From Pilot Projects to Institutional Systems Toward Strategic Readiness for Artificial Intelligence in Evaluation within Developing and Conflict-Affected Contexts

This E-Panel aimed to examine the evolving role of artificial intelligence in evaluation and decision-making, moving from experimental applications toward more institutionalized approaches rooted in trust, governance, and accountability. The panel brought together complementary perspectives from academia, development partners, policymakers, and supreme audit institutions, creating a space for reflection on how AI can strengthen evaluation systems while remaining aligned with public sector values and institutional responsibilities. The discussions showed that AI is no longer limited to isolated pilot initiatives or technical curiosity. Across the presentations, it was presented as a growing set of tools and approaches that can support data analysis, knowledge management, decision support, and learning processes, while also helping institutions improve efficiency and responsiveness. At the same time, the speakers made clear that the value of AI depends not only on its technical performance, but also on the quality of the data, the clarity of the institutional framework, and the presence of human judgment to interpret and validate results. A central message of the panel was that trust must remain at the heart of any effort to integrate AI into evaluation systems. Participants stressed that institutional users must be able to understand how AI tools operate, what assumptions they rely on, and how their outputs are used in practice. This is especially important in evaluation, where credibility, transparency, and methodological rigor are essential to ensure that evidence informs policy decisions in a responsible way. The panel therefore highlighted the need for governance mechanisms that can guide adoption, define safeguards, and ensure that AI strengthens rather than weakens institutional accountability. The contributions from ARABOSAI, IsDB, IDSC, and ENAP illustrated different but complementary pathways toward this transition. Supreme audit institutions brought a governance and oversight perspective, development partners highlighted innovation and operational relevance, while research and training institutions emphasized the importance of building skills, methods, and critical thinking. Together, these perspectives showed that institutionalizing AI in evaluation is not only a technological challenge, but also an organizational and cultural one that requires collaboration across sectors and disciplines. The panel also underscored the importance of capacity building. Several of the presentations suggested that AI should be understood as a tool that augments human expertise rather than replacing it, particularly in fields such as monitoring, evaluation, and learning. This means investing in training, strengthening digital literacy, and supporting practitioners so they can use AI critically and effectively. It also means encouraging institutions to develop internal standards and practices that make AI use consistent with ethical principles and public interest objectives. Another important lesson from the discussion was the need to situate AI within broader institutional priorities, including inclusivity, relevance, and evidence use. The panel showed that AI can contribute to more timely, adaptive, and informed decision-making, but only if it is embedded in systems that are designed to serve people and public outcomes. In that sense, the transition described during the session was not simply about adopting new technologies, but about rethinking how institutions generate, validate, and use evidence in a rapidly changing environment. Overall, the E-Panel demonstrated strong interest in the future of AI for evaluation and governance, as reflected by the participation of an average of around 80 participants (166 inscriptions) throughout the event. The level of engagement confirmed the relevance of the topic and the value of continued dialogue among practitioners, researchers, and institutional actors. More importantly, it showed that the conversation is moving beyond experimentation toward a more mature reflection on how AI can be responsibly integrated into evaluation systems that are trusted, useful, and sustainable

As a follow-up to this E-Panel, a series of webinars will be proposed with the panelists to allow sufficient time for each of them to further present their experience, insights, and expertise on the subject. These webinars will be organized by ENAP as part of its collaboration with GEI under the TAQYEEM program.