OPINION

Australia is in the middle of an extraordinary AI moment.

Microsoft has committed $25 billion to local digital infrastructure and pledged to train three million Australians in AI skills by 2028.

The Anthropic-Australia Memorandum of Understanding signals a deepening partnership between global AI developers and Australian institutions.

Governments, universities, and corporations are investing heavily in making sure Australians can access, understand, and use AI tools.

This is welcome, necessary, and long overdue.

But there is a question that is not being asked loudly enough: what happens after the training?

Because if history is any guide, even the best-intentioned learning investments will run headlong into one of the most persistent structural challenges in professional life: the knowing-doing gap.

The gap that no investment has closed

Research consistently shows that less than 10 per cent of workplace learning translates into lasting behaviour change.

People attend courses, complete certifications, read the right books, absorb powerful frameworks, and then return to the demands of their jobs and implement almost none of it.

This is not a failure of intelligence or motivation.

It is a structural problem.

The systems we use to deliver learning were never designed to close the gap between knowing and doing.

They were designed to deliver information.

This distinction matters enormously right now, because Australia is about to repeat the same pattern at scale.

When three million Australians complete AI skills training over the next three years, many will understand what AI can do.

Far fewer will meaningfully change how they work as a result.

The knowing-doing gap will not disappear because the topic is AI.

If anything, AI makes the gap wider.

The pace of change is faster, the tools are more complex, and the distance between conceptual understanding and practical implementation is greater than in almost any previous technology wave.

Why personalisation is the missing layer

The traditional response to this problem has been more content.

More modules, more resources, more follow-up webinars.

But the issue is not the volume of learning – it is the absence of a personalised bridge between what someone learns and what they specifically need to do differently.

Consider two professionals completing the same AI productivity training.

One is a senior finance manager at a large corporation, primarily concerned with automating reporting workflows.

The other is a solo consultant in the legal sector, trying to reduce time spent on client research.

They have sat through identical content.

But what they need to do next is completely different.

A generic action plan helps neither of them.

This is where AI itself offers a genuine solution, not just as a subject of training, but as a mechanism for personalising the implementation of learning.

The same technology that powers the tools Australians are being trained to use can be applied to the training itself, generating contextually relevant, role-specific action steps that reflect the learner's actual situation, goals, and constraints.

What Australia's AI investment needs to get right

For Australia's AI skills investment to deliver on its promise, the sector needs to think beyond access and toward implementation.

That means asking harder questions of training programs: not just what are participants learning, but what are they doing differently six weeks after they complete the course?

What mechanisms are in place to support the translation of new knowledge into changed behaviour?

Who is accountable for that outcome?

PwC's recent research found that 74 per cent of AI's economic value is captured by just 20 per cent of organisations.

The gap between those organisations and the rest is not primarily one of access to tools or training.

It is one of implementation.

The organisations pulling ahead have built systems and cultures that close the loop between learning and doing.

They have made implementation a discipline, not an afterthought.

Australia has a genuine opportunity here.

As a nation investing heavily in AI capability, we are also well-positioned to invest in what comes after the training, the systems, tools, and design principles that ensure learning leads to action.

That is where the real economic value lies.

Billions spent on AI skills will deliver a better-informed workforce. But a better-informed workforce and a more productive one are not the same thing.

The difference is implementation.

And implementation requires more than good intentions.

It requires design.

If Australia's AI investment is going to move the needle on productivity, it needs to close the knowing-doing gap.

That is the harder problem.

And it is the one worth solving.