Towards a new paradigm of “datafied development”: Epistemic,infrastructural,economic,and geopolitical domains

Introduction

Development is broadly defined as progressive change or improvement in human wellbeing through multifaceted approaches, such as infrastructure enhancement, trade and investment, access to decent work, skills development, and social inclusion. Development is increasingly entangled with data-centric practices and digital infrastructures like artificial intelligence (AI) systems, digital platforms, the internet, computers, and mobile phones (UN Global Pulse, 2012; World Bank, 2021). For example, more than a decade ago, Jeffry Sachs called mobile phones as “the single most transformative technology for development” (quoted from CNN, 2011). While the impact of mobile phones on economic development is still debatable, recent years have witnessed several controversial efforts to integrate big data practices and AI systems in development initiatives, including the contested partnership between World Food Programme (WFP) and Palantir (World Food Programme, 2019), the implementation of biometric system Aadhar in India (Singh and Jackson, 2021), and the UNHCR and Microsoft partnership on refugee registration for humanitarian operations (Lodinová, 2016), among many others.

These cases warrant deeper engagement with the relationships between development and datafication. In particular, there is a need to examine how various digital innovation initiatives intertwine with conflicting visions of enhanced human wellbeing and how these initiatives often perpetuate historically harmful power structures. This special issue offers a sustained engagement with these issues across geographical, disciplinary, and methodological boundaries. It introduces the concept of “datafied development” to critically examine human development in relation to contemporary data-intensive technologies.

First, it acknowledges development's historical roots, theoretical underpinnings, and its political and ethical implications while striving for social justice through praxis (see Carmody, 2019; Hart, 2010; Schuurman, 2000; Willis, 2020). Secondly, it emphasizes the critical role of extraction, not just of natural resources but data extraction as a key force in the politics of development which generates new forms of dependencies (see Couldry and Mejias, 2019). Thirdly, it explicitly moves away from techno-determinism where new forms of technologies are seen as tools to deliver development agenda. Instead, it draws inspiration from science and technology studies (STS) to recognize advanced digital and data-intensive technologies not as artifacts but socially shaped and hence the outcomes cannot be determined (MacKenzie and Wajcman, 1999).

Finally, we acknowledge Gilian Hart's (2001: 650) powerful distinction between “big D” Development and “little d” capitalist development: “big D’ Development defined as a post-second world war project of intervention in the ‘third world’ that emerged in the context of decolonization and the cold war.” The “little d” refers to the “the development of capitalism as geographically uneven but spatially interconnected processes of creation and destruction, dialectically interconnected with discourses and practices of Development” (Hart, 2010: 119). Over the past several decades, practices of Development have been reimagined, undergone reinvention, and implemented, including more recently around the “state capitalism” and the rise of the Majority World (see Alami et al., 2021). In extending this line of thought, our assertion is that varied practices of contemporary datafication (e.g. digital ID, digitization of public records or integration of large language models or chatbots into organizational processes often driven via and by the Big Tech, the state, and the likes of World Bank, the International Monetary Fund, World Economic Forum, etc.) represent “Big D” and the “little d” are the uneven and contradictory outcomes it produces spatially as evidenced through the contributions in this issue. Thus, we argue “datafied development” signals a shift towards a new paradigm whereby datafication is now uncritically and widely linked with progressive change by many scholars, lawmakers and practitioners alike. Yet, as we note in the subsequent sections of this introduction as well as the contributions of this special issue, datafied development is not just about the superficial adoption of technology to solve some of the most pressing issues in the world (e.g. poverty, inequality, global warming), primarily in the Majority world or the Global South. It is also not just about advancing and proposing evidence-based policy interventions designed to facilitate rapid technology adoption. Essentially, it is not datafication for datafication sake. In the next section, after briefly tracing historical roots of datafied development, we provide its conceptual framework and identify key domains within this emerging paradigm.

