In healthcare, it’s our moment to lead.
On June 15, Australia’s updated AI policy reached its first milestone. From that day, every in-scope Commonwealth agency must set out a strategic position on how it will adopt AI: where the opportunities are, and how the organisation intends to embrace them.
It is a genuine moment, and I welcome it. Government is signalling, clearly and at scale, that AI belongs in the work of the public service, including in health. After years of caution, that ambition is the right call. We should want our hospitals, our health departments and our care systems to use these tools well.
What comes next is just as encouraging, if we read it right. The deeper accountability requirements of the policy, the registers of where AI is actually being used, the named accountable owners, the impact assessments, and mandatory AI training for staff, become compulsory in December. So the next six months are not a gap. They are a runway: a window for those of us in healthcare to build the governance capability that will make all this ambition safe, fair and trustworthy by the time it scales.
I noticed this runway, because I have been in this space for a long time.
In 2019 I brought a clinician-led global movement on artificial intelligence in medicine, to Australia for the first time. Back then many people still spoke about AI in healthcare as a question for the future. It was not. It was already in our diagnostics, our administration and our care pathways. Our message that year was that AI was improving patient outcomes now, not one day in the future. Seven years on, that is more true than ever, and the lesson has only sharpened: these systems do their best work when they are adopted and governed together, in step.
That is why I am optimistic about the path Australia has chosen. We have decided to lean on our existing, technology-agnostic frameworks and to trust agencies to govern their own use of AI, rather than wait for a single binding law. In January, two major health reviews found those frameworks fit for purpose. It is a flexible, fast-moving model, and it can absolutely work. But it only works on one condition, and it is worth saying plainly.
A decision to govern through judgement rather than statute does not lessen the need for governance. It concentrates it. The safeguard becomes the people in the room: the people who decide which AI tool a health service buys, how it is deployed, who it is tested on, and who is empowered to ask the hard question. In most sectors that is a serious responsibility. In healthcare it shapes who is diagnosed correctly, who is treated well, and who is kept safe. Get the people right, and a light-touch model is a gift. It lets good clinicians and good leaders move quickly.
So the work ahead is about people, not just policy. We sometimes talk about AI as a purely technical problem to be solved in the code. It is not. AI is socio-technical. Data is selected by people. Models are configured by people. Systems are deployed inside human organisations, and the benefits and the risks are felt by patients, clinicians and communities. AI inherits its strengths and its blind spots from the data, the assumptions, and the rooms where design and procurement decisions get made. Widen those rooms, and we widen what AI can safely do.
There is real opportunity here for clinical AI specifically. A tool trained on populations that reflect all of us serves all of us better. A global mapping of health AI policy recently found that equity and safety are often treated as sub-clauses of broader governance rather than priorities in their own right. In a model built on judgement, we get to do better than that, by design, simply by deciding who sits at the table.
Who is in the room?
Which is the exciting question this week. Who is in the room?
In Australia, as elsewhere, women are still underrepresented in technical AI roles, in AI leadership, and in the forums where AI is procured, deployed and overseen across our health system. First Nations people, people with disability and culturally diverse communities are too often absent from the very decisions that affect them most. We call this the influence gap. The good news is that it is entirely closable, and the next six months are the moment to close it.
The strategic plans agencies file this week are not the finish line; they are an invitation. Government has, in the best sense, handed the health sector the chance to lead on safe and responsible AI: through our boards, our procurement panels, our clinical leaders, and the people who build and buy these systems. The question is simply whether we will use the runway to December to put enough of the right people, with enough diversity of perspective, into those rooms. I think we can.
That is why we built the Govern the Future Scholarship Programme, and why the AI Governance Practitioners Programme, designed by Dr Kobi Leins, exists: to bring capable, diverse, trained custodians of AI into those rooms while the decisions are still being made. Because the real work does not happen only in the code. It happens in the thinking, and in who is invited to do it.
For seven years we have said that AI in medicine is here now, not in the future. The chance to govern it brilliantly is here now too. We are all custodians of AI, and in healthcare that has never been a slogan. It is the job, and this week is a wonderful week to take it up.
I remember your AI in Healthcare conference Luli. It was the beginning of an exciting journey, punctuated by the challenges of the COVID pandemic in between.
I agree that the recent milestones are an important and significant step. However the best solution is ongoing research and education and transparency of its current and future usage.