Framework for Responsible AI in M&E
Put human judgment at the centre of AI-assisted evaluation.
FRAME is a practical framework for using generative AI responsibly across the evaluation lifecycle. It helps evaluators harness AI’s genuine capabilities while honouring evaluation’s responsibilities to the people our work affects.
why frame is different
Not "responsible AI." Responsible people.
Most AI ethics talk puts responsibility on the technology. FRAME starts somewhere else. AI systems have no consciousness, no intention, and no moral capacity. They cannot be responsible in any meaningful sense. Responsibility lives in the people and institutions that create, deploy, and govern these systems.
FRAME positions evaluators as crucial intermediaries between technical capability and societal outcome, and gives them a structured way to keep human judgment in charge.
Evaluators keep the deciding role — AI supports, people judge.
the framework
Five principles at a glance
Five principles, applied across the seven phases of an evaluation, held together by cross-cutting governance that runs throughout.
Principle 1
Accountability Architecture
Establish clear lines of human accountability before any AI is deployed.
Principle 2
Stakeholder Engagement
Give affected communities a genuine say in when and how AI is used.
Principle 3
Epistemic Integrity
Keep evidence rigorous; every Al claim must trace back to source data.
How FRAME works
Five principles. Seven phases. Continuous governance.
Five principles, applied across the seven phases of an evaluation, held together by cross-cutting governance that runs throughout. The result is a framework you can adapt to any organisation, project scale, or evaluation type.
5
Principles
7
Lifecycle phases
37
Checklist items
6
Governance mechanisms
Grounded in NIST, the EU AI Act, OECD, UNESCO, and ISO/IEC 42001.
Who it's for
Built for the people doing the work
Evaluators & MEL teams
Putting AI to work on real evaluations, with a method and a checklist you can apply on a live project.
Commissioners & managers
Putting AI to work on real evaluations, with a method and a checklist you can apply on a live project.
Students & capacity-building
Learning responsible practice and connecting it to the AI for MEL course.
From the field
Where FRAME came from
FRAME emerged from live cases documented in the Routledge volume From Algorithms to Evidence and from practitioners in The Development CAFE’s AI for MEL e-learning course, who have tested AI tools across Asia, Europe, Latin America, and Africa.
It grew out of real cases — with their messy complexities, unexpected failures, and practical workarounds — not from theory alone.
Ready to apply FRAME to your next evaluation?
Download the FRAME checklist, or enrol in the AI for MEL course to put the framework into practice.