L-R; Vari Matimba, Sam Kuuku, Solomon Mwije, Patricia Ainembabazi, Ernest Mwebaze

On October 31st, 2025, we launched our landscape study at Datafest Africa in Kampala, Uganda. In this blog, I share my reflections  and explain how we’ll disseminate this research over the coming weeks.

I had the pleasure of moderating a panel that offered insights on how Africa can use AI tools for monitoring and evaluation while making sure these tools actually work for African contexts and language from a practitioner’s lens. The speakers emphasized that Africa needs to develop its own ways of measuring and assessing AI, rather than just copying approaches from other regions, and that policymakers need skills to judge whether AI evidence is trustworthy. A major theme emerged around language. Including African languages in AI systems means more people can participate in data collection. It’s about redefining whose knowledge counts and training evaluators to confidently use, question, or reject AI tools when needed.

What is the study about?

Our study, “Made in Africa: AI Approaches in Monitoring, Evaluation, Research and Learning,” is the first resource of its kind. It explores how AI intersects with data and evaluation work across the African continent. We interviewed 23 AI experts, technologists, evaluators, civil society leaders, and policymakers from Africa and the global evaluation community. The study addresses key questions: What do MERL practitioners in Africa need? What gaps exist? What priorities should guide the development and use of AI tools in this field to enhance evidence-informed policymaking?

Why MERL is Important for Evidence-Informed Policymaking

Monitoring, Evaluation, Research and Learning (MERL) provides the systematic evidence that policymakers need to make informed decisions. It helps governments and organizations understand what works and what doesn’t. MERL creates accountability to citizens and enables learning from both successes and failures. When policies are based on rigorous MERL evidence, they gain legitimacy and public trust.

Most importantly, MERL creates learning cycles that improve interventions, allowing for systematic examination of results and room to adapt accordingly.

Reflections from Datafest 

The panel and audience reactions sparked three discussion threads that we will also be building on as we further disseminate this research. 

1. African governments and real data use for AI regulation
   Participants emphasized that African governments should focus on real-world AI use cases when collecting data and creating regulations. This means governments need to invest in documenting existing AI applications, collecting data on their outcomes, and understanding both their benefits and harms within specific African communities. Regulation should be use-case based to respond to actual applications, not hypothetical scenarios, allowing for more nuanced policy-making.
2. How AI Can Transform MERL in Africa

Artificial intelligence is revolutionizing how practitioners conduct monitoring, evaluation, research, and learning. AI operates at unprecedented scale and speed, processing vast amounts of data that would take human analysts months or years to review. This capability is generating new types of evidence for policymaking, from real-time program feedback to pattern recognition across thousands of documents.  The promise of AI in MERL for Africa is real, but it must be realized on African terms. This means strengthening capacity among African practitioners to assess AI tools critically, adapting technologies to local needs, and most importantly, developing homegrown solutions that reflect African innovation and knowledge. Africa cannot simply import AI tools built elsewhere and repurpose them for local contexts. 

3. Bringing Three Communities Together
   The conversation highlighted a critical need for collaboration among three distinct communities: technologists who build and influence systems, natural language processing developers who create AI models, and MERL  practitioners who assess impact and effectiveness. Each group brings essential expertise but these groups often work in isolation. Without collaboration, we risk building AI tools that are technically impressive but practically unusable, or evaluation frameworks that fail to account for how AI actually works.

Who will benefit from this publication? 

Our study challenges funders, policymakers, and practitioners to move beyond asking “How can Africa adopt AI?” Instead, we ask: “What approaches to AI genuinely serve African communities?”

We answer this question by examining AI, evaluation practice, and evidence-informed policymaking from African perspectives. The study provides practical insights for practitioners working across the continent. Most importantly, the practitioners’ voices identify some specific conditions under which AI can support African development priorities, and when it cannot.

What is next?

Over the next five weeks, I’ll share our research findings and recommendations through a Landscape Study Research Digest. This will also be a mini-5-part newsletter and you can sign up to receive this digest directly to your inbox. Each post will explore the approaches, principles, and practices that should inform “Made in Africa” AI in monitoring, evaluation, research, and learning. This series will take you deeper into what we learned from practitioners across the continent and unpack key themes from the findings.

Whether you are a funder considering investments in AI for social impact, a practitioner exploring ways to center AI in MERL in Africa, a technologist building tools for African contexts, or a policymaker shaping the regulatory landscape, this series will offer a deep dive into key aspects of this emerging space. 

We welcome feedback and comments and, most importantly, what resonates from this landscape study.