Doing More with Less while Not Losing Trust: How Should Evaluation Standards Evolve.....

AI had the potential to rapidly reshape how evaluations were planned, conducted, synthesised, and communicated, particularly in low-resource and humanitarian contexts where pressure to “do more with less” was acute. However, it simultaneously introduced significant ethical, methodological, and governance risks that directly challenged the profession’s credibility, independence, and trustworthiness. AI added the greatest value when it reduced duplication, lowered transaction costs, and freed human expertise for judgment, ethics, and contextual interpretation, rather than substituting core evaluative functions or obscuring accountability. This created an urgent need for professional standards to evolve, not by endorsing AI wholesale, but by defining where it was appropriate, where it was not, and what competencies and safeguards were required. Drawing on their own use cases, this roundtable explored questions around how evaluation standards, competencies, and institutional frameworks could adapt to ensure that AI strengthened, rather than undermined, ethical practice, methodological rigour, and public trust. The discussion was particularly relevant for evaluators and commissioners grappling with how to translate high-level AI ethics and governance principles into concrete evaluation standards and professional practice.