Education is upstream of workforce policy. Treat it that way.

A surprising amount of AI workforce policy is really deferred education policy. If education systems had already been designed to produce people who can work with, audit, and redesign intelligent systems, the workforce transition would not feel like a shock. It would feel like a handover.

The framing problem

Most AI-in-education debate still centres on tools: which chatbot, what detection, how much to allow. These are real questions. They are not the interesting ones.

The interesting question is whether the curriculum is still calibrated to the labour market it was originally designed for. In most jurisdictions, it is not. We are teaching the previous century's defaults to students who will spend their careers inside systems none of their teachers have used.

Three shifts that actually matter

• From information transfer to judgement. The scarce skill in an AI-saturated labour market is not recall. It is structured judgement under uncertainty.
• From credentials to portfolios of evidence. The credential-as-signal model decays when the underlying work can be produced by a model. Evidence-based credentialing is harder, but more honest.
• From consumer literacy to systems literacy. It is not enough for the next generation to be able to use AI. They will be asked to govern it. They should be educated accordingly.

The public-interest argument

Any country that treats education as upstream of workforce policy will absorb the AI transition better. Any country that treats it as downstream will spend the next decade running adult reskilling programmes to compensate for decisions it did not make in schools.