Using Artificial Intelligence (AI) to Improve Data Quality and Analysis in Development Programmes Development programmes increasingly rely on data to design interventions, monitor progress, evaluate outcomes, and demonstrate accountability to stakeholders. In an era where evidence-based decision-making is central to development effectiveness, the quality of data has become a defining factor in determining the success or failure of programmes. However, despite advances in Monitoring, Evaluation and Learning (MEL) systems, many organizations continue to face persistent challenges in ensuring that their data is accurate, timely, consistent, and usable for decision-making (World Bank, 2023). At the same time, the global development landscape is undergoing a rapid digital transformation. Artificial Intelligence (AI) has emerged as one of the most powerful tools for improving how data is collected, processed, analysed, and interpreted. AI technologies are not only transforming industries such as finance, health, and logistics, but are also increasingly being applied in development programming to strengthen evidence systems and improve decision-making processes (UNESCO, 2023). In development contexts where data is often fragmented, incomplete, or manually processed, AI offers a new opportunity to enhance efficiency, reduce human error, and generate real-time insights that support adaptive programming. However, the integration of AI into development data systems must be approached carefully, with strong attention to ethics, data governance, and contextual relevance (OECD, 2024). This article explores how AI is transforming data quality and analysis in development programmes, the opportunities it presents, the challenges it raises, and how organizations can responsibly integrate AI into Monitoring, Evaluation and Learning systems. Bodmando Insights Why Data Quality Matters in Development Programming Data quality is the foundation of effective development programming. Without reliable data, organizations cannot accurately understand the needs of communities, measure the effectiveness of interventions, or make informed decisions about resource allocation. High-quality data ensures that programmes are based on evidence rather than assumptions, improving both accountability and impact (UNDP, 2024). In development contexts, data quality is defined by several key dimensions including accuracy, completeness, consistency, timeliness, and reliability. When these dimensions are weak, the entire evidence system becomes compromised. For example, incomplete beneficiary data can result in exclusion errors, where vulnerable populations are left out of essential services. Similarly, inaccurate monitoring data can lead to misleading conclusions about programme performance. In many development settings, data is collected through multiple channels including surveys, administrative records, mobile data collection tools, and qualitative interviews. While this diversity strengthens the richness of evidence, it also introduces challenges in harmonization, validation, and integration. As a result, organizations often struggle to consolidate fragmented datasets into meaningful insights that support decision-making. This is where AI becomes increasingly relevant. By automating data validation, detecting inconsistencies, and processing large volumes of information, AI can significantly strengthen the reliability and usability of development data systems. OECD (adapted framing in AI policy discussions) Artificial intelligence is not a substitute for human intelligence, but a tool to amplify human capability and improve decision-making Bodmando Insights Common Data Challenges in Development Programmes Despite significant investments in Monitoring, Evaluation and Learning systems, many development programmes continue to face persistent data challenges. One of the most common issues is data fragmentation, where information is stored across different systems that do not communicate effectively with each other. This makes it difficult to generate a unified view of programme performance. Another major challenge is manual data entry, which increases the risk of human error. In many contexts, data is still collected using paper-based tools or manually entered into spreadsheets, leading to inconsistencies and delays in reporting. These delays reduce the usefulness of data for real-time decision-making. In addition, data duplication and missing values are common problems that affect data integrity. Without automated validation systems, it becomes difficult to detect and correct these issues at scale. Furthermore, limited technical capacity within organizations often constrains the ability to analyse large datasets effectively. These challenges highlight the need for more advanced tools and systems that can enhance data quality while reducing the burden on human resources. AI offers a promising solution to many of these persistent problems. Bodmando Insights Understanding Artificial Intelligence in Development Contexts Artificial Intelligence refers to the ability of machines to perform tasks that typically require human intelligence, such as learning, reasoning, pattern recognition, and decision-making. In development programming, AI is increasingly being used to support data analysis, automate processes, and generate predictive insights that inform programme design and implementation (UNESCO, 2023). AI technologies include machine learning, natural language processing, predictive