Earlier this spring my LinkedIn feed shared a New York Times op-ed that I read with real concern. As each paragraph rolled into another, sharing Silicon Valley’s indoor thoughts on AI’s poise to create a permanent underclass, I stopped at one reader’s reflection in the comments. This person shared that they recently met someone who works at a big AI firm in the Valley, and asked if ethical people who care about the greater good still work there.

“Yes,” the person replied, “but none in positions of power.”

I started my own AI-adjacent career path nearly one decade ago, a lifetime in today’s terms. This was the dawn of our modern ethical AI movement, a team of computer scientists who rang the alarm that this technology could reinforce bias and must be managed like any other influential asset. (Financial markets, insurance, medicine, etc.) In my role as a then-analyst covering tech trends, joining this movement felt impactful, like that elusive greater good would prevail due to common knowledge about AI’s risks. In today’s world, where I work on AI initiatives that serve citizens in my state, I worry about my optimism.

If this sounds bleak, it’s bolstered by the fact that I read this op-ed the week after returning from my fellowship residency at the inaugural Digital Statecraft Academy. Hosted at The University of Cambridge, the Academy unites fellows working on AI governance in the global public sector. 16 fellows from 12 countries left our jobs, families, and to-do lists for one week to gather in person and start answering one of our time’s core questions: How can we solve the public sector’s skills gap in this moment when work is increasingly outsourced to machines?

We learned more that week than I can do justice to here: The fellows in my cohort have done a strong job summarizing key themes, from imagining the state’s role as an AI actor to governance as infrastructure. My main takeaway mirrors one from The New York Times underclass op-ed, which highlights that our current AI movement is creating rare class solidarity where the worries of Ivy League graduates mirror those of factory workers in the U.S. heartland: In today’s AI age, we have more in common than that which divides us.

During our residency, Cambridge hosted public sector leaders from Pakistan, Papua New Guinea, Germany, the Netherlands, Saudi Arabia, and more. Several of us were born in one country while while working in another. Throughout our week together, while we sat in lectures, discussed our fellowship ideas, and watched demos of new AI techniques, you might think we all approached this work from conflicting viewpoints. Instead, I saw the same solidarity described in The New York Times op-ed, alongside shared sense of urgency to leverage AI’s benefits for public good while minimizing its harms.

This fellowship came at an adept time for public servants working in AI. The harms that AI’s misuse can bring are firmly here and visible. In the United States, where I live and work on AI programs for the State of Maryland, resistance to data centers and companies trying to replace people with machines is loud and filled with righteous anger. I still believe in AI’s transformative promise when used as one tool in a bigger kit to solve the right problems and processes. I also wonder, in low moments, if it was naive to think the ethical AI movement would prevail. That view of global power – specifically, who has the most power to control AI – assumes a level of checks and balances that I don’t see today.

This is one of several subjects we discussed during the DSA Cambridge residency: AI sovereignty not as a concept, but as a practical question of ownership and governance. Today, the top five AI companies by revenue are American. Given their well-documented missteps stewarding user data, it’s understandable that other countries (especially in Europe) try to govern AI’s use in alignment with their values, which more squarely center on consumer protection and user privacy. In tandem, international attempts at regulation fall flat when those bodies’ leaders try to regulate technology that was not built in their own countries, and which they often do not understand well enough to govern.

This is a broad simplification; 26 countries comprise the European Union, and saying their citizens all share the same priorities is no less naive than claiming that U.S. citizens in all 50 states want the same policy outcomes. Still, the fact remains: We are at a point in AI development where the most powerful players operate under a policy patchwork coming from U.S. states and global governing bodies, while leveraging lobbyists to stall U.S. federal laws.

Hence why I can’t stop thinking about that response to The New York Times op ed. In AI today, “positions of power” does not simply reference those in leadership roles at U.S. hyperscalers: It dually implies that in the wider AI landscape, those people are the only ones in positions of power. To date, those in government and the civil service have largely lacked the grit to govern well on a topic that is changing the world’s global fabric.

On a virtual call post-residency, one of our fellowship’s guest speakers – a chief scientific advisor for their country’s government – said that when it comes to capability, we must ask if we have the grit to govern well. Pressed on what that meant, they said it involves public leaders – from civil servants to policymakers – who pledge to hold complexity and uncertainty without paralysis.

This article’s draft has been an open tab on my laptop for four weeks; I’ve had so much to say and none of the right words. AI has always been a complex topic; DSA’s residency occurred during a confluence of consequences in this space. With so many swirling thoughts, it has been hard to put them on paper. And yet, that is the problem this residency and fellowship were born to solve: Give public leaders the tools, framework, and community to hold AI’s complexity and uncertainty without paralysis. None of us have all the answers, and there is some comfort in the knowledge that no one does. When new threats loom above society, that society tends to unite.

This is a benefit of the fact that humans are more reactive than proactive. When we stare down the barrel, we reach our hands out to find others who can climb down with us, explore the unknown, and go back up with some solutions to do what we can, with what we have, where are. The Digital Statecraft Academy exists to give public leaders the grit to govern well, to put public good ahead of cash flow. In this moment when AI’s puppet strings are pulled by few hands, the broader public is largely united in a moment when we are divided on much else. It’s the job of public leaders to respond on their behalf. DSA is built to help us reclaim our power.

Read more about the mission of The Digital Statecraft Academy, subscribe to the mailing list for future updates and consider expressing an interest in taking part in future editions.

About the author

Lauren Maffeo is an award-winning author, analyst, and digital systems designer for the U.S. government. In her work, she helps business leaders, educators, and entrepreneurs create structured, actionable data governance plans — because AI is only as good as the data behind it. Lauren is also one of the inaugural Fellows of The Digital Statecraft Academy.

Reflections on The Digital Statecraft Academy was originally published in Data & Policy Blog on Medium, where people are continuing the conversation by highlighting and responding to this story.

Every data transformation in governance privileges what can be counted while quietly sidelining what cannot. That is precisely why family formation is such a useful stress test for data-driven public policy.

Falling birth rates are usually discussed as a demographic question: pension systems, labour markets, intergenerational support and the future demand for public services. Birth statistics record births that happen. They do not routinely record the missing births: the person who wanted a child but did not have one; the couple who delayed because housing never felt secure; the single person or LGBTQ+ couple priced out of assisted reproduction; the patient who could not find — or could not trust — information they needed; the family who could not square care, work, schools and distance from kin (Huddleston et al. 2023, Fledderjohann & Barnes, 2018). These are not marginal details. They are often the operational realities through which reproductive agency is enabled or constrained.

This is not an invitation to fertility specialists alone. It is an invitation and provocation to people working on data-driven governance, public services, urban systems, digital health, public statistics, data protection, citizen participation, information and knowledge infrastructures and research infrastructure to consider family formation as a revealing policy domain.

In the language of Data & Policy Area 1 (Geist et al., 2025) , the core questions for this call are familiar: How does data become evidence? How do public institutions interact with citizens? How are services designed, governed and evaluated? What happens when public, private and voluntary sector data systems do not connect? How do open data, institutions and citizens relate as those systems are transformed?

From fertility decline to reproductive agency

The UNFPA’s State of World Population 2025 report captures the issue, not simply as some countries have too few births, or that some people should be encouraged to have more children (United Nations Population Fund [UNFPA], 2025). The more defensible policy concern is reproductive agency: whether people can have the children they want, when they want them, safely, affordably and with information they can navigate and trust. That includes people who want more children, fewer children, no children, or children under different conditions.

This matters because the data architecture of fertility policy is asymmetrical. We have comparatively strong data on births, maternities, in vitro fertilisation (IVF) cycles and treatment outcomes.

We have much weaker data on the reasons intentions do not become births (UNFPA, 2025). The “non-birth” is difficult to count: it may be caused by housing insecurity, relationship timing, caring responsibilities, local service access, clinic costs, discrimination, the absence of reliable advice at the right time, or simply uncertainty about the future (Vignoli et al., 2020)

Seen this way, family formation is not only a private medical or demographic matter. It is also a question of public legibility. Which experiences become visible enough to be governed for? Which remain known only to citizens, clinicians, charities, local authorities, housing officers, data stewards or service designers?