Datafied development: A genealogy and conceptual framework

Historically, the important role of digital technologies in development has been widely acknowledged since the early 2000s (e.g. Heeks, 2010; Kleine and Unwin, 2009; Walsham, 2017 among others). Collectively, this body of work is commonly referred to as information and communication technologies for development (ICT4D) literature. Within this scholarship, the spread of digital technologies is seen to have wide-ranging social, political, and economic impacts (for a critique of ICT4D debates see Murphy and Carmody, 2015). Concomitantly, observers have also pointed to data-driven practices and infrastructures (e.g. AI, digital platforms) as important contributors to solving global challenges, highlighting the significance and value of data in this endeavor. For example, the United Nations referred to data as the “lifeblood of decision-making and the raw material for accountability” (United Nations Independent Expert Advisory Group, 2014). Similarly, OECD (2015: 17) noted data-driven innovations’ “potential to enhance resource efficiency and productivity, economic competitiveness, and social well-being as it begins to transform all sectors in the economy, including low-tech industries and manufacturing.” Thus, helping “address social and global challenges, including climate change and natural disasters, health and ageing populations, water, food, energy security, and mass urbanisation.” As a result, another strand of scholarship around “data for development” discourse emerged with a focus on big data practices (e.g. applying computational techniques to analyze large datasets such as mobile-banking transactions or online content for analysis to curate patterns and trends) (see Hilbert, 2016; Kshetri, 2014; Mann, 2018; Taylor and Schroeder, 2015; Weber, 2017). It must be noted here that within this literature there is an acceptance of the problematic relationship between digital/data and development, including coloniality and adverse incorporation (see Akbari and Masiero, 2025).

Nonetheless, there is now a latest narrative around AI for good (e.g. Taddeo and Floridi, 2018). While AI and big data practices are considered essential for development, emphasizing their potential for producing fairer, safer, and more sustainable societies (e.g. Sustainable Development Solutions Network and Open Data Watch, 2015; also see Kleine and Unwin, 2009), there are real dangers of entrenchment of existing unequal patterns of power (e.g. Arora, 2016; Couldry and Mejias, 2019; Lehdonvirta, 2022; Madianou, 2019; Taylor, 2017). Then there are risks of ethics washing and ethics dumping (Abu-Elyounes and Gentelet, 2021; Iazzolino and Stremlau, 2024; Ochigame, 2019). Private companies are—perhaps unsurprisingly—the strongest advocates of furthering data-driven solutions in development initiatives. In 2014, in a contested bid to become a basic infrastructure, Mark Zuckerberg invoked the discourse of human rights to call connectivity as basic human right. Over the years, the Bill and Melinda Gates foundation has been a key player in a range of Global Development initiatives, advocating technologies such as CRISPR, AI and even sanitation technologies as transformative tools in reducing poverty and increasing health. More recently, Sam Altman, the CEO of OpenAI (the creator of ChatGPT) recently touted AI to be “the greatest technology humanity has yet developed” which will reshape our economy and society. As a result, our understanding of how AI and big data-centric practices are influencing development is continuously evolving.

The concept of “datafied development” unites existing scholarships on big data studies (e.g. Dalton et al., 2016; Iliadis and Russo, 2016; Kitchin, 2014; Medrado and Verdegem, 2024; Thylstrup et al., 2021; Valente and Grohmann, 2024) with development studies (e.g. Chang and Grabel, 2004; Corbridge, 2017; Desai and Potter, 2014; Horner, 2020; Lewis, 2019; Melber et al., 2023), particularly its expansive subfield of ICT4D (e.g. Carmody, 2025; Masiero, 2022). While research on the links between big data practices and development exists (e.g. Anwar and Graham, 2022; boyd and Crawford, 2012; Cieslik and Margócsy, 2022; Raval, 2021; Taylor and Schroeder, 2015), it often takes place within, rather than across, the fields of development studies and big data studies. The proposed special issue bridges this gap. It does so by bringing critical research on data and algorithmic infrastructures and practices (e.g. Amoore, 2018; Iliadis and Russo, 2016; Kitchin and Lauriault, 2014) into conversation with development studies literature. Overall, we situate this special issue within the growing body of multidisciplinary scholarship that operates across these disciplinary thresholds (see Amrute et al., 2022; Singh and Jackson, 2021; Taylor et al., 2020) with the ambition to facilitate a substantial cross-interdisciplinary conversation on datafied development.

We identify datafied development as a paradigm in which development is increasingly imagined, governed, and evaluated through large-scale data infrastructures, algorithmic systems, and platform-mediated forms of knowledge production reconfiguring power, sovereignty, labor, and accountability. Datafied development marks a historical shift in how development itself is rendered knowable and actionable. While earlier development regimes were already deeply invested in metrics, indicators, audits, and statistical abstraction, datafied development intensifies and reconfigures these older logics of measurement, intervention, and control.

