluated across multilingual educational contexts.

Human-AI learning relationships

Where assistance helps teachers, families and providers without displacing judgement.

Local AI

What can run usefully on affordable, reused or intermittently connected infrastructure.

Safety and evaluation

How to test uncertainty, escalation, safeguarding and harmful failure modes before release.

Access and inclusion

How disability, device constraints, data cost and digital confidence should shape the system.

Research should be inspectable.

Researchers and practitioners can contact us about rigorous, appropriately governed collaboration.