What governance can and can’t see

First, governance can see medical infertility much more clearly than social or structural infertility (Yopo Díaz & Watkins, 2025). For example, in the UK, the Human Fertilisation & Embryology Authority (HFEA) data can tell us a great deal about fertility treatment, including treatment volumes, age patterns and success rates. That is essential. But it cannot, on its own, tell us who never reached the clinic, who could not afford treatment, who lacked local access, who was deterred by eligibility rules, or who delayed family formation until biological and social time no longer aligned. The governance question is: what would have to change for those absences to become visible?

Second, governance can see place, but not always the lived geography of care, especially at neighbourhood and local government scale. Family formation is shaped by factors ranging from housing markets, transport, access to kin, childcare, schools, neighbourhood safety, green and blue space, work patterns and local healthcare pathways (e.g, Bergsvik et al, 2023, Yopo Díaz & Watkins, 2025). Yet these data usually sit in different institutions, at different scales, under different legal and operational constraints. The hard question for practitioners is not “is place important?” but “which data would need to be linked, who would need to trust, enable and maintain that linkage, and what currently prevents it?”

Third, governance can see some digital traces, but not necessarily the information environment citizens inhabit. Many people learn about fertility, cycles, contraception, miscarriage, egg freezing, IVF success rates, donor conception and pregnancy through search engines, apps, social media, commercial clinics and peer networks, which explains why knowledge is patchy and people feel uninformed (Clements et al, 2025). That creates a problem of reproductive — or parenthood — literacy and a governance problem at the same time: citizens are expected to make consequential decisions from fragmented information, while sensitive reproductive data may flow through services that sit outside traditional public-health infrastructures (French et al, 2022). The pressing information literacy question is then: who is accountable for ensuring that timely, trustworthy and usable family-formation information is available in the places people actually seek it? And who is accountable for the destiny of sensitive data generated in those searches to be governed in the public interest?

Why practitioner knowledge matters

Practitioners often know where the data fails long before the failure appears in an academic article. A local authority analyst may know that housing data cannot be linked to family service use. A midwife may know which groups arrive with poor information. A fertility clinician may know which patients are missing from official pathways. A charity may know which eligibility rules exclude LGBTQ+ families or single parents. A data protection officer may know which reuse would be valuable but currently lacks a trusted governance route.

This kind of operational knowledge is too often treated as anecdotal. It should instead be understood as evidence about the functioning of public systems. The aim is not to blame public servants for imperfect data, but rather learn from the people who work closest to the forms, registers, dashboards, procurement rules, consent processes, referral pathways and service gaps through which policy is actually made.

About the Author

Dr Bonnie Boyana Buyuklieva FHEA, FRGS is a Lecturer (research) in Data Science for Society at University College of London’s Dept of Information Studies. Her background is in architecture and computation (Foster and Partners; Bauhaus University Weimar), and she holds a PhD from the Bartlett Centre for Advanced Spatial Analysis (CASA, UCL).

Reading Family Formation: What Data-Driven Governance Can and Cannot See was originally published in Data & Policy Blog on Medium, where people are continuing the conversation by highlighting and responding to this story.

When financial institutions in emerging markets pilot blockchain, the story rarely follows the script written in Frankfurt or Singapore. A bank in Lagos adopts a cross-border payments rail not because it offers a marginal efficiency gain over correspondent banking, but because the correspondent banking option has effectively withdrawn. A cooperative in São Paulo experiments with tokenised receivables not because regulation is settled, but because the alternative — waiting for it to settle — costs working capital it does not have. Our recent systematic review in Data & Policy, “Blockchain Adoption in Emerging Markets: A Systematic Review of Factors and Perceived Impacts,” synthesises 44 peer-reviewed studies to map why these adoption pathways diverge so consistently from those in developed economies — and what that divergence means for policymakers.

The dominant lens for studying technology adoption in firms is the Technology–Organization–Environment (TOE) framework, which sorts adoption drivers into three buckets: the technology itself, the organisation considering it, and the wider environment of regulation, market structure, and institutional trust. TOE has held up well across decades of research on enterprise technology, but it was developed with relatively stable institutional environments in mind. Our review asked a sharper question: when the institutional environment is itself unstable, fragmented, or under-developed, do the same dimensions still carry the same weight?

The short answer is no — and the difference matters for policy.

Adoption antecedents (technological, organisational, institutional) shape the adoption decision; the implementation experience that follows produces a perceived impact that feeds back into either deeper institutionalisation or abandonment. Practical implications differ systematically across emerging-market contexts. Figure adapted from Staley & Amankwa (2026), Data & Policy. CC-BY 4.0.

Adoption is a two-stage pathway, not a binary

Most adoption studies treat the question as a switch: did the firm adopt or didn’t it? Our synthesis pushes back on that framing. Drawing on classic work by Davis (1989) on perceived usefulness and Rogers (2003) on diffusion, we treat adoption as a two-stage pathway. Stage one — the antecedents stage — is where TOE factors (technology, organisation, institutions) shape the decision to adopt. Stage two — the outcomes stage — is where the implementation experience generates a perceived impact: the stakeholder’s subjective evaluation of whether adoption is delivering. That perceived impact then feeds back into the system, supporting either deeper institutionalisation or abandonment.

In emerging markets, the same conditions that shape antecedents — patchy infrastructure, ambiguous regulation, weak institutional trust — also shape the implementation experience. A regulatory void can act as an enabler at stage one (low compliance burden, room to experiment) and an inhibitor at stage two (no enforceable rights when something goes wrong). This dual role is rare in developed-economy adoption studies and helps explain why the same firm can rationally adopt and then quietly abandon within a few years.

Institutional factors do the heavy lifting

The TOE framework treats its three dimensions as roughly co-equal. Across the 44 studies we reviewed, that’s not what we found in emerging markets. Institutional factors — regulatory clarity, government measures, institutional trust, competitive pressure, market demand — appeared as a substantively addressed driver in 86% of the studies. Technological factors appeared in 100%, organisational factors in 75%. But when researchers traced which factors actually moved adoption decisions and post-adoption outcomes, institutional variables consistently carried disproportionate weight.

This is intuitive once you say it out loud. In a stable regulatory environment, the institutional layer is mostly invisible — it’s the air the firm breathes. In an unstable one, every adoption decision is also a bet on what the rules will look like next year. We document cases where regulatory ambiguity functioned as a green light (Brazil’s central bank pilots, India’s regulatory sandboxes) and cases where it functioned as a red light (Argentina’s fragmented frameworks, Pakistan’s enforcement gaps). The variable is not “regulation” but the direction and credibility of regulatory signalling.

Leapfrogging is real but uneven

A recurring theme in the optimistic literature on emerging-market technology is leapfrogging: the idea that countries with weaker legacy infrastructure can jump straight to a more advanced configuration, bypassing the intermediate stages developed economies passed through. Mobile money in Kenya is the canonical example.

The blockchain evidence supports a more nuanced version of this story. Leapfrogging happens, but it is conditional. Countries with reasonable connectivity, an active FinTech ecosystem, and at least some regulatory engagement (Malaysia, the UAE, Brazil) can use blockchain to skip past creaky correspondent-banking or paper-based settlement infrastructure. Countries where the underlying connectivity, electricity, or data centres are themselves the bottleneck (parts of Sub-Saharan Africa, rural South Asia) cannot leapfrog past infrastructure they don’t yet have. Treating “emerging markets” as a single category obscures this distinction. The review pushed us toward a more stratified view: comparable institutional conditions produce comparable adoption pathways, and aggregating across markets blurs the signal.

Policy implications differ by region

The most actionable finding for policymakers is that the highest-leverage interventions vary systematically by regional context.

In Sub-Saharan Africa, where remittance corridors are expensive and financial inclusion gaps are wide, cross-border payments are the natural first application — but only when paired with investment in the digital connectivity layer that blockchain rails depend on. In South and Southeast Asia, where manufacturing and logistics ecosystems are mature, supply-chain finance and provenance tracking are the higher-yield use cases. In Latin America, the bottleneck more often lies in identity and compliance: blockchain-based digital identity and KYC/AML solutions address recurrent regulatory friction.

These are not interchangeable. A policy framework calibrated to Latin American compliance challenges does not transfer cleanly to Sub-Saharan African connectivity gaps. The review’s strongest recommendation against one-size-fits-all prescriptions follows directly from this: regional structural conditions should shape the application portfolio, not the other way round.

