Building new ways to discover, understand and use evidence.
ICED is exploring how AI can strengthen evidence generation, synthesis and uptake, while supporting rigorous evaluation of AI’s effects, limitations and relevance across Africa and the Global South.
Three areas of ICED’s AI development. Each one grows out of work ICED already does in research, evaluation and evidence synthesis. These are not a product catalogue.
A broader, AI-enabled evidence platform built on ICED’s existing Evidence Portal and its Evidence & Gap Maps. ICED lists it as coming soon.
Exploring where AI can strengthen evidence generation, synthesis and uptake in ICED’s research, and studying AI’s own effects, limits and relevance in African contexts.
An institutional AI Policy that will guide how ICED uses AI across its work. It is still being drafted.
ICED’s Evidence Portal already publishes interactive Evidence & Gap Maps that show where studies exist and where they don’t. The AI-Enabled Africa Evidence Portal builds on this work and is being developed into a broader AI-enabled evidence platform.
Climate Change Adaptation EGM. 376 records on agricultural interventions for food security and livelihoods in sub-Saharan Africa: 120 impact evaluations, 28 formative, 213 summative and 15 reviews. Bubble size shows how much evidence sits in each cell. Empty cells show the gaps.
Find studies by intervention and outcome on each map.
Filter by country, evaluation type, design and date. Switch between bubble, mosaic, donut and text views.
Cell-wise summaries of what the evidence says, with confidence ratings of high, medium or low.
Records link to the full article where available, and reference lists can be downloaded.
EGM reports, policy briefs and stakeholder events turn the maps into decisions.
These stages describe the existing Evidence & Gap Maps. The AI-enabled portal is being developed to extend them; its features will be announced as they are confirmed.
Seven published Evidence & Gap Maps, built by identifying and coding hundreds of studies, with summaries, reports and protocols.
Testing how AI tools can help process research, generate summaries and support faster, policy-relevant outputs, as discussed at ICED’s E2A 2026 session on AI-powered evidence synthesis.
A broader platform that brings ICED’s evidence together and makes it easier to discover and explore.
Building an evidence map means finding and coding hundreds of studies by hand. The CCA map alone holds 376 records. As the body of evidence grows, AI offers ways to help researchers keep up, while ICED’s specialists stay responsible for judging quality and meaning.
ICED is exploring where AI can strengthen research and evidence workflows it already runs. Choose an area to see the foundation and what is being explored.
AI is being explored to strengthen this work, not to replace it. Nothing here is a separate AI product.
AI at ICED builds on expertise and infrastructure that already exist and are in use today.
Interactive maps of what has been studied and where the gaps are.
Actionable data to support ongoing research projects across the region.
Primary research and systematic reviews, including an update of Masset et al. (2012) on agriculture and child nutrition.
Programme monitoring, impact evaluations, and M&E technical support and training.
Questionnaires, capacity assessment tools, mobile data collection and KoboToolbox training.
Downloadable datasets from IINDWEGE, ALL-IN and cost-benefit research.
The direction ICED’s AI-enabled evidence work is heading. Each stage shows what ICED does today and where AI is being explored. Not every stage is automated or available.
Searches and expert recommendations identify the studies for each Evidence & Gap Map.
AI tools that help researchers find relevant studies across a fast-growing literature.
AI on its own does not produce credible evidence. ICED’s strength is combining it with deep evidence and evaluation expertise rooted in African contexts.
Evidence & Gap Maps, systematic reviews, evaluation reports and project datasets.
Digital-assisted data collection, phone interviews (CATI), KoboToolbox and agile monitoring systems.
AI tools that could help process research, summarise findings and speed up policy-relevant outputs.
Research and evidence synthesis, M&E, and data and AI specialists, working across 24 African countries.
ICED is developing an institutional AI Policy to guide the responsible use of AI across its research, evaluation and evidence work.
The policy has not yet been approved. Its principles will be published here once it is adopted.
How AI systems used in evidence work are governed, how bias is handled, and who owns them.
AI used without grounded, evidence-based understanding can have negative effects. Its use in development calls for awareness of both limitations and potential.
Digital tools, data science and AI should be used responsibly to accelerate evidence synthesis and build public trust.
Rigorous evaluation of AI’s effects, limitations and relevance across Africa and the Global South.
These are themes from ICED’s public work. They are not the policy’s principles, which are still being developed.
“ICED is exploring how responsible use of AI can strengthen the generation, synthesis and use of evidence for better decisions.”