Reflecting on the M&E Universe in the age of AI 

The M&E Universe is a free resource of over 100 short papers on a wide range of M&E, topics. Drawing on discussion at INTRAC’s event at Glocal Evaluation Week 2026, Nigel Simister, Vera Scholz, Soukeyna Ouedraogo and Paul Knipe consider the ethical and practice implications of AI in M&E. How does the M&E Universe land in the age of AI?  

The M&E Universe is an online platform for curated M&E practitioner knowledge, drawing on years of practitioner-focused, participatory learning and real-world examples. It sits within the broader values-driven approaches of INTRAC, our partners and our network. It was designed specifically to support civil society organisations and to focus on ongoing M&E rather than formal evaluation. It aims to give people the information they need to understand M&E debates, to be practical and based on what works and what is ‘good enough’. It aims to encourage evaluative thinking and common sense, and to avoid or explain jargon wherever possible. It is balanced rather than pushing one agenda, and it recognises other good resources and points people towards them where relevant. 

Generative AI tools are now widely used not just to conduct actual monitoring and evaluation but also on the capacity development side – to synthesise M&E knowledge, answer specific questions, and produce guidance at speed. In principle, a practitioner can now skip original sources as well as hard-won experience entirely and go straight to an AI-generated summary of tools, methodologies, or suggested M&E plans. But what is that summary actually drawing on? 

When one of our facilitators asked an AI tool to generate guidance on M&E for capacity strengthening portfolios – a topic covered in the M&E Universe – and then asked what sources it had used, the response was disarmingly honest. “I produced guidance that sounds authoritative,” the response stated, “is reasonably coherent, and carries no traceable accountability. You cannot check it against experience. You cannot know whose knowledge it erased in the process of synthesising a consensus view.” 

The inability to check against experience and erasure of knowledge poses a problem for rigorous and ethical M&E. AI tends to reproduce consensus on how best to approach M&E tasks and challenges – a consensus that, one might argue, has often led to M&E that does not really support social transformation, inclusiveness, or the redistribution of power. Based on what we know about how AI tools get trained, the most popular models have in all likelihood absorbed predominantly well-documented institutional frameworks, from the UN, World Bank, OECD-DAC, USAID and large INGOs, as well as academic literature from minority-world universities. Conversely, many areas pertaining to practitioner, unorthodox, and values-driven M&E will be underrepresented. These include oral traditions, indigenous knowledge systems, grey literature, informal practitioner guidance, content in minority languages, wider tacit knowledge within the field, and, critically, honest accounts especially of what doesn’t work.  

At our Glocal Evaluation Week 2026 session, we put three questions to participants. The responses revealed consistent and compelling themes. 

On what is at risk of being lost 

Participants pointed to the living, participatory dimension of community-based MEL. These include the personal relationships that bring the most marginalised into the room, the contextual intelligence that shapes data collection, the tensions, cultural dynamics and nuances. There was real concern about a shift from participatory to extractive approaches, and about AI tools relying heavily on minority-world produced sources and the loss of authentic, marginalised voices and knowledge. 

On the continuing value of curated knowledge spaces 

Participants valued human-driven, accountable guidance, alongside resources with named practitioners behind them who adhere to professional integrity. AI is well suited to completing tasks, such as answering a question, drafting a framework, summarising a methodology. But developing as an MEL practitioner involves something slower and harder to shortcut. Professional development means building the judgement to know which approach fits which context, to recognise when something feels wrong, to navigate organisational realities around evidence, and to make imperfect decisions around priorities and trade-offs. Strengthening these skills comes both from actually doing the work but also from engaging over time with the experience of others and a vast body of pre-existing knowledge. The M&E Universe is not just a reference resource: it is a learning resource, one that shapes how practitioners think, not just what they do in a given moment. 

On what responsible guidance would look like 

Practitioners called for honest, critical guidance on what AI can and cannot do in M&E. This includes acknowledgement of AI’s limitations, and guidance calibrated to small civil society organisations with limited M&E budgets. Practitioners also wanted to see ethical principles around consent and data sharing. Crucially, that guidance needs to be shaped by voices from the majority world, speaking to the importance of relational, facilitative approaches to bring in marginalised voices and ensure they are heard. It is also essential that M&E practitioners critically review AI generated outputs, conscious of its limitations and uses.  

There was broad consensus that M&E Universe-style guidance and the accountability and experience it is based on are still needed in the age of AI. This follows in the footsteps of evaluation theorists such as Thomas Schwandt, who emphasised that evaluation should be less marked by scientific rules and technical excellence and more by “practical wisdom”.   

This session was one input into a forthcoming M&E Universe paper on AI. This resource will incorporate these questions and findings around practice and ethics into a simple guiding paper and do what AI cannot: give practitioners accountable, values-grounded, and experience-based guidance.  

As for the wider M&E Universe, there are options we are exploring to ensure it continues to support monitoring and evaluation practitioners in the future. How could the Universe weave connections between people doing this work over time and across places? Could we integrate an AI tool to make it easier for users to query, interact with and document reflections continuously? 

We invite you to join us and continue this discussion. Get in touch: pknipe [at] intrac.org