What this means for the next round of research

Two methodological gaps stood out. First, the literature is overwhelmingly cross-sectional. We can describe adoption decisions and early implementation experiences, but few studies follow the same firms long enough to observe whether adoption sticks, what changes the perceived impact over time, or whether early enthusiasm gives way to abandonment. Longitudinal work is overdue. Second, “emerging markets” is treated as a homogeneous bucket far too often. Stratified analyses that link individual findings to specific institutional, infrastructural, and developmental classifications would let researchers say more precise things about transferability.

For policymakers and financial institutions designing blockchain interventions in emerging markets, the practical takeaway is straightforward: the institutional and infrastructural baseline isn’t context; it’s the variable that most determines whether adoption produces the impact stakeholders expect. Calibrate accordingly.

The full peer-reviewed article, “Blockchain Adoption in Emerging Markets: A Systematic Review of Factors and Perceived Impacts,” is open access in Data & Policy (https://doi.org/10.1017/dap.2026.10071).

About the Authors

Ian Staley is an independent researcher in fintech and tokenisation infrastructure and a doctoral candidate in IT (Blockchain and Digital Assets) at IICUT. Eric Amankwa is in the Department of Computer Science at Ghana Communication Technology University.

Why Blockchain Adoption Looks Different in Emerging Markets was originally published in Data & Policy Blog on Medium, where people are continuing the conversation by highlighting and responding to this story.

Artificial Intelligence (AI) is often presented as the defining technology of our era — transformative, inevitable, and capable of reshaping economies, governments, and societies. Yet, when AI meets organizational reality, the story becomes far more complicated.

Across discussions on digital government, intelligent systems, data governance, and organizational transformation, one recurring insight increasingly holds: AI rarely fails because models are not advanced enough. More often, it fails when institutions are not ready for the responsibilities intelligence creates.

This observation is becoming visible across sectors. Governments, companies, and public institutions are rapidly experimenting with AI-enabled systems, yet many initiatives struggle to move beyond pilots or produce sustainable value. The challenge is rarely purely technical. Instead, the failures frequently emerge from fragmented governance, unclear ownership, weak data foundations, limited organizational capacity, or poorly designed human–AI relationships (Campion et al., 2022; Nikiforova et al., 2026; Sun and Medaglia, 2019).

Responsible AI, therefore, cannot be reduced to model performance or “ethics-by-design” checklists alone. This reflects broader calls to contextualize digital artifacts within the organizational, institutional, and infrastructural environments in which they operate (Currie, 2009; Benbya et al., 2020).

Responsibility should therefore be understood not merely as a property of AI systems themselves, but also as a property of the manner in which they are adopted and governed (Nikiforova et al., 2026). The critical question is not only whether AI systems are responsible, but whether public administrations and organizations are positioned to adopt AI responsibly — in ways that uphold legality, accountability, fairness, sustainability, and inclusion (Nikiforova et al., 2026; Sun and Medaglia, 2019). Responsible AI adoption thus becomes a question of governance capability, institutional readiness, and socio-technical design.

AI Does Not Begin with Models — It Begins with Data

Public debates around AI often focus on algorithms, automation, or generative systems. But AI systems do not emerge in isolation. They depend on infrastructures, data ecosystems, organizational arrangements, legal frameworks, and human practices that shape whether these systems become useful, trustworthy, or harmful.

This is particularly visible in public and organizational data ecosystems — complex socio-technical networks in which actors interact to create, consume, publish, share, archive, govern, repurpose and reuse data. Such ecosystems are not merely technical infrastructures. They consist of interdependent relationships between institutions, technologies, governance structures, legal frameworks, and users, continuously evolving through interaction and co-dependence (Dawes et al., 2016; Nikiforova et al., 2025).

Discussions of AI governance though still too often overlook the data dimension itself (Aaltonen et al., 2023). Yet persistent weaknesses across the data lifecycle continue to undermine AI adoption in practice. Responsible AI adoption in the public sector further demonstrates that many adoption barriers are fundamentally data-centric and organizational rather than purely technical (Nikiforova et al., 2026):

• fragmented or low-quality data,
• interoperability gaps,
• inconsistent metadata and documentation,
• unclear ownership and stewardship,
• weak accountability structures,
• organizational silos,
• and insufficient governance coordination.

As Wang and Strong (1996) and Strong et al. (1997) argued long before the current AI wave, data quality is contextual and deeply tied to organizational use. AI does not eliminate these problems. In many cases, it amplifies them.

This becomes especially critical because AI systems increasingly depend on interconnected data infrastructures operating across organizational and institutional boundaries. Data curation and infrastructure work are themselves anticipatory and deeply embedded in broader socio-technical arrangements (Parmiggiani et al., 2024). AI adoption therefore inherits not only datasets, but also the governance weaknesses, assumptions, and institutional dependencies embedded within them. This challenge is increasingly visible in practice as public institutions attempt to operationalize data-driven policy under fragmented governance conditions (Buyuklieva et al., 2026).

Responsible AI Requires Lifecycle Thinking

One of the central limitations of contemporary AI governance is fragmentation. Governance is frequently approached as a compliance exercise focused narrowly on AI systems themselves, while the broader lifecycle dependencies enabling those systems remain underexamined.

Recent research increasingly suggests the need for lifecycle-oriented and ecosystem-based approaches spanning both data and AI governance.

This includes:

• governance across the full data lifecycle (creation, collection, storage, sharing, use, reuse, archiving, deletion),
• governance across the AI lifecycle (planning, development, deployment, operation, retraining, retirement),
• recognition of different organizational and data roles,
• and understanding AI systems as embedded within broader socio-technical and infrastructural ecosystems.

Importantly, governance challenges differ across lifecycle phases and organizational roles. Research on data governance has long demonstrated that “one size does not fit all” (Weber et al., 2009). Governance requirements vary depending on organizational context, responsibilities, and forms of authority. This becomes even more complex with AI.

Moreover, AI systems increasingly perform different organizational roles. For example, recent work on AI-assisted data quality management (Roht et al., 2026) identified several distinct AI roles emerging in practice:

• the translator (turning human intent into technical logic),
• the explainer (summarizing and interpreting patterns),
• the resolver (suggesting or implementing fixes),
• and the integrator (connecting systems and datasets).

Each role creates different forms of authority, dependency, and risk. An AI system acting as an explainer requires different oversight mechanisms than one autonomously integrating records across systems.

As AI becomes embedded into organizational ecosystems, it reshapes how data is produced, interpreted, governed, and acted upon, while simultaneously transforming relationships between humans, institutions, and technological systems. This reflects broader shifts toward human–AI hybrid arrangements and delegation relationships (Giest and Klievink, 2024; Rai et al., 2019; Baird & Maruping, 2021; Roht et al., 2026).

This insight is critical: responsible AI is not a single governance problem. Different AI roles generate different governance needs, accountability structures, and oversight requirements. An AI system acting as an explainer requires different safeguards than one autonomously integrating records across systems.

Governance therefore cannot remain static, model-centric, or one-size-fits-all.

Question 0: Do We Need AI Here at All?

Organizations often begin with the question:

“Where can we use AI?”

But perhaps the more important question comes earlier — what Virginia Dignum (2019) describes as “Question 0”, namely

“Do we need AI here at all?”

Not because AI can be deployed, but because it creates meaningful public, organizational, or societal value.

This perspective is especially important in public administration and policy contexts, where AI adoption increasingly intersects with democratic legitimacy, accountability, public trust, and institutional resilience.

Sometimes the most responsible strategy is not automation, but redesign.

Long before AI, business process reengineering already warned organizations:

“Do not automate, obliterate.”

Simply layering AI onto fragmented or dysfunctional processes risks accelerating inefficiency rather than solving it.

Responsible adoption therefore also requires asking:

• Is the task itself appropriate for AI?
• Is the process ready?
• Is the institution ready — technically, legally, organizationally, and culturally?
• Are governance structures capable of supporting long-term accountability and oversight?
• Is the data ecosystem mature enough to sustain responsible deployment?

Without these foundations, even technically sophisticated systems may fail in practice.

Beyond Systems: Governing Infrastructures and Dependencies

Another emerging challenge is that many AI governance frameworks remain heavily software-centric. Existing approaches often focus primarily on models, applications, and risk classifications, while underestimating the broader infrastructures, material and environmental dependencies that make AI systems possible.

