The question hinges on how you define ‘risk’.
For all the talk of promises and threats posed by artificial intelligence, the technology has so far only had slight impacts on general practice.
Often, that talk turns to risk; but risk can be interpreted in many different ways.
Here’s what industry experts told delegates at The Medical Republic’s Burning GP 2.0 conference in Noosa over the weekend.
How risky is AI… when ‘risk’ means taking my job?
GP and AMA Queensland director Dr Maria Boulton acknowledged that AI would likely become better at the knowledge aspect of medicine than most humans.
But that, she said, was not the point of general practice.
“If you have an 87-year-old with multiple illnesses, 13 medications, multiple specialists, facing treatment that may extend their life for six months but will reduce their quality of life – optimisation isn’t enough,” Dr Boulton said.
“Someone needs to understand the whole person and help them to decide what matters. And that is us, the patient’s GP.
“If we look further ahead, AI will almost certainly know more about medicine than any human could possibly learn in a lifetime, but medicine isn’t ultimately an information problem. It’s a human problem.
“In a world saturated with AI, automation, and synthetic relationships, authentic human attention may become extraordinarily valuable. So, a doctor sitting opposite another human being and saying, ‘I’ve known you for 20 years. I know what matters to you. Here’s what I think you should consider,’ may become one of the most valuable things that medicine has to offer.
Managing that risk, she said, starts with training GPs to act as decision making guides, rather than focussing on the amount of information they can retain.
Related
How risky is AI… when ‘risk’ means patient safety?
According to medical writer, educator, and GP Dr Justin Coleman, the answer to this question depends on where AI is being used.
“With a patient in front of you [as a decision support tool], that can be highly harmful,” he said.
“If you have AI giving you the wrong impression about the importance of something in your learning [or CPD], it is theoretically harmful, but it’s a less risky situation.”
MBSPro co-founder Dr Casey Going said he was seeking a “middle ground”.
“There’s a risk of doing something, and a risk of not doing it,” he said.
“I mean, we know that, as GPs, the risk of a scribe that just summarises [a conversation] – that’s pretty low risk. You, as a GP, are going to check it, hopefully, and look at it.
“There is a risk of misspelling and putting in wrong things. I think that risk is low because most GPs that would then read that note, even with the words that are not correct, would probably surmise by the context of the conversation, … what was actually meant to be written there.
“Then you have the risk of clinical decision support, where maybe a new GP, new registrar, or new intern is using a tool to directly manage a patient in front of them without consulting their supervisor or reputable guidelines.
“That is … probably a higher risk than just using a generic scribe to write. So there’s some sort of middle ground.”
How risky is AI… when ‘risk’ means getting sued?
Best Practice chief technology officer Jess White said the practice management software behemoth was concerned that technology was quick outpacing regulation.
“That’s certainly our concern as one of the largest PMS providers, because we want to make sure that you guys are safe, that everything you do is safe,” she said.
Avant commercial lawyer Erica Reed said that clinical responsibility for errors tended to rest with the practitioner, even if an AI tool they used was faulty.
“[If] a patient sues the practice because technically [the practice is] the one holding the records, then do I have to indemnify the practice for its loss and damage? The answer is: most likely,” she said.
“And you’re just going to end up in this big fight. … The thing I would recommend to every doctor in this room is to consider your agreements with your providers.
“Ensure that they comply with Australian law, in the sense that they’re not sending things off out of Australia without permission, they’re not breaching the Privacy Act, those kind of things. And you should be talking to them and getting real written guidance about how you are using decision-making tools and whether you need to disclose those to your patients in a privacy policy.”
Burning GP 2.0 was held at Peppers Noosa Resorts and Villas on 26 September 2026.
