The future is here: How the vet profession can navigate AI tech responsibly

Turning up to veterinary practices as a locum and finding different Practice Management Systems, clinical equipment and ways of working is nothing new. Recently, though, a new type of technology has been popping up – Artificial Intelligence (AI)-powered tools, which work in new and different ways. These new bits of software or kit are exciting but have also got me thinking about how they impact the work we do as vets and vet teams.

Don’t always trust the machine

Arriving at one practice in 2025 to find a new piece of kit next to the microscope and finding out it was supposed to assess my pathology slides for me was exciting. I’d seen this tool at BVA Live and had wanted to try out it in the real world. The novelty of seeing the machine do its thing- take my prepared slide, flash lights at it and move it around- was certainly cool. Then getting a report of which bacteria and yeasts were in each ear, looking pretty similar to one I’d get from an external lab, was even more so.

Technology like this all looks great, and it certainly could be a step forward in how vet teams carry out both clinical and non-clinical work. It is, however, important to assess these tools critically. After all, just because the machine can write a report doesn’t mean I can trust it and base my treatment on it. I decided to look at the slides myself for a few weeks. When it was prepared well and the correct part of the slide was analysed, I usually agreed with the AI-generated report – but not always. Would I always agree with a pathologist’s report when it comes back two days later from the lab? Maybe, maybe not. Remembering that these tools are fallible is important and our human instinct to trust the computer needs to be challenged. When using these tools it is on me, the vet, to decide if and when I trust them.

General principles for using AI

That’s why I was thrilled to be approached to chair BVA’s working group on AI in veterinary medicine. We brought together a team of experts from the veterinary and technological fields and consulted widely with more. Over a period of 10 months, we collaborated to bring together a guide for how vets can make the most of this technology for our work and for the animals we care for, while being mindful of the risks and hazards that could crop up.

The BVA policy document has some really useful tools for starting to evaluate how we use AI tools in practice – from overarching principles for approaching AI tech safely, ethically and effectively, to looking at how to question if these tools are worthy of that trust.

This evaluation would certainly have helped a colleague of mine, who, after using a radiography software to assess for cruciate ligament disease on some cases, discovered that although the tool was great for labradors, its skill at detecting cruciate disease in a wonky Frenchie was less so. On questioning the supplier of that tool, it turned out that the training data was missing these smaller, less standard breeds. This didn’t mean it wasn’t good at the job it was trained for – but that it had limitations and these could have been made clearer.

Asking these questions and developing the literacy in how AI works are cornerstones of the policy position’s advice to vets and the resources it will inspire. In both of the cases above and in many more, both the opportunity and the risk are clear to see. Now is the moment where our profession can really have an impact in making sure that this new wave of technology is developed ethically, responsibly and sustainably, with animal welfare at its heart.

More key recommendations of BVA’s AI policy

The policy also calls on companies producing AI tools to make sure that vets are involved in the development of these tools, that we can feedback our experiences and that tools and the way they work are explainable. It calls on regulators and government to think carefully about the gaps in regulation in veterinary technology. If we as vets are going to carry the liability for the outputs from these tools, we need companies to step up and help us understand them.

With such an array of tools on the market (just look at the exhibitor lists for London Vet Show and BVA Live!) the opportunities are wide. Picking the right one for you and your workflow is important. The policy position also includes a helpful Risk Pyramid – as a way to consider how these different tools impact our work, some more than others. This is particularly important when we think about giving tools more autonomy. Vets may want to start lower down and work up or think carefully about how the risks are managed for tools higher up the pyramid.

Overall, the new wave of AI technologies will likely change how we work as vets - whether that’s how we communicate with our clients, test for or diagnose disease, or keep the public safe from zoonotic diseases. Being aware of how AI can help, where it can’t, or shouldn’t, and guiding its development responsibly, is how we make the most of this opportunity.