Yet AI systems are deeply embedded within data infrastructures, cloud and compute dependencies, semiconductor supply chains, hardware production, energy systems, procurement decisions, and organizational capabilities.

Emerging research therefore increasingly calls for a shift from governing isolated AI systems toward governing AI ecosystems, infrastructures, and lifecycle dependencies. This broader perspective also brings sustainability into AI governance discussions. AI governance is not only about fairness, transparency, explainability, or accountability. It is also about:

• environmental sustainability,
• compute-intensive scaling,
• energy consumption,
• hardware dependencies,
• procurement practices,
• and long-term institutional resilience (Nikiforova and Cole, 2026).

Recent work on supply chain-aware AI governance (Nikiforova and Cole, 2026) argues that current governance frameworks often externalize material and environmental impacts while neglecting upstream and downstream dependencies across the AI lifecycle. Governance fragmentation, rebound effects, and infrastructure invisibility increasingly challenge the sustainability of large-scale AI adoption.

As one joke circulating in AI governance discussions summarized it “With great AI also come great electricity bills.” Humorous perhaps — but increasingly true.

From Ethics to Institutional Capability

Responsible AI is often framed primarily as an ethical issue. Ethics remains essential. But ethics alone is insufficient if institutions lack the capability to operationalize responsibility in practice.

The challenge is therefore not simply building “ethical AI,” but building institutions capable of governing AI responsibly over time.

This requires:

• governance capacity,
• interdisciplinary collaboration,
• organizational learning,
• AI and data literacy,
• adaptive regulatory approaches,
• lifecycle-oriented governance,
• stronger integration between data governance and AI governance,
• and ecosystem-level thinking.

Ultimately, responsible AI is not merely a property of systems. It is a property of institutions.

Looking Ahead: A Call for Contributions

As the Data for Policy 2026 Conference approaches, with its focus on “Governance of/with AI: Implications for Data, Infrastructure, and Tech Sovereignty,” Focus Area 1 — Digital & Data-Driven Transformations in Governance — provides an important space to advance these debates.

We welcome contributions exploring topics including:

• data governance as a prerequisite for AI governance,
• AI governance across data and AI lifecycles,
• human–AI delegation and accountability,
• institutional readiness and governance capability,
• AI sustainability and infrastructural dependencies,
• organizational and socio-technical perspectives on AI adoption,
• interoperability, coordination, and ecosystem governance,
• AI in policymaking and public administration,
• responsible AI adoption in public and private sectors,
• and broader questions surrounding trustworthy, sustainable, and democratic AI governance.

As AI continues to evolve, the key challenge may not be whether systems become more intelligent. The real question is whether our institutions become wise enough to govern them responsibly.

If you are working on one or more of these topics, we warmly encourage you to submit an abstract or full paper proposal by the deadline of June 1st 2026 and join us in advancing this conversation.

About the Author

Anastasija Nikiforova is an Associate Professor of Applied AI and Information Systems at the at University of Tartu, Estonia. She is one of the Area Editors of the Data & Policy journal (Cambridge University Press), covering Focus Area 1: Digital & Data-Driven Transformations in Governance. Other members of the Area Committee include:

• Mitzi Bolton (Monash University, Australia)
• Bonnie Buyuklieva (UCL, UK)
• Sarah Giest (Leiden University, Netherlands)
• Sharique Manazir (Kautilya School of Public Policy, GITAM University, India)
• Sujit Kumar Sikder (Leibniz Institute of Ecological Urban and Regional Development, Germany).
• Austin Zwick (Syracuse University, USA)

References:

Aaltonen, Aleksi, Cristina Alaimo, Elena Parmiggiani, Marta Stelmaszak, Sirkka L. Jarvenpaa, Jannis Kallinikos, and Eric Monteiro. “What is missing from research on data in information systems? Insights from the inaugural workshop on data research.” Communications of the Association for Information Systems 53, no. 1 (2023): 475–490.

Baird, Aaron, and Likoebe M. Maruping. “The next generation of research on IS use: A theoretical framework of delegation to and from agentic IS artifacts.” MIS quarterly 45, no. 1 (2021): 315–341.

Benbya, H., Nan, N., Tanriverdi, H., & Yoo, Y. (2020). Complexity and information systems research in the emerging digital world. MIS Quarterly, 44(1), 1–17.

Buyuklieva, B., Williams, J., Zileli, S., Wood, M., Gozzi, M., Hera, V.-N., McKeating, A., & Chaytor, S. (2026). Delivering Data-Led Local Policy: How to Build Data Capacity and Culture in Local Government. Data for Policy Community Report.

Campion, Averill, Mila Gasco-Hernandez, Slava Jankin Mikhaylov, and Marc Esteve. “Overcoming the challenges of collaboratively adopting artificial intelligence in the public sector.” Social Science Computer Review 40, no. 2 (2022): 462–477.

Dawes, Sharon S., Lyudmila Vidiasova, and Olga Parkhimovich. “Planning and designing open government data programs: An ecosystem approach.” Government information quarterly 33, no. 1 (2016): 15–27.

Dignum, Virginia. Responsible artificial intelligence: How to develop and use AI in a responsible way. Vol. 2156. Cham: Springer, 2019.

Giest, Sarah N., and Bram Klievink. “More than a digital system: how AI is changing the role of bureaucrats in different organizational contexts.” Public Management Review 26, no. 2 (2024): 379–398.

Nikiforova, A., Cole, N.L., “From Code to Cobalt: A Supply Chain–Aware Lifecycle for AI Governance and Sustainability” (2026) ECIS 2026. TREO

Nikiforova, Anastasija, Martin Lnenicka, Mariusz Luterek, Petar Milic, and Manuel Pedro Bolívar Rodríguez. “Theorizing the evolution of public data ecosystems: An empirically grounded multi-generational model and future research agenda.” Government Information Quarterly 42, no. 3 (2025): 102062.

Nikiforova, Anastasija, Martin Lnenicka, Ulf Melin, David Valle-Cruz, Asif Gill, Cesar Casiano Flores, Emyana Sirait, Mariusz Luterek, Richard Michael Dreyling, and Barbora Tesarova (2026) Responsible AI Adoption in the Public Sector: A Data-Centric Taxonomy of AI Adoption Challenges. In Proceedings of the 59th Hawaii International Conference on System Sciences

Parmiggiani, Elena, Nana Kwame Amagyei, and Steinar Kornelius Selebø Kollerud. “Data curation as anticipatory generification in data infrastructure.” European Journal of Information Systems 33, no. 5 (2024): 748–767.

Rai, Arun, Panos Constantinides, and Saonee Sarker. “Editor’s comments: Next-generation digital platforms: Toward human–AI hybrids.” MIS quarterly 43, no. 1 (2019): iii-ix.

Roht, Karen; Nikiforova, Anastasija; and Lefebvre, Hippolyte, “A Typology For Responsible Delegation Of Data Quality Management To Generative Ai” (2026). ECIS 2026 Proceedings. 16. https://aisel.aisnet.org/ecis2026/datasc_isresearch/datasc_isresearch/16

Strong, Diane M., Yang W. Lee, and Richard Y. Wang. “Data quality in context.” Communications of the ACM 40, no. 5 (1997): 103–110.

Sun, Tara Qian, and Rony Medaglia. “Mapping the challenges of Artificial Intelligence in the public sector: Evidence from public healthcare.” Government information quarterly 36, no. 2 (2019): 368–383.

Wang, Richard Y., and Diane M. Strong. “Beyond accuracy: What data quality means to data consumers.” Journal of management information systems 12, no. 4 (1996): 5–33.

Weber, Kristin, Boris Otto, and Hubert Österle. “One size does not fit all — -a contingency approach to data governance.” Journal of Data and Information Quality (JDIQ) 1, no. 1 (2009): 1–27.

When AI Meets Reality: Why Responsible AI Adoption Begins with Data and Governance was originally published in Data & Policy Blog on Medium, where people are continuing the conversation by highlighting and responding to this story.

Algorithmic governance is no longer a future prospect. It is already embedded in how governments deliver services, how decisions are made, and how power is exercised across societies. From automated welfare systems to predictive analytics in public policy, algorithms are increasingly shaping the infrastructures of governance. Yet, the success of these systems does not depend solely on technological sophistication. It depends on whether institutions and citizens are equipped to understand, question, and meaningfully engage with them. For the editorial committee of the Data for Policy Conference and Data & Policy journal focused on Algorithmic Governance (Area 5 of the six focus areas working across the two venues), this raises a fundamental question: how do we ensure that the expansion of algorithmic systems enhances, rather than undermines, good governance?

