Priorities and Recommendations for Using Artificial Intelligence (AI) to Improve Equid Health and Welfare
Format of work:
Journal Article
Event presented at / Journal Name:
Animals
Speaker / Contact Author's Name:
Sarah Freeman
Speaker / Contact Author's E-mail Address:
sarah.freeman@nottingham.ac.uk
-
Research aim:
This study aimed to establish stakeholder consensus on how artificial intelligence (AI) should be developed and applied to maximise benefits for equid health and welfare in the UK. Specifically, it sought to identify key welfare priorities, priority areas for AI development, barriers to implementation, and potential solutions. By integrating perspectives from across the equid sector, the study aimed to provide a strategic framework to guide future research, funding, and responsible deployment of AI technologies in equid welfare contexts.
-
Background:
Artificial intelligence is increasingly applied in animal health and welfare, including for equine care, where it is used for behaviour monitoring, disease detection, and clinical decision support, among other applications. However, development of AI tools is not always guided by welfare priorities, despite the likelihood that such a strategy would deliver greater benefits for equid welfare than the current approach. There is also limited regulation of AI in the veterinary sector, alongside ethical, technical, and interpretative challenges. This study addresses the need to align AI development with real-world welfare priorities and stakeholder perspectives.
-
Approach:
The study used a stakeholder-driven Delphi approach. Forty-one participants from across the equid sector (including veterinarians, owners, researchers, welfare organisations, and AI developers) attended a structured workshop to generate ideas on welfare priorities, AI applications, barriers to implementation, and solutions. These ideas were refined into statements and evaluated through three rounds of online Delphi surveys, with consensus defined as ≥75% agreement. In total, 134 statements were assessed, with 106 reaching agreement. This iterative process enabled systematic prioritisation of key issues and development pathways.
-
Key finding:
Stakeholders identified unmet behavioural needs, poor management practices, and limited owner understanding as key welfare concerns. AI was considered most valuable for monitoring equid behaviour, health and overall wellbeing, and improving population traceability. However, participants identified several barriers, including poor-quality or biased data, challenges in model validation, and difficulties interpreting AI outputs. Participants also highlighted risks such as over-reliance on AI and erosion of human expertise, emphasising the need for careful and responsible implementation.
-
Industry or policy relevance:
The findings provide a clear, stakeholder-informed framework to guide the development of AI tools in the equid sector. They highlight the need for policy that ensures AI systems are evidence-based, transparent, and ethically developed, as well as codes of practice to guide responsible use. The study also underscores gaps in current UK regulation of AI. More broadly, it facilitates the alignment of technological innovation with welfare priorities - an approach which could be applied to other areas of animal health and welfare.
-
Route for practical application:
The study identifies priority applications for AI in equid welfare, including development of tools for monitoring behaviour, health, and wellbeing, as well as improving population-level traceability and supporting management decisions. Tools should be supported by high-quality, representative datasets, robust model validation, and user-friendly interfaces. Education may help to ensure correct interpretation of AI outputs. Collaboration across sectors, including data-sharing initiatives, may facilitate more reliable systems. Adoption of tools is likely to be influenced by affordability, accessibility, and integration into existing practices.
-
Confidence in findings and next steps towards realising impact:
This study provides a robust, consensus-based assessment of priorities for AI in equid welfare, drawing on a diverse group of stakeholders and a structured Delphi methodology. The iterative design and clear agreement threshold strengthen confidence in the identified priorities and recommendations. The consistency of findings with previous research on equid welfare further supports their validity. A number of limitations should, however, be considered. The participant group was relatively small and skewed towards highly educated individuals, researchers, and those already engaged with AI or welfare issues, which may limit generalisability to the wider equid-owning population. Participant dropout across survey rounds may also have influenced final consensus outcomes. Additionally, findings are based on expert opinion rather than empirical testing of AI tools. Future work should focus on translating these priorities into validated AI tools and evaluating their real-world impact on welfare outcomes. This includes improving data quality, developing standards for validation and regulation, and establishing clear codes of practice alongside expanding education to increase AI literacy. Long-term impact will depend on interdisciplinary collaboration, responsible governance, and ensuring that AI complements, rather than replaces, human expertise.
Funders:
Animal Welfare Research Network (AWRN), funded by the Biotechnology and Biological Sciences Research Council (BBSRC) Technologies for Animal Welfare Seeding Award
Links to Open Access Publications or DOI:
Citation:
Young, P. L., Hyde, R., Douglas, J., & Freeman, S. L. (2026). Priorities and recommendations for using artificial intelligence (AI) to improve equid health and welfare. Animals, 16, 1082.
