In international development, evaluations often report broadly positive findings. Meanwhile, progress towards SDGs has slowed or halted, democracies are backsliding, persistent poverty, the ecological crisis, etc. This strand explores whether, and how far, development evaluation should be seen as being “too positive” and whether development evaluation as a practice is shaped by institutional and methodological biases. At the same time, it highlights approaches that make biases visible, strengthen democratic ownership in partner countries and support genuinely sustainable development pathways. The strand aligns with the conference focus on evaluation for vibrant democracies by examining how bias in data, methods, and institutional incentives affects the credibility and usefulness of development evaluation in politically contested and ecologically constrained contexts. It also asks how AI-enabled tools improve or worsen these problems. It is often said that locally led evaluation and new forms of training can help re centre accountability toward affected populations, support decolonisation of evidence, and address trade offs between growth, rights and sustainability. We will invite examples where this has been achieved. Theory: clarifying how problem framing, power and bias shape evaluative judgement in development cooperation, and how AI and new methods alter these dynamics. Practice: showcasing designs, methods and governance arrangements that move beyond “polite positivity” towards candid, constructive findings. Policy dialogue: bringing together evaluators, commissioners and local actors to discuss how evaluation can better inform strategic decisions on democracy, sustainability and AI use in development.

We invite submissions evaluators or researchers in development and humanitarian sector. We welcome contributions that show how evaluation – including the use or critique of AI – has challenged “too positive” narratives, shifted who defines problems and success, and helped create more democratic and sustainable development practices, including concrete tools, methods, or training models to strengthen such work and help us explore ways in which evaluation practices can be better placed to include marginalised voices, identify what is not working in order to learn how to do better, and strengthen accountability. Younger contributors are particularly invited.