Last week I had the pleasure of both presenting at and listening in on the ASAIMH conference in Sydney. It was one of those rare days where I came home with more energy than I left with, which is not something I say often about a conference.
I want to share what I took from the day, because it wasn't any single talk. It was something that built quietly across all of them.
Essentially, what I came to during the day is: The decision is ours, to choose to become involved in shaping how AI is used, or let others decide.
We are the ones who decide how AI gets used in mental health. Not the big international technology companies. Not the regulators, though they matter. Not whoever happens to build the loudest product. Us: the clinicians who sit in the room with people treating patients, who write the reports, who carry the professional and ethical responsibility for what happens next, the researchers who provide the information that helps us know what Ai tools are needed, and validate the technological advances, the founders, passionate about what they are creating to help the clients and clinicians.
Every choice we make from here about what we adopt, what we refuse, what we insist on, and what we will not tolerate - that is what determines whether this technology ends up helping our clients or getting in the way of them. It is not a passive process. Someone is going to decide. It may as well be the people who understand the work.
That sounds obvious written down. It did not feel obvious in the room, because the honest truth is that a lot of our profession has been sitting this one out. Not out of laziness. Out of a very reasonable professional caution, and a sense that this is happening to us rather than with us.
I don't think that's true. I think it's the opposite. But it only stays untrue if we turn up.
##Why I'm optimistic
You may already be able to tell that I am an optimist about this, and the day did nothing to change that.
What I saw was researchers asking genuinely good questions. Not "is AI good or bad" - that's a question for a panel show, not a profession.
The questions were sharper than that. What actually changes in a person when they talk to an AI instead of a human, and can we measure it properly? How accurately do clinicians judge their own judgement about these tools? Why does automated content moderation keep failing the young people it's meant to protect, and what would have to change in the design for it to work? Which subgroups do our averaged findings hide, and can we find them? Those are our questions. They come from our discipline, our methods, our concerns. That is what it looks like when a profession engages rather than reacts.
And the possibility underneath it is real.
AI can accelerate what we know and how quickly we know it. It can sharpen our techniques. It can take work off us that was never clinical in the first place and hand back the hours and the mental capacity that go with it. It can help us reach people we currently miss, and support the ones we already see between the sessions where most of the actual change happens.
None of that replaces us. All of it makes us better at the part only we can do.
The risk if we sit it out
The flip side is straightforward, and it's the part I'd like my colleagues to sit with.
Australia is currently one of the most cautious countries in the world about AI. In the Melbourne Business School and KPMG global study of 48,340 people across 47 countries, we came last on the question of whether AI's benefits outweigh the risks. And only about a quarter of us have had any training in it.
I understand that caution. In a profession that carries statutory responsibility for what it writes and says, scepticism is a professional instinct doing its job. I'd be more worried about a psychologist who wasn't sceptical.
But caution and leadership are not opposites, and caution is not the same as absence. If we stay out of the room, the decisions still get made. They just get made somewhere else, by larger countries with larger markets and companies with no particular understanding of what a psychological assessment is for. And then the tools arrive here fully formed, built around someone else's assumptions about our work, and our only remaining choice is whether to use them or not.
That is a much smaller choice than the one we have now.
We are a small profession in a small market, but we are a genuinely good one. Our training standards are high. Our regulatory framework is thoughtful. We have researchers doing world-class work and clinicians willing to build things. There is no reason at all that Australia should be a downstream adopter of decisions made elsewhere. We have every opportunity to be the ones setting the standard for what safe, useful, clinically sound AI in mental health actually looks like.
But you cannot lead something you have opted out of.
What that looks like in practice
I don't think leading requires anyone to become technical, or to adopt anything they're uncomfortable with. It's much more ordinary than that. It looks like being curious enough to understand a tool well enough to say no to it specifically, rather than generally.
It looks like asking where your data goes and expecting a straight answer. It looks like judging a tool by whether it leaves you with more capacity for the next client, not by how fast it produces something.
It looks like doing a bit of training so your professional judgement about this is actually informed. And it looks like saying out loud what you already do and don't do, so the rest of us can calibrate against something real.
Mostly it looks like turning up to the conversation. Which, last Friday, people did.
Thank you
A conference is only ever as good as the people who speak at it, and this one was genuinely excellent.
My thanks to every one of them, and to ASAIMH for building the room in the first place.
Dr David Camfield — Founder and Managing Director, ASAIMH, who opened the day and whose vision this society is.
Dr Erika Penney — Schema Therapy & Supervision Australia and UTS, for a keynote that gave the whole day its frame: AI in Mental Health: Then,
Dr Oren Griffiths — University of Newcastle, on Quantifying the differences between human and AI therapists.
Dr Biya Tang — Black Dog Institute and UNSW Sydney, on Beyond One-Size-Fits-All: How AI is Transforming Mental Healthcare and Mental Health
Marika Conomos — Ecko Health, on AI Clinical Twins in Healthcare.
Assoc Prof Róisín McNaney — University of Melbourne and CSIRO Next Generation Graduates in AI Mental Health, on Thigh Gaps and Filtered Snaps: translating contextual nuance into content moderation design for eating disorders.
Steven Bernardi — National Manager Mental Health, Healthscope, on The Neuro-Ergonomics of AI: Protecting Clinical Brains and Scaling Resilient Healthcare Systems.
Dr Tuki Attuquayefio — PsychReady and Western Sydney University, on Flight Simulator for Mental Health Training.
Sally-Anne McCormack and Kiera Macdonald — ANTSA, on Between-Session Care: Strengthening Support Across the 10,000-Minute Gap.
Dr Matthew Davidson — UTS, on Assessing attitudes toward the use of Gen-AI in Allied-Health Care Settings.
Steven Read — Chief Product and Technology Officer, Sonder.
And to the panel on AI literacy, current gaps in training, and challenges for mental health professionals: Taimi Allan (Mental Health Commissioner for South Australia), Rebecca Frost (Healthdirect Australia), Dr Anthony Joffe (Black Dog Institute), Dr Erika Penney, Dr Oren Griffiths, Dr Matthew Davidson and Assoc Prof Róisín McNaney.
I spoke on The Future Psychologist Is Now: Playing with AI to Amplify Human Insight and Client Recovery - the founder of Virtuosa AI, an Australian-built report-writing application for psychologists.
If you're a psychologist who has been watching all this from a distance and quietly wondering whether you should be paying more attention: yes. And you'd be very welcome.