From Automation to Governance Infrastructures

Algorithmic governance is often framed in terms of automation. Governments adopt AI to improve efficiency, optimize service delivery, and process large volumes of data. These developments reflect what we might call governance with algorithms, where AI supports decision-making processes. But the reality is more complex. Increasingly, we are witnessing governance by algorithms, where systems directly influence or make decisions, and governance through algorithms, where digital infrastructures shape behaviour, access, and visibility. Karen Yeung’s work on algorithmic regulation is useful here, as it shows how algorithmic systems can structure conduct through automated, data-driven forms of control. This transformation is particularly visible in areas such as fraud detection, social protection, and urban management, where algorithmic systems structure how individuals interact with the state.

Research has shown that these systems are not neutral tools. They are embedded in institutional contexts, shaped by policy choices, and capable of reinforcing existing power dynamics. As a result, algorithmic governance must be understood not only as a technical development, but as a reconfiguration of governance itself.

Good Governance Through, With, By, and Of Algorithms

Ensuring “good governance” in this context requires a shift from principles to practice. Transparency, accountability, and fairness remain central, but they are not sufficient on their own. Hauer, Krafft & Zweig have highlighted that transparency mechanisms, for instance, often fail to produce meaningful accountability if stakeholders cannot interpret or act on the information provided. Similarly, algorithmic systems used in public administration can introduce new forms of rigidity and opacity if not designed with institutional realities in mind.

This suggests that good governance in the age of AI must address multiple layers:

• governance of algorithms
• governance with algorithms
• governance by algorithms
• governance through algorithms

Bringing these layers together is essential for building governance systems that are not only efficient, but also legitimate and accountable.

Algorithmic Agency and the Limits of Control

The rise of generative and agentic AI introduces new challenges. These systems are not simply executing predefined tasks; they can generate content, adapt to new contexts, and increasingly participate in decision-making processes. This raises critical questions about algorithmic agency. If AI systems can influence outcomes in complex and sometimes unpredictable ways, who is responsible? How do we ensure accountability when decision-making is distributed across human and machine actors? Recent contributions by Bell, Nov and Stoyanovich point to the need for more robust governance frameworks that go beyond risk classification and compliance, and instead address institutional design, oversight, and accountability in dynamic systems.

Human Agency in Algorithmic Governance

A central tension in algorithmic governance is the relationship between human and machine agency. While AI systems can enhance decision-making, they can also reshape how humans think, act, and participate in governance processes. Trustworthy governance is not only about regulating systems; it is about ensuring that individuals and institutions have the capacity to understand and engage with them. If decision-makers rely uncritically on algorithmic outputs, or if citizens lack the tools to question automated decisions, governance risks becoming less transparent and less democratic. In this sense, algorithmic governance is also about epistemic empowerment: preserving the ability of humans to interpret, challenge, and shape decisions.

Human–Machine Collaboration in Critical Decision-Making

Rather than framing the future as a choice between human and machine decision-making, the more productive approach is to focus on human–machine collaboration. Research across the Data & Policy community highlights that effective collaboration depends on:

• clearly defined roles and responsibilities
• meaningful human oversight
• institutional capacity to evaluate and contest algorithmic outputs

Accountability in Algorithmic Governance: From Principle to Practice

Accountability has become one of the central pillars of algorithmic governance, frequently invoked as a condition for legitimate, trustworthy, and lawful uses of artificial intelligence. However, despite its prominence across ethical frameworks, policy documents, and regulatory instruments, accountability remains conceptually unstable and operationally underdeveloped in practice. As recent research shows, accountability often functions as a “cluster concept,” co-occurring with transparency, explainability, human oversight, auditability, and fairness, rather than as a clearly delimited legal or institutional standard.

This conceptual ambiguity poses a significant challenge for Area 5 concerns, particularly in public-sector contexts where algorithmic systems increasingly mediate decisions related to welfare, policing, health, and education. Public authorities are not only expected to comply with legal requirements but to remain answerable for the use of AI systems within existing frameworks of administrative law, data protection, and fundamental rights. Yet, without a clear understanding of what accountability entails, translating this expectation into practice becomes difficult.

A useful way to address this challenge is to distinguish between three closely related but analytically distinct concepts: accountability, responsibility, and liability. From a governance perspective, accountability must therefore be understood as lifecycle-based and socio-technical.

Importantly, accountability in algorithmic governance is not only a legal or technical issue but also an institutional and epistemic challenge. As AI systems mediate knowledge production and decision-making, the ability of individuals and institutions to understand, question, and contest algorithmic outputs becomes central. Without sufficient interpretive capacity, transparency risks becoming purely procedural, and accountability risks devolving into formal compliance without substantive oversight.

Looking Ahead: A Call for Contributions

As the Data for Policy 2026 Conference approaches, with its focus on “Governance of/with AI: Implications for Data, Infrastructure, and Tech Sovereignty,” Focus Area 5 on Algorithmic Governance provides a critical space to advance these debates. We invite abstract and full paper contributions by the deadline of June 1, 2026 that explore:

• automation of government processes and services
• governance through, with, by, and of algorithms
• algorithmic agency and decision-making
• human agency and epistemic capacity
• human–machine collaboration
• socio-economic impacts of algorithmic behaviour
• Accountability, liability and responsibility in practice.

At a time when AI systems are rapidly reshaping governance, there is a pressing need for interdisciplinary research that bridges technical, legal, and social perspectives. Algorithmic governance is not just about building better systems. It is about shaping the conditions under which those systems operate, and ensuring that they serve society in ways that are fair, accountable, and ultimately human-centred.

If you are working on one or more of the topics outlined above, we warmly encourage you to submit an abstract or full paper proposal. As a dynamic and evolving research community, we are particularly interested in contributions that bring critical, interdisciplinary, and forward-looking perspectives to the study of algorithmic governance. We welcome work that advances understanding of how algorithms shape decision-making, accountability, human agency, and institutional design across different contexts. By contributing to Area 5, you will help shape emerging debates on governance of, with, by, and through algorithms, and co-create knowledge at the intersection of technology, policy, and society. Join us in advancing this conversation and contributing to the Data for Policy Conference.

Sources

1. Veale, Michael, and Irina Brass, ‘Administration by Algorithm? Public Management Meets Public Sector Machine Learning’, in Karen Yeung, and Martin Lodge (eds), Public Sector Applications, in Karen Yeung, and Martin Lodge (eds), Algorithmic Regulation (Oxford, 2019; online edn, Oxford Academic, 24 Oct. 2019), https://doi.org/10.1093/oso/9780198838494.003.0006.
2. Yeung, K. (2018), Algorithmic regulation: A critical interrogation. Regulation & Governance, 12: 505–523. https://doi.org/10.1111/rego.12158
3. Tan E, Petit Jean M, Simonofski A, et al. Artificial intelligence and algorithmic decisions in fraud detection: An interpretive structural model. Data & Policy. 2023;5:e25. doi:10.1017/dap.2023.22
4. Wieringa M. “Hey SyRI, tell me about algorithmic accountability”: Lessons from a landmark case. Data & Policy. 2023;5:e2. doi:10.1017/dap.2022.39
5. Danaher, J. (2016). The Threat of Algocracy: Reality, Resistance and Accommodation. Philosophy & Technology, 29, 245–268.
6. Hauer MP, Krafft TD, Zweig K. Overview of transparency and inspectability mechanisms to achieve accountability of artificial intelligence systems. Data & Policy. 2023;5:e36. doi:10.1017/dap.2023.30
7. Inga Ulnicane, Governance fix? Power and politics in controversies about governing generative AI, Policy and Society, Volume 44, Issue 1, January 2025, Pages 70–84, https://doi.org/10.1093/polsoc/puae022
8. Papyshev G. Reversing the logic of generative AI alignment: a pragmatic approach for public interest. Data & Policy. 2025;7:e30. doi:10.1017/dap.2025.9
9. Bell A, Nov O, Stoyanovich J. Think about the stakeholders first! Toward an algorithmic transparency playbook for regulatory compliance. Data & Policy. 2023;5:e12. doi:10.1017/dap.2023.8
10. Hossain S, Fernando M, Akter S. The influence of artificial intelligence-driven capabilities on responsible leadership: A future research agenda. Journal of Management & Organization. 2025;31(5):2360–2384. doi:10.1017/jmo.2025.10010
11. Corrêa AM, Elbi A, Garsia S. Better together? human oversight as means to achieve fairness in the european AI act governance — CORRIGENDUM. Cambridge Forum on AI: Law and Governance. 2025;1:e32. doi:10.1017/cfl.2025.10017
12. Sidhu D, Magistro B, Allen Stevens B, Loewen PJ. Why do citizens support algorithmic government? Journal of Public Policy. 2024;44(3):659–677. doi:10.1017/S0143814X24000114
13. Hauer MP, Krafft TD, Zweig K. Overview of transparency and inspectability mechanisms to achieve accountability of artificial intelligence systems. Data & Policy. 2023;5:e36. doi:10.1017/dap.2023.30
14. Flórez Rojas, M. L. (2025). Algocracy in the Judiciary: Challenging Trust in the System. Revista De Estudios Sociales, 93, 107–128. https://doi.org/10.7440/res93.2025.06
15. Flórez Rojas, M.L. (2025). The Shaping of AI Policies in Latin America: A Study of International Influence and Local Realities. In: Goossens, J., Keymolen, E., Stanojević, A. (eds) Public Governance and Emerging Technologies. Springer, Cham. https://doi.org/10.1007/978-3-031-84748-6_12

