Healthcare organisations across Australia are being encouraged to become “AI-first”. With workforce shortages, rising patient demand and mounting pressures on budgets, it is easy to see why. But too many healthcare organisations are still approaching AI as a technology challenge when, in reality, it is a data and decision-making challenge.
The organisations seeing the greatest value from AI are not necessarily those deploying the most tools. They are the ones building trusted, governed, high-quality information foundations before they scale. In a sector where a single decision can affect a patient’s health, that distinction matters more than almost anywhere.
The race to adopt AI is outpacing data readiness
AI in Australia has moved well beyond the innovation lab. CSIRO’s AI trends for Healthcare 2026 report shows AI is now embedded in real-world clinical settings, from clinical decision support and medical imaging to disease management and personalised care. The models are advancing just as quickly. A study published in Science found a large language model outperformed physician baselines across a series of clinical reasoning tasks.
Capability, however, is not the same as readiness, and the data foundations needed to deploy AI safely and reliably have not kept pace.
Most healthcare organisations continue to operate with fragmented information environments. Clinical records, diagnostic reports, administrative documents and patient communications regularly sit across multiple systems and departments, and often in physical archives. Much of it is unstructured and locked in legacy platforms that are difficult to search, govern or connect to modern AI tools.
The skills and governance picture is just as uneven. Amazon Web Services’ Unlocking Australia’s AI Potential 2026 report highlights that just 26% of workers in healthcare are seen as confident using advanced AI tools, one of the lowest levels of any sector, while only 13% of organisations have a comprehensive responsible AI strategy in place. The appetite for AI is running well ahead of the frameworks needed to use it safely.
For healthcare leaders, this raises a very important and uncomfortable question: is our data prepared for AI, or are we deploying AI across a flawed data estate?
AI is only as trustworthy as the information behind it
AI systems are only as reliable as the data they are trained on, connected to, and governed by. When information is fragmented, duplicated, out of date, or incomplete, AI does not correct these problems. It scales them.
In most industries, the cost of that may be a poor recommendation or a wasted investment. In healthcare, it can have life-or-death consequences. A clinical decision support tool that cannot see the scanned discharge summary sitting in a legacy archive has not suffered a technology failure. It has suffered an information failure, and it is the patient who bears the consequences.
Unstructured information is the biggest blind spot. Some of the most clinically valuable information sits in scanned documents, PDFs, clinician notes, diagnostic reports and archived records that are not readily available for analysis. If information cannot be easily found, understood and governed, AI will draw conclusions from partial datasets, with no signal that anything is missing.
Australia has made significant progress in developing national AI strategies, implementation frameworks and healthcare-specific guidance. What is still missing is practical, sector-wide evidence on how to implement, evaluate and scale AI effectively across healthcare settings.
This issue is already on the radar of policymakers. A report by the Department of Health, Disability and Ageing, Safe and Responsible Artificial Intelligence in Health Care, states that unmanaged data and consent risks need to be managed across the entire AI life cycle. Stakeholders also warned of risks ranging from physical harm to the “long-term pollution of health records”, including AI-generated errors that may be difficult or impossible to reverse. The review even called for a centralised, trusted source of information to help clinicians and consumers make informed decisions about AI products. Trust in AI, in other words, ultimately starts with trusted information.
Four foundations for AI-ready healthcare
Moving from pilot to practice starts with a different question. Not “Where can we use AI?” but “Which decisions do we need to improve, and can we trust the information behind them?” Working back from the decision puts information governance, records management, data quality and accessibility where they belong: ahead of any AI initiative, not behind it.
As healthcare organisations move from experimentation to implementation, the focus should shift from adopting more AI to strengthening the information foundations that enable AI to deliver value safely.
In our work supporting organisations to build the data foundations Australia’s AI standards call for, four fundamentals consistently make the difference:
- Establish trusted and high-quality data. Organisations need a clear understanding of data lineage, ownership and quality. Address redundant, duplicate, and outdated information before it enters AI systems, ensuring decisions are based on reliable information.
- Embed privacy and patient safety by design. Privacy protections cannot be treated as an afterthought. Healthcare providers should ensure sensitive information is appropriately protected while maintaining meaningful human oversight for high-impact patient decisions.
- Build secure and compliant information environments. Data should be stored, managed and accessed in ways that align with regulatory obligations, cybersecurity requirements and data sovereignty considerations, while remaining accessible to authorised users.
- Integrate governance across the organisation. AI governance should not operate separately from existing privacy, records management, cybersecurity and clinical governance frameworks. Aligning these disciplines improves accountability, reduces duplication and creates a stronger foundation for responsible AI adoption.
Trust is the real accelerator
AI will reshape how care is delivered in Australia. But it also raises the stakes on trustworthy data, clear accountability and informed decision-making. Skipping the foundations to move faster is a false economy. Correcting a patient record after an AI system has acted on it is far more difficult and costly than governing that information in the first place.
The healthcare organisations that get the most from AI will not be the ones that adopted it first. They will be the ones that know what information they hold, where it came from and whether it can be trusted. That is how AI delivers on its promise: not as a technology investment, but as a means of improving care and patient outcomes. Before healthcare goes AI-first, it must go data-first.
Viewpoint articles are the author’s opinion, produced without payment or sponsorship.