analytics, and computer vision. These tools can process large and complex datasets far more quickly than traditional methods, enabling organizations to identify trends, detect anomalies, and generate insights that would otherwise remain hidden. In Monitoring, Evaluation and Learning systems, AI does not replace human expertise. Instead, it complements it by handling repetitive and data-intensive tasks, allowing practitioners to focus on interpretation, contextual analysis, and strategic decision-making. Bodmando Insights How AI Improves Data Collection and Quality One of the most significant contributions of AI in development programming is its ability to improve data collection processes. AI-powered digital tools can automate data entry, reduce human error, and ensure that data is captured in real time. Mobile-based data collection platforms integrated with AI can also validate responses instantly, reducing inconsistencies at the point of entry. For example, AI algorithms can detect incomplete responses in surveys and prompt data collectors to correct them immediately. This significantly improves data completeness and accuracy. In addition, AI can standardize data formats across different collection tools, making it easier to integrate datasets from multiple sources. In humanitarian and fragile contexts, AI-enabled systems can also support remote data collection, reducing the need for physical presence in insecure areas. This not only improves efficiency but also enhances the safety of field teams. Bodmando Insights AI for Data Cleaning and Validation Data cleaning is one of the most time-consuming aspects of Monitoring, Evaluation and Learning. Traditional methods require manual review of datasets to identify errors, duplicates, and inconsistencies. AI significantly reduces this burden by automating data cleaning processes. Machine learning algorithms can identify patterns in

How AI is Changing Monitoring, Evaluation and Learning Monitoring, Evaluation, and Learning (MEL) has long been a cornerstone of effective development programming. It enables organizations to measure progress, assess impact, and generate evidence for better decision-making. Across sectors such as health, education, agriculture, governance, and livelihoods, MEL systems play a critical role in ensuring that programmes are accountable, effective, and aligned with intended outcomes. However, as development challenges grow more complex and the volume of data continues to increase, traditional MEL approaches are struggling to keep pace. Manual data collection processes, delayed reporting cycles, and limited analytical capacity often hinder the ability of organizations to fully utilize the data they generate. As a result, valuable insights remain underutilized, and decision-making processes are not always informed by the best available evidence. Artificial Intelligence (AI) is now emerging as a transformative force in this space. By enabling faster data processing, deeper analysis, and more adaptive learning systems, AI is reshaping how MEL functions in the development sector. Organizations that integrate AI into their MEL frameworks are better positioned to generate actionable insights, respond to emerging challenges, and improve overall programme effectiveness. AI IN MEL systems The Growing Need for Smarter MEL Systems Development programmes today generate vast amounts of data from multiple sources, including household surveys, field reports, administrative systems, and digital platforms. While this data has the potential to provide valuable insights, managing and analyzing it using traditional methods can be both time-consuming and resource-intensive.     In many cases, organizations collect more data than they can effectively use. Large datasets are stored but not fully analyzed, and important patterns remain hidden. This creates a situation where data exists, but its potential to inform decision-making is not fully realized. AI technologies offer a solution to this challenge. Tools such as machine learning, natural language processing, and automated data extraction allow organizations to process large volumes of data quickly and efficiently. These technologies can identify patterns, detect anomalies, and generate insights that would be difficult to uncover through manual analysis alone. According to the World Bank, data-driven technologies are increasingly shaping how development decisions are made, enabling organizations to move toward more responsive and adaptive systems (World Bank, 2021). As a result, MEL systems are evolving from static reporting mechanisms into dynamic tools that support real-time learning and decision-making. AI IN MEL systems Automating Data Collection and Processing One of the most immediate and visible impacts of AI in MEL is the automation of data collection and processing. Traditional methods often involve manual data entry, which is both time-consuming and prone to errors. In large-scale programmes, this can significantly delay analysis and reduce data quality. AI-powered tools are helping to streamline these processes. Technologies such as Optical Character Recognition (OCR) can extract data from scanned documents, handwritten forms, and images, converting them into structured digital formats. This reduces the need for manual data entry and accelerates the overall data processing cycle. In addition, AI systems can automatically clean and organize datasets by identifying inconsistencies, removing duplicates, and flagging potential errors. This