Those Who Teach Every Day Also Need to Be Taught: AI Education’s Dual-Expertise Gap

On AI teacher training, the market’s default answer is: teach teachers to use AI tools. Run a workshop, issue an attendance certificate, expect the classroom to change. The answer fails quietly — because it trains at right angles to the actual gap.

The real gap spans two professions. On one side, technical understanding: how models are trained, how data shapes output, why generative AI errs so fluently. On the other, cognitive development: what a ten-year-old’s working memory can hold, how reflection is scaffolded, under what conditions a child outsources thinking wholesale. Teachers strong in the first mostly come from computing and find the second unfamiliar; teachers deep in the second span every subject and hesitate before the first. Both are professionals. AI education sits in the no-man’s-land between them. Harder still is the structure of time: teachers teach every day. They are not postgraduates with two spare years; nor will students wait — generative AI is already in the homework, the chat window, the child’s sense of self. An honest training system must accept this constraint: it must let teachers grow while teaching, starting from strength — not open by declaring teachers deficient and prescribing wholesale retraining.

UAICS training is designed from exactly there. First, it is organised by the four certification blocks (BDG / F / A / E), each stage self-contained — a Badge Stage teacher need not understand backpropagation; an Engineer Pathway teacher must; the path is visible and chosen. Second, it distinguishes two destinations: cross-curricular teachers carry AI into their own subjects, specialist teachers build the depth — breadth spreads through the first, height stands on the second, and neither is secondary. Third, every block shares one core module, Learn with AI: recognising cognitive load, preventing cognitive hollowing, coaching metacognition — because whatever the age band, what a teacher most needs to know is the same: what is happening to the child in front of the tool.

Education spent a century learning to train teachers of mathematics. AI education does not have a century — but it can decline to repeat one mistake: treating teachers as vessels to be filled. Teachers are the strongest lever this transition has. Provided someone builds them a proper road.