Why healthcare AI must improve work without intensifying burnout, according to ADHA Chief Digital Officer

Healthcare productivity is often measured too narrowly.

Internal efficiency still matters, but in health the larger gains come from what the system enables for clinicians and patients.

In this ADAPT Insider conversation, Peter O’Halloran, Chief Digital Officer at the Australian Digital Health Agency, argues that real value is created when clinicians have better information, unnecessary tests are avoided, and patients move through care with less delay and duplication.

Listen to the full episode on Apple Podcasts, Spotify, or YouTube.

Key takeaways:

• Measure healthcare productivity by the value created for clinicians and patients, not only by internal efficiency gains.
• Build trust in AI through smaller use cases that show clear benefit and keep human oversight in place.
• Redesign work alongside automation so productivity gains improve care without intensifying burnout.

Productivity in healthcare is created at the point of care

Peter’s view is that the most meaningful productivity gains do not sit inside the agency that funds the work, but across the wider health system it helps improve.

When clinicians can access the right information at the right time, they can avoid unnecessary pathology tests, reduce repeat visits, and make faster decisions about diagnosis and treatment.

Those gains compound well beyond any internal efficiency measure because they improve care while also reducing waste across the system.

That is why digital health has such a large upside.

Better information-sharing improves what happens in front of the clinician and the patient as much as it improves the administration behind them.

For a system under pressure from rising demand and limited workforce capacity, that is where productivity becomes most valuable.

Trust sets the pace of AI adoption in healthcare

Healthcare does not adopt AI on promise alone. The tolerance for failure is low, which means trust has to be built carefully.

Peter makes the point that digital systems in health are often expected to be near flawless in ways that other tools are not.

That creates a very different adoption environment from other sectors.

The practical result is that AI has to prove itself through smaller, visible gains.

One example is breast cancer screening, where AI can help prioritise higher risk scans for clinician review.

That supports faster assessment and better outcomes, while keeping the clinician firmly in control of the final judgement.

In his view, that is the right path for healthcare AI, start with uses that improve decision making and safety without asking clinicians to surrender trust in the process.

Capacity gains only matter if the work is redesigned properly

The conversation also sharpens the difference between automation and productivity.

Freeing up time is helpful, but that does not automatically mean work improves.

In some clinical settings, AI can push the most complex or high risk cases to the top of the list, which improves speed and clinical value.

At the same time, it can increase the intensity of the workload if clinicians are left dealing only with the most demanding tasks all day.

AI should help clinicians operate at the top of their scope, but in a way that is sustainable.

If lower intensity work disappears and only the heaviest cognitive load remains, burnout becomes a real risk. The goal is not only to make clinicians faster.

It is to structure work so that productivity gains support better care without eroding the human capacity needed to deliver it.