improves data accuracy and reliability, ensuring that analysis is based on high-quality information. Automation not only increases efficiency but also allows MEL practitioners to focus on higher-value tasks such as data interpretation, learning, and strategic decision-making. By reducing the time spent on routine processes, organizations can allocate more resources toward generating meaningful insights. AI IN MEL systems Enhancing Data Analysis and Insight Generation Beyond automation, AI is significantly enhancing the analytical capabilities of MEL systems. Traditional data analysis methods often rely on predefined statistical techniques, which may not capture the full complexity of development programmes. Machine learning algorithms can analyze large datasets to identify patterns, correlations, and trends that are not immediately visible. These insights can help organizations understand which interventions are most effective and why certain outcomes are being achieved. Natural language processing (NLP) tools further expand analytical capabilities by enabling the analysis of qualitative data. Interviews, focus group discussions, beneficiary feedback, and narrative reports can be processed and categorized, transforming unstructured data into actionable insights. This is particularly important in development contexts, where qualitative information often provides critical context for understanding programme outcomes. By combining quantitative and qualitative analysis, AI enables a more comprehensive understanding of programme performance. Michael Quinn Patton, Evaluation Expert Data alone does not create impact. It is the ability to analyze, interpret, and learn from data that drives meaningful development outcomes. AI IN MEL systems Supporting Predictive and Adaptive Programming One of the most transformative capabilities of AI in MEL is predictive analytics. By analyzing historical and real-time data, AI models can forecast future outcomes, identify potential risks, and highlight opportunities for improvement. For example, predictive models can identify patterns that indicate when a programme is likely to fall behind schedule or when certain interventions may not achieve expected results. This allows organizations to take proactive measures, adjusting strategies before challenges escalate. In complex and dynamic development environments, this ability to anticipate change is particularly valuable. Programmes often operate in contexts influenced by economic shifts, climate variability, and social dynamics. AI enables organizations to respond more effectively to these changes by providing timely and relevant insights. Adaptive programming is strengthened through this approach. Instead of relying on periodic evaluations, organizations can continuously monitor performance and make adjustments in real time. This leads to more responsive and effective programmes, ultimately improving development outcomes. AI IN MEL systems Improving Learning and Knowledge Management Learning is a critical component of MEL, yet it is often underutilized. Organizations frequently generate large volumes of reports and data, but these are not always systematically analyzed or used to inform future programming. AI has the potential to significantly strengthen learning processes by organizing, synthesizing, and interpreting knowledge across projects and datasets. AI-powered tools can summarize reports, identify recurring themes, and highlight key lessons learned from multiple programmes. This enables organizations to move beyond fragmented information toward structured knowledge management systems. Insights from past interventions can be captured, shared, and applied to future

Evaluations in the Global South Evaluations in the Global South The context of Program Evaluation in the Global South.The context of Program Evaluation in the Global South. Developing nations are providing increasing evidence that underscores the necessity for improved evaluation frameworks to ensure the long-term sustainability of South-South cooperation. Nations in the global South stress the importance of creating, testing, and consistently applying monitoring and evaluation approaches specifically designed for the principles and practices of South-South and triangular cooperation. Presently, there exists a significant gap in this area, indicating potential shortcomings in the design, delivery, management, and monitoring and evaluation (M&E) of these initiatives. It is crucial to note that the observed challenges do not suggest inherent issues with this form of cooperation but rather indicate possible deficiencies in various aspects (United Nations Office for South South Cooperation, 2018). To fully realize the developmental benefits of South-South and triangular cooperation, especially in reaching excluded and marginalized populations, greater attention must be given to addressing these challenges. As interest in these cooperation modalities grows, stakeholders are calling for discussions on methodologies to assess the impact of these initiatives. However, numerous technical challenges hinder the evaluation process, such as the absence of a universal definition for South-South and triangular cooperation, the diverse nature of activities and actors involved, and varying perspectives on measuring contributions. Various frameworks have been proposed by stakeholders to tackle these challenges. Examples include the framework detailed by China Agricultural University based on China-United Republic of