New AusDoc research finds that 76% of Australian GPs use at least one AI tool in their day-to-day work, despite 79% identifying accuracy and hallucination risk as a major concern. Among GPs concerned about accuracy and hallucinations, 83% are already using AI.
The assumption behind most healthcare AI go-to-market plans is straightforward: doctors are cautious, doctors are risk-averse, so the first job is to win their trust before they’ll touch the product. New Australian GP research conducted by AusDoc on AI adoption in general practice says that assumption is wrong, and it’s costing vendors time they don’t need to spend.
Among the Australian GPs and practice leaders surveyed, 79% rate accuracy and hallucination risk as their leading concern about AI in clinical practice; the single biggest worry on the list, ahead of privacy, medico-legal exposure, cost, patient safety and loss of clinical skill. It is, by any measure, a serious and specific reservation about the technology doing the wrong thing at the wrong moment in a patient’s care.
“Concern re hallucinations / patient privacy and I’m happy with the process I have. I don’t want to spend the end of each session proof checking previous consultations.”
— GP, on their biggest concern about AI
It is also not stopping them. 83% of the GPs who hold that exact concern use an AI tool anyway.
The fear is specific. The adoption is broad.
What makes this finding useful rather than just counterintuitive is how precisely GPs can articulate what worries them. This isn’t diffuse technophobia.
What concerns Australian GPs most about AI?
Concern breaks down into a clear, rank-ordered list:
- Accuracy and hallucination (79%),
- Privacy (71%),
- Medico-legal risk (70%),
- Cost (60%),
- Patient safety (57%), and
- Loss of clinical skill (42%).
GPs are not undecided about AI. They’ve formed a specific, itemised risk assessment and then adopted the technology regardless.
Adoption itself is already mainstream, not early-majority. According to the AusDoc GP AI Adoption Research (2026), 76% of GPs use at least one AI tool day-to-day.
- ChatGPT leads at 55% usage, followed by:
- Lyrebird Health (34%)
- Gemini and Copilot (23% each)
- Heidi Health (22%)
- Claude (18%) and
- Perplexity (6%)
Clinical scribe tools alone, Heidi and Lyrebird combined, are already embedded in 44% of respondents’ daily practice, sitting in on consultations and generating notes in real time. That’s not a pilot. That’s infrastructure.
Perhaps the sharpest data point: 45% of GPs who rate themselves “not at all familiar” with AI are already using a tool. Self-assessed literacy isn’t gating adoption either. Doctors don’t seem to be waiting to understand AI, feel confident about it, or trust it fully before putting it to work. They’re using it, forming views on it in parallel, and continuing to use it anyway.
“I constantly challenge ChatGPT by referring back to original literature. Lyrebird is the best thing to hit general practice since computer-generated prescriptions… I may take a little longer checking the validity of what’s been transcribed, but the quality of the recorded notes far exceeds any notes from the past.”
— GP, describing how they already use AI in practice
That’s the paradox in one clinician’s own words: managing the accuracy risk themselves, in real time, while remaining an enthusiastic daily user. Not waiting for the concern to be resolved but working around it.
AI didn’t win GPs over. It just showed up first.
This is the part vendors should sit with. The prevailing trust-building playbook, explain the technology, address the concerns, then earn adoption, assumes doctors are waiting on the other side of that process. They’re not. They’re already inside the tools, already forming specific, well-reasoned objections, and continuing to use AI at the same time. Adoption arrived first; trust-building is happening after the fact, or not at all.
That’s a different opportunity than the one most vendor strategies are built for. The job isn’t to convince a sceptical, non-adopting GP that AI is worth trying; three in four already believe that, functionally, with their behaviour if not their stated confidence. And GPs are asking for this directly: offered a list of reasons to attend further AI training, “understanding safety and risk” ranks second only to practical, immediately usable tools (49.5% vs 53.5%), well ahead of curiosity or CPD points.
“To stay current, and understand the risks.”
— GP, on their strongest reason to pursue further AI training
The job is to be the vendor who resolves the itemised concern list for doctors who are already customers of somebody’s AI tool, or about to be.
Broad awareness campaigns and “AI is safe” reassurance messaging are aimed at a doctor who, on this data, barely exists. The doctor who does exist is already using AI and is evaluating vendors, consciously or not, on which one addresses their specific concerns credibly. Accuracy transparency, explainability, and medico-legal reassurance aren’t trust-building exercises ahead of a sale. They’re the sale.
What this means for AI vendors
For any AI provider building or marketing a tool into Australian general practice, three things follow directly from this data:
Don’t wait for trust to resolve before you go to market
It won’t, and doctors aren’t waiting for it to. Build for a clinician who is already using a competing or adjacent tool, not one who needs to be convinced AI belongs in the consult room.
Compete on the concern list, not on awareness
Accuracy claims that can be substantiated, medico-legal positioning, and privacy handling that can be explained in plain terms will move a GP further than a broader trust or safety campaign, GPs are asking for exactly this:
“Robust real-world data on clinical safety, efficacy and usefulness.”
— GP, on their strongest reason to pursue further AI training
Watch the scribe category specifically
With 44% of GPs already using an AI scribe, this is the most mature adoption point in the market and the clearest evidence that specific, high-utility use cases beat general-purpose trust-building every time.
How does AusDoc support those who want to connect with doctors?
AusDoc gives pharma brands and other healthcare organisations a trusted, editorially independent way to engage with 62,000 AHPRA-verified Australian doctors through clinical education, promoted content and accredited CPD that supports better-informed clinical practice and patient care.
Explore our solutions → here
Help Australian GPs build practical AI skills they can use immediately in your practice.
Sources:
Help shape AI training for Australian doctors and practices, August 2026 (n = 207 MED / 205 GP)