Australia will not get an AI productivity boom by training workers alone

Australia's regional adoption gap makes this management issue more urgent. Smaller firms outside major metropolitan areas often have fewer specialist staff and less room for failed experiments. They need simple management routines that help them test technology in bounded workflows, compare outcomes and expand only when the evidence is strong.

Public policy can help by shifting some attention from generic AI awareness to manager-level implementation skills. Industry programs can teach workflow selection, measurement, escalation design and change management alongside technical use. Training providers can require participants to bring back a real workplace experiment rather than a certificate. Business advisers can ask for evidence of reduced rework, improved service or better decisions before calling an AI project successful.

This approach also offers a way out of Australia's increasingly polarised AI argument. The country does not need to choose between "AI will transform everything" and "AI is mostly hype." Both claims are too broad to guide action.

The better question is whether a particular organisation has managers capable of turning a tool into a better way of working.

Australia can train millions of people to use AI. Without managers who know how to redesign work, measure outcomes and stop bad deployments, those skills will produce more activity than productivity.