Within this broader landscape, extractivism provides a particularly useful lens for understanding these trends within datafied development. Extractivism refers to a “transnational system of labor exploitation and resource appropriation […] that enables the accumulation of commodities across vast networks of social actors, institutions, and territories” (Posada, 2026). Extraction as a concept has a long history in explaining patterns of exploitation and accumulation of value from the periphery to the core through different industries (Posada, 2026). Ye et al. (2020) note that extractivism is built on multiscalar dimensions from global value chains to local power dynamics and operates both across human and nonhuman entities (e.g. water and land) (also see Ricaurte, 2023a). In the context of datafication today, value extraction is continuous, automated, and infrastructural rather than episodic. Extractivism functions as a system of exploitation ranging from surveillance to formalized labor and land exploitation (Ricaurte, 2023b). Questions around data production and distribution are central to development concerns but also increasingly tied to the issue of value extraction from the Majority World. Hence, datafication is a process of data generation, processing, and exploitation for large scale value extraction and capture (Couldry and Mejias, 2019).

There are further issues around power imbalances when it comes to asymmetrical data flows vis-a-vis platforms (i.e. big tech firms, see Srnicek, 2017), leading some scholars to argue for the emergence of “data colonialism” (Couldry and Mejias, 2019). A parallel school of thought has also consolidated around the decolonial turn in both development studies and critical data studies (Singh, 2023) and a revitalization of dependency theories to explain the persistent inequalities through the lenses of colonialism and coloniality. The central argument within this body of work has been that forms of development (including industrialization, poverty reduction, employment generation, etc.) can take place in dependent capitalist relationships, albeit shaped by class compromise between local bourgeoisies, the state, and multinational capital. Thus, dependency and development are not necessarily mutually exclusive, but shaped by the multiscalar power arrangements which could lead to uneven development (Marini, 2022).

One of the underlying arguments we are making here is that these frameworks allow analyses of current datafied development in terms of subordination, imperialist dynamics, and global value chains. One of the key exponents of Marxist dependency theory, Theotonio dos Santos (1970), placed particular emphasis on the role of technology in reproducing dependency and, by extension, in conditioning technological development in peripheral countries via patents, rents, and royalties both on natural resources and machine tools. From microsocial and environmental perspectives, extractive nature of datafication invites closer examination of both the land and resource demands of data-intensive technologies such as so-called AI and the ways these datafied processes affect communities. Therefore, understanding datafied development also means understanding what hinders or blocks that development.

Datafied development represents an updating of super exploitation and the international division of labor within AI and data value chains, mediated by mechanisms of rentierism (Sadowski, 2020) and a dependency that is primarily infrastructural in character (Rikap, 2026; Srnicek, 2026). Infrastructural capacity, encompassing data centers, cloud computing, submarine cables, and associated systems, constitutes the material foundation of this dependency, consolidated within the monopolistic positions of large technology corporations (van der Vlist et al., 2024). Even where states and communities pursue the development of their own technologies, they encounter, at some point along the value chain, a structural dependency on infrastructures that they do not own or control. The recent proliferation of “stack” discourse, with its ambition to achieve control over entire digital infrastructure chains, emphasizes the importance of the dependency framework, offering a productive lens for examining why such full-chain control remains structurally difficult, if not impossible, for any actor beyond those already dominant.

Understanding datafied development then requires both the recognition of dependency and, dialectically, an attention to the possibilities of addressing digital sovereignty. Digital sovereignty is a plural and plastic concept (Couture and Toupin, 2019), whose meanings have been even co-opted by large technology companies as sovereignty-as-a-service (Grohmann and Costa Barbosa, 2026). In its more conventional sense, however, sovereignty refers to the control of technologies and infrastructures, with the state occupying a central role. Reclaiming digital sovereignty, according to a collective of authors led by Rikap et al. (2024) means, in synthesis, (1) digital infrastructures governed by nonprofit democratic international consortia; (2) universal digital commons, such as search engines, foundation AI models, cloud services, governed by public institutions with state and civil society representation; (3) it is grounded in an ecological internationalism, stating that the real rules of the digital economy are the leading corporations. For this reason, analytical frameworks such as dependency, sovereignty, and autonomy, understood dialectically, are productive resources for datafied development.

