Australia is not moving fast enough in the effective use of generative AI (genAI), and professionals across the therapeutic goods sector – including regulatory affairs, clinical research, and medical technology – risk missing the moment. The time to engage with AI is now, not after everyone else has figured it out.

While AI has existed for decades, the launch of ChatGPT by OpenAI in late 2022 marked a major turning point, making powerful generative tools available to the masses. Since then, early adopters globally have been testing, learning, and building new ways to embed AI into their workflows. Meanwhile, many Australian organisations – including those in healthcare and life sciences – have hesitated, unsure how to balance opportunity with risk.

But genAI is not just another technology trend. For ARCS members, it represents a once-in-a-generation shift that could reshape how regulatory submissions are compiled, how trial data is analysed, and how patient-facing information is generated and translated.

Where We Are vs. Where We Could Be

Around the world, genAI is being used to improve diagnostic accuracy in medical imaging, automate literature reviews for pharmacovigilance teams, and accelerate the review of clinical trial applications. In Europe, regulators are exploring how AI can help manage pharmacovigilance signal detection. Some Australian medtech startups are using AI to streamline device documentation and reduce time to market.

Yet most professionals in Australia remain unsure where to start. An EY survey last year showed 65% of CEOs acknowledge AI's potential to improve efficiency, but only 37% are actively investing – far behind the global average of 47%. The Australian therapeutic goods sector, with its strict compliance demands and complex data environments, may be particularly cautious.

Adding to the hesitation is a critical skills gap. The Tech Council of Australia estimates the country needs an additional 200,000 AI-capable workers by 2030 – a figure that seems staggering when many ARCS members are already stretched managing evolving regulatory requirements and clinical obligations.

From Curiosity to Capability

One ARCS member organisation, a mid-sized CRO based in Victoria, recently began trialling genAI to assist with drafting investigator brochures and converting clinical protocols into plain language summaries. Using AI tools such as Microsoft Copilot, they found they could produce first drafts in a fraction of the usual time. Importantly, the content still requires expert review – but the efficiency gains are significant.

However, without a clear AI strategy, even the best tools risk being underutilised. “We had staff saying, ‘We should use AI,’ but nobody could explain how or why,” said one project lead. “We needed a framework – a roadmap – before jumping in.”

This story is typical. Organisations recognise the potential, but lack confidence in how to adopt AI safely and purposefully.

Getting Organisational Buy-In

For AI adoption to take root, it must be led by people who understand both the opportunities and limitations of the technology. It’s not just an IT issue – regulatory, clinical, medical affairs, and operational teams all have a role to play.

A key step is defining who “owns” AI strategy. In some ARCS member companies, the responsibility has been given to data governance teams. Others have embedded AI champions within departments, encouraging small pilot projects to build confidence and identify value.

Crucially, organisations must also invest in education. AI literacy cannot be developed through a single lunchtime webinar. One ARCS-affiliated biotech has created a learning pathway that includes AI ethics, prompt writing, and risk assessment – tailored for non-technical staff. Their staff now participate in weekly “prompt labs” where they share best practices and success stories.

Mitigating Risk While Unlocking Potential

Of course, any meaningful use of AI must be accompanied by robust risk controls. Privacy, data protection, bias, and transparency are serious concerns. But AI can also be a tool for managing risk.

One Australian startup working with ARCS is trialling a conversational AI tool designed to flag early signs of burnout in clinical site coordinators by analysing anonymous feedback and meeting transcripts (with consent). The AI provides insights to line managers and HR teams, allowing for timely intervention.

This type of use aligns with new SafeWork psychosocial hazard codes – and demonstrates how AI, when deployed thoughtfully, can actively contribute to a safer, more responsive workplace.

A Human-Centred Approach

The key to effective AI adoption is maintaining a human in the loop. That’s the principle behind Dragonfly Thinking, a tool developed by ANU researchers to support complex decision-making. It doesn’t tell users what to do, but helps them explore different paths, risks, and trade-offs. Several policy teams and industry bodies are using it to navigate regulatory reform scenarios.

ARCS members dealing with global regulatory strategies, for instance, could use similar tools to model the potential impact of differing US, EU, and Australian requirements on product development timelines.

Governance in an AI Era

Using AI responsibly is not just about policies. It’s about building governance structures that support continuous learning, transparency, and ethical oversight.

Some ARCS members have formed AI steering groups to explore use cases, challenge assumptions, and share knowledge. Even small teams can start by identifying champions and developing minimum standards for privacy, data handling, and transparency.

Trying to force AI into existing IT risk frameworks rarely works. AI presents unique risks – from model drift to hallucinations – that require specific mitigation strategies. That’s why organisations are now turning to AI-specific governance frameworks, like the one co-developed by the Human Technology Institute at UTS, which outlines eight core elements of responsible AI use:

1. Clear roles and responsibilities

2. Skills and a learning culture

3. Fit-for-purpose governance structures

4. Ethics-aligned policies and principles

5. Operational practices and controls

6. Stakeholder engagement

7. Infrastructure readiness

8. Ongoing monitoring and reporting

This model isn’t just for tech companies or boards – it’s applicable to any organisation navigating AI, including regulatory teams, education providers, and clinical networks.

Stimulating Innovation Through Trust

The Australian Government’s proposed mandatory AI guardrails focus on high-risk applications. While the intent is to build public trust, the real aim should be making AI more trustworthy.

The therapeutic goods sector already operates within some of the most stringent regulatory frameworks in the world. This gives ARCS members a strong foundation to lead on responsible AI use – provided they lean in, stay curious, and commit to learning.

As one senior clinical operations manager said after testing a new AI-powered risk monitoring tool: “It doesn’t replace my expertise – it sharpens it.”

For ARCS members, the message is clear: AI is here to stay. Whether used to draft plain-language summaries, support trial recruitment, or improve training programs, it’s not about replacing people. It’s about empowering them – safely, ethically, and effectively.