ID: 33911
Artificial intelligence (AI) is transforming educational technology, yet most AI tools in education lack a critical foundation: they have no formal understanding of how curriculum knowledge actually fits together. AI tools can discuss topics fluently, but cannot reliably sequence learning, identify what a student needs to know before tackling a new concept, or verify that a lesson genuinely addresses the standards or objectives it claims to. They guess because, typically, national curricula have existed only as documents written for humans, not in any form that digital tools can meaningfully interrogate.This is a solvable problem. At Oak National Academy, a UK government-backed provider of free teaching resources used in three in four schools in England [1], we are building an answer. With over 100,000 open-licence resources spanning 17 national curriculum subjects, including more than 13,000 lessons, we hold a distinctive position: a large-scale, publicly trusted body of curriculum-aligned content, all freely available [2]. This session presents our work to build a knowledge graph that encodes curriculum structure in a form that AI tools can read, query, and reason about, along with the lessons we have learned so far from releasing it openly to the sector.Our knowledge graph is a structured map of how curriculum knowledge connects, specifying which concepts must come before others, how topics in one subject relate to another, and what a sequence of learning actually looks like from a pupil’s perspective. When that map is machine-readable and openly available, it transforms what AI tools can do. Rather than guessing at curriculum relationships, AI tools can query structured data. The result is AI tools that genuinely support learning progression, plan well-sequenced, coherent curriculum materials, and identify gaps in available resources. This provides the solid foundations for safe and effective AI in education in any national context. We have begun sharing our early knowledge graph capability [3] with a range of external organisations, including AI developers and edtech companies, to develop and stress-test this shared foundation.England's forthcoming curriculum refresh offers a concrete illustration of why this matters. As the curriculum changes, AI tools trained on old frameworks risk undermining rather than supporting learning. We have been supporting the Department for Education in creating a digitised curriculum, designed with structured, machine-readable capabilities, such as knowledge graphs, in mind. When available, this will significantly increase the depth and richness of our knowledge graph, enabling even more reliable and sophisticated AI applications across the sector.Open education principles are at the heart of this work: every organisation rebuilding curriculum structure from scratch is duplicating effort. A common, openly licensed foundation redirects that energy toward the applications and innovations that actually benefit learners. In this session, we will share our experience and learnings, including the architectural decisions made, the standards chosen, and our licensing strategy, to enable any institution seeking to make its domain knowledge machine-readable to build upon this work.
Speakers 
Interim CEO, Oak National Academy
John Roberts is the interim CEO of Oak National Academy, the UK government-backed provider of free, open-licence teaching resources used in three in four schools in England. A co-founder of the organisation, he previously led the product and engineering team behind Aila, the UK's...
Read More →
Wednesday October 7, 2026 4:20pm - 4:50pm
EDT
6 DR4
MIT Samberg Conference Center, 50 Memorial Drive, Cambridge MA 02139 USA