About the Author:

Maria Lorena Flórez Rojas is a Colombian lawyer, currently working as an Assistant Professor of Law at Groningen University. She is part of the Area 5 Editorial Committee on Algorithmic Governance working across the Data for Policy Conference and Data & Policy journal.

Other members of the Area 5 Committee on Algorithmic Governance include:

• Yanto Chandra (City University of Hong Kong, Hong Kong)
• Asif Gill (University of Technology Sydney, Australia)
• Jacqueline C.K. Lam (University of Hong Kong, Hong Kong)
• David Uriel Socol de la Osa (Hitotsubashi University, Japan)
• Riccardo Vecellio Segate (Groningen University, Netherlands)
• Leid Zejnilovic (Nova School of Business and Economics, Portugal)

Algorithmic Governance Needs More Than Code: Why Policy and Literacy Matter Now was originally published in Data & Policy Blog on Medium, where people are continuing the conversation by highlighting and responding to this story.

By Migle Laukyte, Manuel Portela and Vladimir Estivill-Castro

This blog post coincides with the announcement of the 2026 Data for Policy Conference, which takes place at Universitat Pompeu Fabra, Barcelona on 8–10 September, 2026. View the Call for Papers and consider submitting (deadline: 1 June 2026).

Against the backdrop of shifting global order, uncertainty of international law, growing military technology demands and hybridization of armed conflicts, climate crisis, democracy backlashes, political instabilities, deepening inequalities between Global North and Global South, the questions about governance of disruptive technologies, such as AI, are gaining significance. And yet it is not receiving the necessary attention.

Let us contextualize this question within the reality of how technological innovation is being articulated right now and by whom: here the clash between public sector and private sector is the most evident, but not the only one. Indeed, the private sector — especially thanks to the big tech companies — has emerged as an international superpower, which in many ways shadows and pushes the States aside. This aggressive and ultrarich private superpower has an agenda of its own, where the liberal democratic ideals and values, compliance with regulations, ethics and fundamental rights are not among the priorities and which in many cases wants to substitute the public sector in its rights, tasks and responsibilities. At the same time, regional and national economies need small and medium companies that add value and develop new data-driven and AI-based solutions in a highly competitive market. Paradoxically, a broad spectrum of the economy is using AI from global suppliers, enabling them to fast develop new solutions but increasing technological dependency. The public sector, tight to big budgets and expenditure in foreign services, are questioning these decisions, moving torwards to more sovereign and local technologies.

This is why the question about AI governance is so important: it is not only a question of who is making decisions on AI as such, but also a question about who is shaping the world with and through it, how we create discourses, rules and roles in this new economy. In a world where agents are put into charge, taking important decisions, we should seek of debating how AI is shaping our world. Formulated like this, the question seems too big to address, but we suggest to look at AI governance through the following three bullet points that we unpack in what follows:

1. data governance,
2. infrastructure ownership, and
3. technological sovereignty.

Data governance: AI heavily depends on access to large datasets, besides relying on computational resources and technical expertise and knowhow. Much of the world’s digital data infrastructure is controlled by a small number of multinational technology companies, including Google, Microsoft, Meta and Amazon, which determine what the models learn, how they perform and whose interests count most (leaving massive surveillance and profiling aside). As a result, governments and organizations seeking to build AI systems often rely on external providers and hence have limited ability to ensure transparency, enforce accountability, or protect sensitive information. Data availability is key to develop new AI systems, and controlling data is not only defending the rights of people and companies, but also controlling the economy. Key regulations have been created last years to increase data availability and defining the roles of data-economy actors.

Infrastructure ownership: Training advanced AI models requires enormous computational resources, besides coming at increasingly less and less sustainable environmental costs. These resources are unevenly distributed globally, often concentrated in countries with strong semiconductor industries and home to big tech superpowers. Consequently, AI governance increasingly intersects and clashes with a variety of industrial policies, supply chains, and economic interests. Diverse solutions are approached, from state-based AI Factories, to private-owned companies to push the boundaries of AI. However, all these infrastructures are fragile, needing natural resources, and global presence; even in places of conflict.

Technological sovereignty: The capacity of states or regions to maintain meaningful control over their digital infrastructure and technological capabilities lately have become a focus and an urgency around the world, GAIA X being one among many. However, we should bear in mind that technological sovereignty does not mean building walls and fighting interoperability, but rather the need for strategic autonomy in key digital domains and independence from foreign private powers. Sovereignty could be strategic and expensive at the same time, cutting down ties with the global economy. The dilemma is broad in this sense, and a global network of collaboration could be as much as fragile as the connecting underground cables that bring internet around the world.

This is where the conference Data for Policy 2026 comes into play. This year’s theme is Governance of/with AI: Implications for Data, Infrastructure, and Tech Sovereignty and it will take place on September 8–10 at the Pompeu Fabra University in Barcelona, Spain. The conference calls for participation of those who want to discuss new ideas and approaches and see the need for a responsible and ethical societal response to the aforementioned challenges. Evidence-based research is very welcome, but theoretical reflections will not be omitted. The topic is broad and inclusive, and the discussions promise to be revealing, unsettling and inspiring: there is a big need to meet in person and talk face to face about the questions that bother and affect us all: namely, what role are we going to grant the AI with in our societies, who defines our choices, what trade-offs are on the table and what solutions, ideas or approaches could offer a way forward.

About the Authors:

Migle Laukyte, Manuel Portela and Vladimir Estivill-Castro are all based at Universitat Pompeu Fabra in Barcelona and are local chairs of Data for Policy 2026. Migle Laukyte is Associate Professor in Artificial Intelligence and Law, based in the Department of Law. Manuel Portela is a Research Associate in the Department of Information and Communication Technologies (DTIC). Vladimir Estivill-Castro is a Professor in the Department of Engineering.

A Critical Moment for AI, Data, Infrastructures and Technological Sovereignty or Why Governance is… was originally published in Data & Policy Blog on Medium, where people are continuing the conversation by highlighting and responding to this story.

Generative AI (GAI) is quickly transforming how societies learn, teach, and work. From ChatGPT to image generators and automated tutors, these technologies are becoming embedded in everyday life. But while much of the public conversation has focused on how AI can help students write essays or assist teachers in managing workloads, a deeper and more pressing question is emerging: Will AI create a more inclusive educational future, or will it deepen existing divides?

In our recent study “The future of learning or the future of dividing? Exploring the impact of generative artificial intelligence on higher education”, published in Data & Policy, we investigated this very question. We examined how leading universities are responding to the generative GAI revolution. Our goal was to understand how prepared these institutions are to train students for an AI-driven future and whether current approaches are helping to build a more equitable educational landscape. What we found was both exciting and concerning: there are some significant gaps in readiness, equity, and long-term vision among universities. While GAI offers powerful tools for learning and innovation, its adoption is uneven, and this unevenness risks exacerbating educational inequalities

Generative AI in Higher Education: Promise and Uneven Progress

GAI tools are already making their way into classrooms and research labs. But integrating such technologies into education requires more than simply giving students access to ChatGPT. It demands a thoughtful redesign of curricula, new ethical guidelines, and investments in digital infrastructure. More importantly, it requires ensuring that everyone, not just those in elite institutions or wealthy countries, can benefit from AI-powered learning.

