In association with Global Affairs Canada, IOD Parc, Foreign, Commonwealth and Development Office.
Executive summary
Effective use of existing global evaluation evidence is central to the adoption of successful foreign and development policies. Artificial intelligence (AI) technologies have the potential to significantly strengthen our ability to learn from existing evidence, distil key messages, and enhance evidence uptake. Meanwhile, combining AI with robust Knowledge Management (KM) practices and high-quality evidence synthesis can help decision makers and programme leads understand what works and what does not, enabling them to take an evidence-based approach to designing and implementing policy and programmes.
From 24-26 March 2025, experts and practitioners from around the world, including the global south, gathered at Wilton Park to discuss strategies for harnessing the potential of AI and KM in evaluation and evidence synthesis. Participants from a wide range of technical and policy perspectives shared their experiences, forged cross-sector partnerships, and committed to practical actions designed to shape the future use of AI and KM in advancing impactful and sustainable evidence-based interventions [1].
Throughout the discussions, several recurrent themes emerged. These included:
- The opportunities and challenges inherent in using AI for evaluation and evidence synthesis
- The potential for AI to address or entrench existing inequalities across geographies and thematic areas
- The development of principles and standards around the use of AI
- The importance of human supervision and quality assurance
- The value of robust KM structures
- The role of AI and KM in supporting and enhancing evidence uptake
- The need for sustainability and appropriate funding models
- The benefits of greater cross-organisational collaboration
Participants agreed a series of action points and commitments to drive forward the use of AI and KM in evaluation and recognised the need to maintain momentum gathered during this conference via discussions at key upcoming events.
[1] A glossary was issued to participants ahead of the event to ensure that discussions were based upon a common understanding of key terms and concepts. This glossary also informs the language used within this report, and can be found here.