Interrogating the “Black Box”: Trust, Perception, and AI in Monitoring and Evaluation

As artificial intelligence becomes embedded in monitoring and evaluation (M&E) systems, it is transforming how evidence is generated, interpreted, and used in decision-making. Yet many of these systems operate as “black boxes” - producing outputs that are not always fully transparent or easily understood by those who rely on them.

This session focuses on how these shifts are experienced within evaluation practice itself, from how evidence is interpreted to how it is ultimately used in decision-making. It moves beyond a purely technical lens to examine how AI is reshaping the relationship between evidence, judgement, and decision-making - and what happens when evaluation systems begin to rely on forms of evidence that are not fully interrogable.

Participants will leave with:
• A clearer understanding of what “black box” AI means in evaluation practice, and its implications for evidence use
• Insight into how AI is reshaping the evaluation function and its role in decision-making
• A grounded perspective on the opportunities and risks of using AI in M&E systems, particularly in public sector contexts
• A deeper appreciation of the interaction between human judgement and algorithmic systems
• Practical considerations for responsibly integrating AI into evaluation while maintaining credibility and transparency