Sandra Leaton Gray, a professor of education futures at UCL, argues that Britain’s push to test educational AI risks measuring what is easy to count while missing what it may mean for children’s learning and development. Her proposed remedy is an independent National Observatory for AI and Child Development.
Watch Desk analysis
What happened
In an opinion article published by Times Higher Education on 29 October, Leaton Gray points to England’s £24 million EdTech Testbed, which is due to involve more than 1,000 schools and colleges, and ministers’ ambition for AI tutoring to reach up to 450,000 disadvantaged pupils a year. The programme aims to establish which technologies improve educational outcomes.
Leaton Gray argues that product trials can test selected outcomes but may miss longer-term effects, such as what happens when children routinely delegate intellectual tasks to AI, or use AI companions as part of ordinary life. She also warns that the same technology could supplement excellent teaching in one school but stand in for scarce human expertise in another.
She proposes combining experimental and longitudinal research with children’s own experiences and scrutiny of commercial claims. She says she raised the concerns in oral evidence to the House of Commons Education Select Committee on 20 October. Read Leaton Gray’s article in Times Higher Education.
Why it matters
A system can improve a measured result without answering whether it expands children’s opportunities, changes how they learn or widens differences between schools. Those are not side questions if AI is becoming part of the infrastructure of childhood, as Leaton Gray argues.
The £24 million testbed gives the debate a practical edge: public money is being used to evaluate educational technology, while the article questions whether the measures used will capture the effects that matter most. It is a case for widening the research, not a claim that educational AI has already harmed children.
Our read
Leaton Gray’s strongest point is that a tidy evaluation can still leave the important question out of frame. Testing whether a tool improves a chosen outcome is useful; it is not the same as understanding what happens when children grow up with AI woven into learning and relationships.
Her observatory proposal is an argument, not an established government plan. But it offers a clear challenge to policymakers: assess not only whether a product works, but what it is replacing, for whom and over what timescale. The spreadsheet deserves a place at the table. It should not get the only chair.
What to watch
- Whether the Education Select Committee examines the developmental and inequality questions raised by Leaton Gray.
- What outcomes the EdTech Testbed will measure, and whether its findings address longer-term effects.
- Whether her proposal for a National Observatory for AI and Child Development gains support.
Discussion spark: Should education authorities judge classroom AI mainly by measurable short-term gains, or require evidence about its longer-term effects on children and inequality too?
Sources and evidence
- Source update (29 October 2026, 00:00 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.