Examining datafied development therefore requires, on the one hand, attending to the possibilities, constraints, and meanings of development at both global and local scales. On the other hand, the importance of historicizing datafied development remains crucial. This means analyzing what is historically continuous and persistent in the current datafied conjuncture, and what constitutes new elements and layers. What is distinctive in the current conjuncture, we believe, is the growing centrality of digital data infrastructures, platforms, cloud services, algorithmic systems, and AI models in shaping how development problems are defined, what types of interventions are designed and how, and how outcomes are assessed. If earlier development regimes relied on censuses, household surveys, and institutional statistics, contemporary datafied development is increasingly organized through real-time data capture, predictive analytics, interoperable databases, entrepreneurial epistemologies, platform-mediated coordination, and privatized technical infrastructures. This marks a shift from development as a project primarily administered through states and multilateral institutions toward one increasingly mediated by technology firms, consultancies, philanthropic actors, and outsourced data infrastructures.

Rather than purely theoretical this paradigm became visible to us as researchers active in the field in the mundane pragmatics of expertise and procurement. One of us was for instance approached by a World Bank consultant seeking a report on “data responsibility” in East Africa; it soon became clear that the brief centered less on situated forms of social policy than on data warehouses, predictive governance, and partnerships with platform firms, including the use of proxy indicators such as commercial price fluctuations or pothole data as signals of conflict. Another of us come to think of the tech entrepreneur Leila Jannah's slogan “Give Work,” displayed in Sama's offices: a promise of opportunity that simultaneously obscures the labor processes, interfaces, infrastructures, and value chains through which AI work is organized.

Datafied development therefore signals both continuity and rupture. It continues long-standing developmental aspirations to know, manage, and improve populations through technical means. At the same time, it introduces new modalities of governance: from retrospective measurement to prediction and preemption; from public statistics to proprietary data assets; from territorial state infrastructures to transnational cloud and platform systems; and from development expertise grounded in field-based knowledge to knowledge practices increasingly shaped by datasets, proxies, dashboards, and algorithmic outputs.

Contributions

To study these transformations, we propose four interrelated domains to analyze datafied development: epistemic, infrastructural, economic, and geopolitical domains.

First, we argue for sustained attention to the epistemic domain, that is, how development knowledge is produced, validated, and authorized. What counts as evidence in datafied development? Which forms of social life become legible through data-intensive systems, and which remain obscured or distorted? Whose categories, classifications, and models shape developmental truth claims? This domain foregrounds questions of epistemic injustice, abstraction, proxy reasoning, and the displacement of situated forms of knowledge by computational forms of legibility. Several contributions in this SI examine the epistemic politics of datafied development, focusing on how development problems are rendered knowable through data-intensive systems. A commentary by Toupin and Siad (2025) on Artificial Intelligence for Development (AI4D), for instance, offers a mapping of the plurality of competing frameworks through which AI is mobilized in development discourse, ranging from techno-liberal developmentalism to decolonial critiques. Their analysis thus shows that rather than a singular project, AI4D unfolds as a contested field structured by divergent political and epistemological assumptions. Crucially, their critical review of literature on AI4D and related expressions shows that while the notion of AI4D applies broadly to the Majority World, the term is often used in reference to AI development on the African continent. Similarly, Ravn's (2025) article on synthetic data explores how emerging data practices reconfigure long-standing concerns in data justice. Through a case study of global synthetic datasets on human trafficking, Ravn's article thus demonstrates how claims to privacy and innovation may obscure persistent inequalities in visibility, representation, and political economy, raising concerns about what the author terms “synthetic-washing” and questions about what synthetic data justice might consist of.

Second, we draw attention to the infrastructural domain, which addresses the material and organizational architectures through which datafied development operates. Who owns and governs the infrastructures of data storage, computation, connectivity, and interoperability? Where are these infrastructures located, and under whose jurisdiction do they fall? Here, the concern is not only with visible tools and applications, but with the deeper stack of cloud services, data centers, platforms, software standards, biometric systems, and vendor dependencies that make datafied development possible. A set of articles examines how datafication reshapes institutional practices, labor, and governance in situated contexts, particularly within humanitarian and public-sector settings. Clausen et al. (2025) for instance analyze the use of big data in humanitarian localization, showing three ways in which the mobilization of big data to convey “local” perceptions of needs, what they call “datafied localization,” can perpetuate epistemic injustices and paternalistic logics in humanitarian practice. Iazzolino and Dhungana's (2025) contribution focuses on data curation practices in digital humanitarianism, demonstrating how datasets are produced through fragmented value chains, asymmetries in expertise, and contested collaborations, ultimately reshaping how crises are known and acted upon. Arriagada and Cotoras (2025) examine the implementation of predictive systems in Chile's Public Defender's Office, showing how ethical principles such as fairness and accountability are negotiated in practice rather than simply applied, revealing the limits of abstract AI ethics frameworks. And finally, Cornet et al. (2026) turn to AI data labor in the Majority World, highlighting how workers navigate precarious conditions while actively constructing strategies, identities, and aspirations within platform-mediated labor markets. Their analysis underscores how developmental narratives of opportunity often obscure the unrecognized skills and forms of labor that sustain AI systems.

