AI , no Mr Fix it
If we Don’t change how care works, AI won’t fix it
We spoke to Frank Lorfino, here's our write up of the conversation, we began by asking for a brief bio...
I am a Senior Research Fellow at the University of Sydney’s Brain and Mind Centre, where I lead digital mental health research. My work focuses on how technology, including AI, is shaping the future of mental health care.
What is clear to me is this. AI will be part of that future. The real question is not whether we use it, but how we use it.
"things" not "thing"
There are already technologies available that we are not using well enough.
- Systems that can support decision-making at the point people enter mental health care.
- Systems that can help triage, help people navigate services, and help match care to need.
This is not theoretical. These tools can make care more efficient, reduce cost, and get people access to the right support much earlier. If we do not adopt them, that is also a risk. People are already presenting in distress. People are presenting to hospitals with self-harm. Continuing to do things the same way and expecting different outcomes is not a solution.
At the same time, we cannot approach AI as a replacement for care.
There is a tendency to frame this as a choice between human care and technology. That is the wrong framing. We should be asking a different question.
There are already clear examples of where AI is useful.
Administrative tasks are an obvious one. Note-taking systems and AI scribes are already reducing workload for clinicians. That matters, because clinician burnout is real, and reducing that burden improves the system overall.
There are also areas like psychoeducation, where AI can deliver information in ways that people prefer. There is evidence that people respond well to how AI summarises and presents information.
And there are low-risk interventions, such as coaching around lifestyle changes, where AI can be effective. These are not marginal gains. They are practical improvements.
The challenge is not just technological. It is structural.
We have a system that is built around episodic care. People enter the system, receive a set number of sessions, and then exit. That structure is tied to funding, to policy, to how services are delivered.
But it does not reflect how people experience mental health. People’s needs change over time. Progress is not linear. What we need is a system that can respond dynamically.
Technology makes that possible. We can track outcomes. We can monitor progress. We can adjust care in real time. Increase support when needed. Reduce it when it is not.
We have seen that change is possible.
During COVID, telehealth was adopted rapidly. That shift happened because it was clearly valuable. The same is true for some AI tools. When the benefit is obvious, adoption follows. But where change challenges the underlying structure of care, it slows down. That is where we are now.
The most important direction is personalisation.
Mental health care does not have the same diagnostic tools as other areas of health. We do not have clear tests. The process of finding the right care is often slow and based on trial and error. That takes time, and it can be frustrating for people.
At the moment, decisions are often made based on brief interactions. A session once a week, or even once a month. These are snapshots. From those snapshots, we try to make decisions about care. We can do better than that.
The future can be bright
Not a system where AI replaces clinicians, but one where technology supports more personalised, more continuous, and more effective care. AI will be part of that. It will exist in multiple forms, working alongside clinicians and within systems.
