AI will either build human capability or quietly erode it. That's not a prediction — it's a design decision. Most schools are making it by accident.1
Fifteen years engineering education that demonstrably works — curriculum, assessment, and the unglamorous systems underneath. The gap between what learning science knows and what schools actually do is the problem I keep returning to. AI has made it urgent.
Right now that means a research programme on the neuroscience of school design — five structural projects on the ways schools are built against what we know about the brain — published in the open at The EdJournal, with the diagnostics and advisory that turn it into decisions schools can defend. Finishing an MSc in Applied Neuroscience at King's College London.
most schools have already decided. they just haven't noticed yet.
§1 Selected work
every claim carries an evidence tier. where it thins, it says so.
Fig. 1 — Flagship · research programme, live
The State of the Evidence
A 138-page review of the cognitive, biological, affective and developmental neuroscience for the design of schools — 10 chapters, 22 graded principles, 406 references. From it, five structural projects, each on a different way schools are built against what we know about the brain.
Five ways schools are built against the brain
01
The Shape of the Day
The day is built around bus routes and staff contracts, not adolescent sleep and attention.
02
Manufactured Disadvantage
The architecture produces the differences it then measures, and records them as the child's.
03
The Assessment Engine
The cycle masses retrieval at exactly the point the evidence says it does least good.
04
The Device
The most consequential attention variable in a school, handled as behavior rather than designed as structure.
05
Designing for the Developing Brain
Provision built around a single notional learner, while children vary systematically across the years and within them. In build.
it stages AI by developmental readiness, not by year group or by task
Fig. 2 — Body of work · evidence synthesis + design-based research
AI by Design
A research-and-design programme for when, not whether, AI belongs in a young person's learning. Six layers carry it from evidence to proof: the argument (The Formed Mind), a model that stages use across three developmental phases — foundations forming, judgment forming, judgment formed — then the practice, the school policy, and an evaluation design built and waiting on its first school. Every claim carries an honest mark against the evidence beneath it, including the joints that don't yet hold.
One line, evidence to proof — follow it down, or tap a stop
the grading bias turned out to be the interesting part
Fig. 3 — Research · adversarial, on purpose
Stress-testing AI on high-stakes assessment
Trained an AI grading system on real exam papers and anonymized student responses, then went looking for where it was wrong. Where it breaks is the finding.2
building the tools, not just writing about them
Fig. 4 — Shipped & in build · classrooms, not slide decks
Teaching assistants & apps
A chemistry assistant producing the full teaching content stack, from lesson plans to differentiated tasks to guidance through complex practical work, built and then rebuilt around what teachers actually needed once they had it in front of them. And now a set of learning tools in active development, running on the same evidence the writing argues from, because I would rather build the thing than only describe it.
§2 Observations
every number here has a story where it nearly didn't work
350+
practitioners trained across a multi-site trust
70%
of teacher planning time removed via AI workflows
4
continents reached by a venture built from nothing
80%
of a 250-student cohort met or beat targets, two years running
1
TEDx talk arguing grades don't matter. Still the argument I'd start with.
Advisory, speaking, collaboration — or an argument about where AI and learning are actually going. I'd rather have the argument. If it's a decision your leadership is carrying now, book a call. Otherwise, write below.