Third, there is an increased need to examine the political-economic domain, that is, how value is extracted, distributed, and accumulated in and through datafied development. This includes the labor of data production, annotation, verification, moderation, and maintenance; the role of platforms and intermediaries; and the ways development initiatives may become sites for new forms of extraction, rentiership, and market-making. This perspective draws attention to how developmental promises can mask labor exploitation, enclosure of data resources, and the consolidation of corporate power. Fourth, we note the increasing significance of the geopolitical domain, which includes the reworking of dependency, sovereignty, and power across scales. Datafied development unfolds within an uneven international order in which access to compute, cloud infrastructures, foundational models, proprietary datasets, and platform ecosystems is highly concentrated. Dependency thus no longer operates only through trade, finance, or industrial production, but increasingly also through infrastructural and computational asymmetries of control and resilience. At the same time, projects of digital sovereignty, autonomy, and collective infrastructural control increasingly emerge as attempts to contest these dependencies.

The special issue foregrounds the political economy and geopolitical dimensions of datafied development, particularly in relation to extraction, infrastructure, and global power asymmetries. Safir and Sharma's (2025) commentary on AI for climate action reveals how the technology firms from the Minority World dominate infrastructural and discursive spaces, advancing techno-solutionist narratives while marginalizing actors from the Majority World. Hung's (2024) contribution furthermore conceptualizes AI as a planetary assemblage of coloniality, mapping a vertically integrated system spanning global corporate power, national policy regimes, and localized labor practices. Together, these contributions highlight how datafied development is embedded in uneven “techno-geoeconomic structures” that reproduce dependency and stratification across scales driving a tiered global data economy and new formations of AI as “planetary assemblages of coloniality.”

Taken together, we argue that these four domains provide a framework for understanding datafied development not as a singular field or sector, but as a wider transformation in the logics, institutions, and infrastructures through which development is imagined and enacted. Moreover, we propose that the stakes of this shift extend well beyond questions of efficiency, innovation, or technical inclusion. Rather, what is being transformed, we argue, is the very architecture of developmental governance: datafied development thus reorients development from diagnostic and redistributive logics toward predictive and preemptive forms of intervention; from public accountability toward opaque partnerships with vendors, platforms, and philanthropic actors; and from nationally anchored institutional capacities toward externally governed digital infrastructures. As a result, questions of justice in development can no longer be approached only through distributional frameworks, but must also be (re)understood as struggles over knowledge production, infrastructural control, labor conditions, and technological sovereignty.

This special issue contributes to an emerging cross-disciplinary conversation on the relationship between development and data-intensive digital infrastructures by bringing development studies, ICT4D, critical data studies, platform studies, political economy of tech and STS into closer dialogue. While these fields have each engaged important aspects of contemporary technological transformation, they have often done so in relative isolation: development studies has long examined the politics of intervention, inequality, and dependency; ICT4D has analyzed the promises and limits of digital inclusion and technological diffusion; and critical data studies has explored the politics of datafication, platformization, data justice, and algorithmic power. This special issue argues that contemporary transformations in development demand that these conversations be brought together more systematically.

In doing so, the special issue makes three contributions. First, it proposes datafied development as a conceptual frame for analyzing how development is increasingly reorganized through data infrastructures, platforms, and AI systems. Second, it offers a multiscalar and multisited account of these transformations, tracing them across humanitarian governance, state institutions, labor platforms, synthetic datasets, and geopolitical infrastructures. Third, it foregrounds the contested politics of this paradigm, showing how datafication does not simply modernize development but fundamentally reworks long-standing inequalities around knowledge, labor, territory, and sovereignty. Across the contributions, development emerges as a contested field of intervention in which digital technologies are mobilized to promise inclusion, efficiency, and optimization while often reproducing asymmetries of power. The special issue therefore invites readers to rethink development through the lens of datafication, and to rethink datafication through the historical and political questions that development studies have long posed.