20 Jun 2025
In the second part of their series on artificial intelligence, Liz Barton and Victoria Johnson look at the role veterinary regulators and organisations have to play in ensuring the responsible use and development of the technology in the sector…
Image: Adobe Firefly / AI Generated
Considerable work has already been done across industries and governments to advise on the risks and necessary guardrails around artificial intelligence (AI).
The EU AI Act and Guiding Principles for Good Machine Learning Practice for Medical Device Development provide solid foundations.
Veterinary regulators are already considering how such questions apply to veterinary practice. While it is widely agreed that robust regulation of AI will be impossible to achieve in theory or practice due to the complexity of the technology and its development channels, there is broad agreement that guidance and support need to be provided for veterinary teams to understand and apply these tools safely and effectively in practice.
The RCVS hosted a round table discussion and continues to engage with stakeholders to determine the best approach. The American Association of Veterinary State Boards (AAVSB) and the American College of Veterinary Radiology (ACVR) and European College of Veterinary Diagnostic Imaging have recently released white papers urging transparency, validation and championing the central role of veterinary professionals at all stages in the life cycle of AI.
Complex questions remain, such as how to assess the risk level for an AI tool, and thus the need for obtaining informed consent for its use. There is more work to be done, and fortunately a wide recognition of the importance of this work and an appetite for exploring what and how to move forward. Pooling collective efforts through collaboration will be important to ensure consistent guidance for applying these technologies globally.
Firstly, the best use of AI will be in solving the biggest challenges of veterinary work and animal health and welfare. Determining this requires veterinary input at every stage of the AI life cycle – from ideation through to follow-up after deployment.
As a values-based and purpose-driven profession that is uniquely charged with safeguarding animal health and welfare, ideation is perhaps the most important stage to involve vets and nurses. We need to be vocal about how we want AI to work for animals and our teams. For this to work, we cannot be seen as putting up barriers and stifling “progress” from a desire to maintain the status quo.
With the current state of affairs, we are also reliant on individual companies choosing to self-regulate, holding themselves accountable to the principles of AI best practice. For example, VET.CT has been seeking to lead the way by publishing its position statement on AI in veterinary radiology as early as 2022, with principles aligned to best practices in medicine, including transparency, accuracy and patient-centred design.
The flip side of asking, rather than requiring, AI developers to adhere to best practice guidelines, is providing the individual user with the wherewithal to adopt these technologies to best effect. Usability is key to successful implementation.
Having AI systems that are useful, understandable, explainable, transparent in their use of data and accuracy, and provide clear guidance on appropriate use cases and reliability of outputs is vital to ensure effective integration into practice. Feedback mechanisms that ensure errors are consistently fed back to the developers are also vital. This is the responsibility of the companies developing AI tools, and it is in their best interests.
The regulated entity is the veterinary professional applying the technology in practice, so individuals should be provided with the tools they need to do this consistently and well. AI tools may soon augment the expertise of the clinician to such a degree that it may be considered negligent not to use them. Education and training will be key here, at all stages of veterinary practice, from universities through to vets at all ages and stages.
The AAVSB gives the analogy that a scalpel is not regulated, but the person wielding it who has the knowledge and skill to do so responsibly. However, the scalpel isn’t telling you if, when, where or how deep to cut. Predictive algorithms and surgical robots do exactly that. The human-in-the-loop signing off on those decisions at each step requires a level of expertise, making it more challenging for new graduates to integrate decision-making aids without the contextual knowledge that comes with seasoned experience. How can we empower vets with varied clinical experience to apply these tools safely and effectively?
Truly, we are in uncharted waters and the debate on how to steer the global, cross-industry juggernaut of AI development is infinitely complex, requiring broad input and opinion from experts in science, tech, business, ethics, philosophy and sociology for starters.
Within the field of veterinary medicine, we need to be actively watching and learning from these discussions, engaging with stakeholders from across the animal health and related industries, and forming our own guardrails for the development of AI and education at all levels on its responsible integration into practice.
It’s encouraging to see the efforts of global veterinary organisations to date. The hope is that the fruits of these efforts result in applicable, agile and robust guidance for both the development and use of AI technologies.
There is no doubt, this is a VUCA (volatile, uncertain, complex and ambiguous) technology unleashed on a VUCA world, which can be – indeed, should be – unsettling. However, there is also enormous potential to improve health and welfare for many more animals with good AI.
For each of us as individuals, our choice is whether to lean into the challenges and help guide AI in the direction that we deem to be most beneficial. Rather than bury our heads in the sand, we need to cultivate curiosity, openness and hope (that most human of characteristics) that AI will work for us and our patients. Truly, this is where our transferable veterinary skillset of values-based, impact-driven, critical thinking and compassionate care could have the biggest influence on the future of veterinary medicine.