Our research focused on the top 25 universities in Asia, based on global rankings by QS and THE. We reviewed publicly available university policy documents and undergraduate core curricula to assess how these universities were addressing the rise of GAI. We also developed a framework of four essential capacities that students need in the AI era: AI ethics, AI literacy, human-AI collaboration, and human-distinctive skills such as creativity, critical thinking, and emotional intelligence.

Out of the 25 top-ranking Asian universities we studied:

• Just under half (12 universities) had any formal policy on generative AI.
• Chinese universities, despite being global leaders in AI research, had no public GAI policies in their top universities.
• Hong Kong institutions were among the most proactive, with all five top universities having GAI policies in place.
• When it came to actual curriculum reforms, only one university had integrated all four skill areas into their core curriculum.

These policy documents varied widely in their scope and depth. However, very few institutions touched on the broader ethical implications of GAI, its long-term impact on the future of work, or the need for comprehensive curriculum reform. Most policies were reactive rather than visionary, focused more on managing risks than harnessing opportunities. This disconnect raises important concerns. As generative AI becomes more prevalent in workplaces, those who graduate without competent AI literacy may find themselves at a disadvantage. Without targeted policy interventions, GAI could reinforce existing educational inequities between countries, between institutions, and even within classrooms.

The Risk of Reinforcing Inequality in the AI Era

In many cases, GAI-related courses were optional electives, not required for all students. This means that students could graduate from even top institutions without any formal learning about AI ethics, human-AI collaboration, or even basic AI literacy. This means students could complete a degree at some of Asia’s most prestigious universities without ever learning how AI works, how to use it ethically, or how to critically engage with its social and economic consequences. These findings reveal a troubling trend: even among the most prestigious institutions, the integration of GAI into higher education is selective, uneven, and often superficial.

The implications of this divide extend beyond individual students. Higher education plays a central role in preparing societies for technological change. Universities shape the workforce, influence public discourse, and drive innovation. If they fail to adapt to the realities of the AI era, entire nations may find themselves left behind in the global knowledge economy. Conversely, those that harness AI effectively, especially in inclusive and ethical ways, could gain a competitive edge not only economically, but socially and culturally.

This is not to say that universities are unaware of the challenges or opportunities. Many are experimenting with AI in creative ways. Some have introduced AI-driven tutoring systems to support student learning, while others are using GAI tools to accelerate research or improve administrative efficiency. These efforts are promising, but they remain fragmented. What is missing is a coherent, institution-wide strategy that embeds AI literacy, ethics, and collaboration into the fabric of higher education.

Charting a More Inclusive AI Future for Universities

So, what needs to change? First and foremost, universities must move beyond piecemeal approaches and develop comprehensive frameworks for GAI integration. These frameworks should include clear institutional policies, robust ethical guidelines, and a commitment to curriculum reform. AI should not be relegated to computer science departments; it must be part of the general education of all students, regardless of discipline.

Second, universities need to invest in faculty development. Many educators are eager to embrace AI but lack the training or institutional support to do so effectively. Workshops, professional development programs, and interdisciplinary collaboration can help bridge this gap. Faculty are not just users of AI, they are also its interpreters, responsible for helping students understand its implications.

Third, equitable access must be a priority. Not all students have the same access to devices, software, or stable internet. Universities should ensure that AI tools are made available to all, and that students receive the training and support needed to use them. Otherwise, GAI could exacerbate digital divides and leave the most vulnerable further behind.

Fourth, collaborations between universities, both regionally and globally, can play a vital role. Institutions can share best practices, co-develop AI curricula, and even offer joint degrees or certifications. Regional networks can help ensure that GAI adoption does not become a race between elites, but a cooperative effort aimed at raising standards for all.

Finally, we must recognize that the skills needed in the AI era go far beyond technical know-how. While understanding algorithms and data is important, it is equally essential to cultivate the human capacities that AI cannot replicate. Critical thinking, empathy, creativity, and ethical reasoning are more valuable than ever. These are the qualities that will allow students not only to use AI, but to shape it and to ensure that it serves the public good.

Our study is a snapshot of a rapidly evolving landscape. As GAI continues to advance, we expect that more universities will adopt policies, revise curricula, and invest in infrastructure. But the pace of change will not be uniform. Institutional culture, national policy, and resource availability will all shape how GAI is implemented. Without careful planning and a focus on equity, we risk creating a two-speed system of higher education, one where some institutions race ahead, while others struggle to keep up.

This is a critical moment. The choices we make today will determine whether GAI becomes a tool for democratizing knowledge or a driver of deeper inequality. Higher education has a unique opportunity and responsibility to lead this transition. If universities can rise to the challenge, they can help shape a future where AI enhances learning, empowers students, and strengthens societies. But if they fail to act, the promise of AI could become just another line in the growing story of educational division.

The future of learning is in flux. It is up to us to decide whether it will be a future of opportunity or a future of division and exclusion.

References:

Wong, W., Aristidou, A., & Scheuermann, K. (2025). The future of learning or the future of dividing? Exploring the impact of generative artificial intelligence on higher education. Data & Policy, 7, e44

About the authors:

Wilson Wong is the Founding Director and Associate Professor of the Data Science and Policy Studies Programme, School of Governance and Policy Science, the Chinese University of Hong Kong. He is also a Senior Research Fellow of the School of Management, UCL.

Angela Aristidou is an Assistant Professor of the School of Management at UCL. She is also a Faculty Fellow at the Stanford Digital Economy Lab and the Stanford Institute for Human-Centered AI, Stanford University.

Will AI Bridge or Broaden the Gap in Higher Education? was originally published in Data & Policy Blog on Medium, where people are continuing the conversation by highlighting and responding to this story.

The 2025 Europe edition of the Data for Policy conference, held 12–13 June at Leiden University in The Hague, gathered researchers, policymakers, and practitioners around the theme “Twin Transitions in Data and Policy for a Sustainable and Inclusive Future.” Over two days of presentations and discussion, participants explored how data systems and digital technologies are reshaping governance and what it will take to make them serve climate goals, social equity, and democratic accountability.

Here are five reflections from the event.

1. Europe Wants to Align Digital and Green Goals, but the World Isn’t All In.

Andrea Halmos of the European Commission opened the conference with a call to integrate Europe’s digital and environmental agendas. She argued that digital tools and sustainability can be mutually reinforcing, supporting emissions tracking, predictive resource management, and real-time environmental monitoring. She urged public institutions to lead this integration, pointing to initiatives like climate-neutral data centers and digital twins for monitoring circular economy efforts.

Halmos warned, however, that local governments lack the technical capacity to implement these goals. Municipalities are missing up to 70 percent of the skills needed to meet EU climate and innovation targets. Her keynote also highlighted an underlying tension: green innovation depends on digital infrastructure, but that infrastructure through energy use, material demands, and surveillance risk can undermine climate commitments. The EU is developing initiatives such as the Cloud and AI Development Act and sustainable digital standards to manage these tradeoffs, but Halmos emphasized that implementation will lag without stronger coordination and investment in local operational capability.

A comparative study by researchers at the University of Copenhagen added an international perspective. Analyzing how Europe, the U.S., and China frame “green IoT” in policy, they found that only Europe treats digital and environmental goals as interlinked. U.S. policy focuses on market growth and national security, while China emphasizes industrial expansion and platform integration. These divergent priorities shape how each region approaches tech development, regulation, and collaboration.

2. AI and Digital Systems Must Be Built Around Justice Instead of Efficiency

Several sessions pushed back on the idea that digital innovation alone will deliver a just and sustainable future. Instead, they asked whether today’s AI and data systems are even compatible with those goals.

Researchers from University College Dublin analyzed how EU and national policies fail to account for the environmental impacts of AI infrastructure. Focusing on Ireland’s data center boom, they showed that hyperscale facilities, essential to AI development, consume nearly a quarter of the country’s electricity while operating with limited transparency. Regulatory blind spots and trade secrecy claims, they argued, sustain an artificial separation between AI applications and the infrastructure that supports them. This disconnect hampers governance and obscures the true operational costs of emerging technologies.

