Journal of Development Effectiveness

Rapid and Responsive Evidence Synthesis for Sustainable Development

Journal of Development Effectiveness Special Issue: Rapid and Responsive Evidence Synthesis for Sustainable Development

The landscape of evidence synthesis is undergoing a fundamental transformation, driven in large part by the rapid advancement of artificial intelligence (AI). AI-enabled Digital Evidence Synthesis Tools (AI-DESTs) now offer the prospect of dramatically accelerating and reshaping multiple stages of the systematic review process — from search and retrieval to citation screening and data extraction to synthesis and dissemination. These developments are creating new possibilities but also pressing questions about how to ensure that speed does not come at the cost of rigour, transparency, or equity. As AI capabilities continue to outpace the frameworks and standards developed to govern their use, the field faces an urgent imperative: to respond with robust methods, clear reporting norms, and principled guidance for responsible synthesis.

Yet technology alone cannot make evidence synthesis truly responsive and impactful. Alongside the rise of AI, there is growing recognition that synthesis processes are shaped by the institutional, financial, and knowledge environments within which they are produced and used. In sustainable development contexts, this raises important questions about whose evidence is represented, how synthesis priorities are determined, and how disparities in access to data, infrastructure, language, and AI-enabled tools may influence evidence production and use.

Addressing these challenges requires a more fundamental reorientation toward the needs, timelines and decision contexts of policymakers, practitioners, funders, and communities. The most effective synthesis products are increasingly understood to be those designed with, not merely for, the people and institutions they are meant to serve. Responsive models of evidence production, including co-produced reviews, demand-driven evidence mapping, and iterative engagement with policymakers and practitioners throughout the synthesis process, are becoming an increasingly important complement to technological innovation.

Together, these developments — AI-enabled efficiency and stakeholder-centred co-production — are reshaping what it means for evidence synthesis to be fit for purpose in sustainable development.

This special issue of the Journal of Development Effectiveness invites contributions that explore the full range of innovations, technological, methodological, and relational, that are reshaping evidence synthesis in development policy and practice. We define "rapid and responsive" not merely as a matter of speed, but as a broader commitment to synthesizing evidence in ways that are better calibrated to the rhythms of decision-making across governments, donors, philanthropies, and implementing organizations working toward sustainable development.

Contributors are encouraged to engage critically with the opportunities and trade-offs presented by a range of approaches — including living systematic reviews, automated screening tools, machine learning-assisted data extraction, stakeholder-engaged and co-produced synthesis, AI-assisted evidence mapping, and related innovations.

Papers that interrogate the quality, equity, and ethical dimensions of accelerated synthesis processes are especially welcome, as are contributions from low- and middle-income country contexts that have historically been underrepresented in methods-focused literature. We are interested in a wide range of contribution types, including but not limited to:

• Methods development and testing of AI tools: Benchmarking and validation studies evaluating the validity and reliability of large language models for search and retrieval, citation screening or data extraction; comparisons of AI-assisted versus human-led review stages; validation models assessing the long-term reliability and stability of automated workflows against 'algorithmic drift' in continuous update cycles; protocols for benchmarking machine-learning performance across heterogeneous data structures (e.g., qualitative and multi-sectoral evaluation data); frameworks for auditing AI tools for bias, hallucination, reproducibility, epistemic inclusion or error in synthesis workflows across multi-sectoral data and diverse study designs (e.g., mixed-methods and qualitative data).
• Other methodological innovations: Development and validation of rapid review protocols; assessments of the sensitivity and specificity trade-offs in abbreviated search strategies; novel approaches to evidence mapping; advances in meta-analytic or qualitative synthesis methods suited to rapid timelines and policy responsiveness.
• Applied examples and case studies: Examples and accounts of rapid or living evidence products commissioned and used in real policy or programming decisions; reflections on what was gained and lost in accelerated processes; documentation of stakeholder-engaged or co-produced synthesis in low- and middle-income country contexts.
• Knowledge translation and evidence use: Innovations in communicating and disseminating synthesis findings for policy and practice audiences, including interactive evidence platforms, and adaptive evidence products designed to support evidence uptake in dynamic implementation settings.