Tanzania collaboration, the NeST Africa chapter’s framework drawn from extensive multi-stakeholder engagement, and the South-South Technical Cooperation Management Manual published by the Brazilian Cooperation Agency (ABC). Additionally, AMEXCID (Mexico) has outlined a strategy for the institutionalization of an evaluation policy, including pilots to assess management processes, service quality, and project relevance and results. While India lacks an overarching assessment system, the Research and Information System for Developing Countries (RIS) think tank has conducted limited case studies to develop a methodological toolkit and analytical framework for assessing the impact of South-South cooperation. In contemporary times, there is widespread acknowledgment that program evaluation initiatives have surged in the Global South. However, the primary focus in the evaluation discourse revolves around narrower aspects such as monitoring and auditing, often driven by the requirements of donors or funders. Moreover, the emphasis on evaluating “impact” often leaves program implementers with insufficient information to enhance program performance or comprehend the underlying mechanisms of program success or failure. This paper explores the gaps and challenges associated with evaluation in the Global South and proposes recommendations to embrace contemporary evaluation approaches that recognize the complexity and context specificity of international development sectors. It also advocates for intentional efforts by researchers, policymakers, and practitioners to build local capacity for designing and conducting evaluations. Program evaluation, the process of generating and interpreting information to assess the value and effectiveness of public programs, is a crucial tool for understanding the success and shortcomings of public health, education, and various social programs. In the Global South’s international development sector, evaluation plays a vital role in discerning what works and why. When appropriately implemented, program and policy evaluation assists policymakers and program planners in identifying development gaps, planning interventions, and evaluating the efficacy of programs and policies. Evaluation also serves as a valuable tool for understanding the distributional impact of development initiatives, providing insights into how programs operate and for whom (Satlaj & Trupti, 2019). Evaluations in the Global South Methodological Bias Currently, impact evaluations employing experimental design methods are considered the gold standard in the international development sector. However, there is a growing recognition among evaluation scholars and practitioners of the limitations of “impact measurement” itself. Some argue that a program may not be suitable for a randomized control trial (RCT) and might benefit more from program improvement techniques like formative evaluation. Scholars emphasize the need to reconsider “impact measurement” as the sole criterion for evaluating program success. The discourse has also shifted towards acknowledging the complexity of causality, advocating for evaluators to be context-aware and literate in various ways of thinking about causality. Despite this, the dominance of methods like RCTs often hinders the use of complexity approaches, even when they may be more suitable. Evaluations in the Global South Human-Centered Design and Development evaluation Developmental Evaluation (DE) is a form of program evaluation that informs and refines innovation, including program development (Patton, 2011). Formative and summative evaluations tend to assume a linear trajectory for programs or changes in knowledge, behavior, and outcomes. In contrast, developmental evaluation responds to the nature of change that is often seen in complex social systems. DE is currently in use in a number of fields where nonprofits play important roles, from agriculture to human services, international development to arts, and education to health. Another technique that has gained salience around addressing complexity and innovation is human-centered design (HCD) –it shares many parallels with developmental evaluation and attends specifically to the user-experiences throughout the program design process. More generally, it involves a cyclical process of observation, prototyping, and testing (Bason, 2017). Although human-centered design is seemingly focused upon initiation (or program design) and evaluation on assessment after the fact, human-centered design and developmental evaluation share a number of commonalities. Both support rapid-cycle learning among program staff and leadership to bolster learning and innovative program development (Patton,2010; Patton, McKegg & Wehipeihana, 2015). Evaluations in the Global South Theory-Driven Evaluation In recent years, theory-driven evaluations have gained traction among evaluators who believe that the purpose of evaluation extends beyond determining whether an intervention works or not. This approach posits that evaluation should seek to understand how and why an intervention is effective. Theory-driven evaluations rely on a conceptual framework called program theory, which consists of explicit or implicit assumptions about the necessary actions to address a social, educational, or health problem and why those actions will be effective. This approach enhances the evaluation’s ability to explain the change caused by a program, distinguishing between implementation failure and theory failure. Unlike