Barrie Sander of Leiden University extended this critique, describing AI as an active participant in climate governance: a resource consumer, a tool for mitigation and adaptation, an enabler of securitized responses to climate migration, and a shaper of narratives through corporate lobbying. Rights-based regulation, he argued, may protect individuals, but it falls short against systemic risks and planetary boundaries. What’s needed is a shift that enables legal frameworks to match the scale and systemic logic of industrial AI.

Researchers from the Tilburg Institute of Law, Technology & Society added a sectoral view, focusing on the EU’s clean energy transition. Their work showed how digitalization is framed in technocratic terms optimizing for competitiveness or system efficiency , while sidelining civic democratic values like equity and justice. Market-based tools such as tariff adjustments or digital consumer dashboards cannot address deeper problems like energy poverty or infrastructure inequality. They called for a re-framing of the twin transitions that centers inclusion and justice rather than optimization alone.

3. Data Governance Aims High, But Stumbles on the Ground

Europe’s data policies often seek to promote the public good. Yet multiple presentations showed how those same frameworks can stall sustainability innovation, particularly in cases that depend on using personal or biometric data.

Andres Chomczyk Penedo and Anna Capellà Ricart examined how GDPR complicates the development of extended reality (XR) tools for green innovation. These tools can support urban planning or energy management, but regulatory uncertainty, limited infrastructure for managing consent, and privacy-by-design constraints restrict experimentation and slow deployment.

Adrianna Michałowicz offered a similar critique of the EU’s Data Governance Act. While it encourages voluntary data sharing through mechanisms like data altruism, she found that participation has been low. Weak incentives, vague definitions, and a lack of institutional support have left the concept underdeveloped. Unless trust and accountability mechanisms improve, these policy tools may struggle to deliver.

4. Decentralized Models Offer a Way Forward

Despite these regulatory barriers, several presentations showcased alternatives that put privacy, participation, and sustainability at the center of digital design.

Researchers from Dublin City University shared insights from the SHINE project, which uses indoor air quality sensors in social housing. To overcome concerns about data privacy and loss of control, the team developed a sensor system that embeds transparency and community control directly into its design. Privacy was not treated as an afterthought but as a foundational principle, shaping both the technical architecture and how residents engage with the system.

At the London School of Economics, researchers explored person-centric data architectures like Personal Data Stores. These tools allow individuals to manage how their data is shared, offering an alternative to corporate surveillance and centralized state data collection and creating space for data practices that are granular, participatory, and accountable. Their project examined how these systems could contribute to national statistics and circular economy metrics while allowing people to choose what data they share, with whom, and for what purposes.

Adding an institutional dimension, Federico Bartolomucci shared research performed with co-authors on the MIT City Science Network. This network links urban innovation labs across multiple countries, each serving as a testbed for participatory data practices shaped by local needs. Through interviews and comparative analysis, the researchers found that these labs succeed not by standardizing approaches but by fostering adaptable governance, shared learning, and community-led experimentation. The result is a decentralized model of innovation that prioritizes collective outcomes without imposing uniform solutions.

5. Public Institutions Need Better Ways to Access Private Data

Many of the data sources most relevant to public policy ,on mobility, consumption, platform use are held by companies. Governments need this data to make informed decisions, but access remains limited, fragmented, and often informal.

A dedicated track at the conference tackled this issue. Researchers from Utrecht University and the University of Manchester analyzed responses to the EU’s public consultation on the Data Act. They found a sharp divide: public actors favored mandatory sharing for public interest purposes, while private firms favored voluntary sharing, safeguards, and price-setting mechanisms. While both sides acknowledged the problem of asymmetrical access, there was no consensus on how to address it.

Another Utrecht team examined B2G dynamics in Amsterdam and Barcelona. Drawing on discourse analysis, they found that trust gaps and unclear legal frameworks often stall collaboration. While governments try to insert data-sharing clauses into procurement and permitting, companies remain cautious without clear incentives.

The GovLab offered a possible bridge. Adam Zable presented the “9Rs Framework,” a taxonomy of business motivations for sharing data, including Reputation, Reciprocity, Regulatory compliance, and Revenue generation. The framework offers a structured way to identify and surface what drives companies to share data, aiming to help policymakers and private organizations design data collaboration initiatives that benefit all parties. Zable outlined plans to test and refine the framework through interviews with corporate data stewards.

Conclusion

Data for Policy 2025 made clear that aligning digital and green transitions will require not only deploying new tools but, more crucially, rethinking the infrastructures, incentives, and institutions that govern how data is produced, accessed, and used. Across sectors, presenters emphasized that a deep shift is needed toward data practices that are participatory, transparent, and accountable to the communities they affect. For the twin transitions to succeed, governance must be shared, values clearly stated, and technology anchored in public purpose.

About the Author:

Adam Zable is a Research Fellow at The GovLab, where he explores the intersection of emerging technologies, data governance, and democratic innovation. He previously served as Director of Emerging Technologies at the Digital Trade and Data Governance Hub, where he led the Global Data Governance Mapping Project and organized international conferences on XR policy and generative AI governance. Adam holds a Master of Public Policy from the Willy Brandt School of Public Policy and is based in Freiburg, Germany.

Five Takeaways from “Data for Policy 2025 Europe” was originally published in Data & Policy Blog on Medium, where people are continuing the conversation by highlighting and responding to this story.

The European Commission’s AI Continent Action Plan, released on April 9th, sets another milestone in the EU’s long march towards digital sovereignty. The Plan acknowledges the key role of data in ‘unlocking the full potential of AI’ and introduces a new Data Union Strategy to address the current ‘scarcity of robust and high-quality data’ needed for training and validating AI models (Action Plan, p. 11).

The EU has never been shy when it comes to launching data-related initiatives: the 2020 European Data Strategy, the Common European Data Spaces (CEDS), or Gaia-X are the most notable examples. Yet, the implementation of these programs and initiatives has faced significant challenges that are far from being solved. At this juncture, the announcement of the new Data Union Strategy begs the question: will it overcome the shortcomings of its predecessors, or will it struggle with the same difficulties?

Potential strengths

First, the proposed Data Union Strategy seeks to overcome the scarcity of high-quality data, an indispensable condition for the EU to close the ‘innovation gap’ and claim a frontrunner position in the AI global race. Compared to the broad objectives of the 2020 EU Data Strategy, the Data Union focus on AI development is targeted, application-oriented, and explicitly aligned with the AI policies that the EU is deploying. Furthermore, the strategy’s emphasis on ‘streamlining existing data legislation’ indicates a clear awareness of the regulatory challenges that keep hampering previous EU data initiatives.

Second, to address the persistent issue of data fragmentation, the strategy introduces the Data Labs, a new tool for federating data from various AI Factories that should also provide essential services to AI developers. The Data Labs are set as key components of the AI Factories initiative, acting as central hubs that bring together and organise data originating from various AI Factories operating within the same sector. Furthermore, Data Labs will be linked to the CEDS, so that the data held within these sectoral ecosystems can be accessed by AI developers. This interconnected approach aims to provide AI developers with access to substantial volumes of high-quality data across diverse domains (up to 14 so far). Beyond providing access to data, the Labs are also expected to offer a range of value-added services–e.g. cleaning and enriching datasets– and technical tools–e.g. standardised formats, synthetic data, shared technical building blocks (idem). Similarly, the Data Labs are expected ‘to offer data-pooling services that would help companies to share data while adhering to antitrust rules, drawing on the Data Governance Act framework for trusted data intermediaries’ (idem). Ultimately, the goal of Data Labs aligns with the CEDS mission–facilitating the cross-border sharing and exchange of data across sectors–while specifically addressing the needs of AI research and development.

Third, the Strategy acknowledges the importance of specific data types, such as language data and health data. It mentions the Alliance for Language Technologies (ALT-EDIC) and the European Health Data Space (EHDS), respectively– as examples where the pooling of data can deliver tangible results, and shows a use-case driven approach that could foster engagement and demonstrate the real-world benefits of the strategy.

Persistent risks and unresolved challenges

While the Data Union Strategy sets the clear goal of facilitating access to ‘robust and high-quality data’, it provides limited specifics on how we will get there.

A comprehensive approach to data quality requires more than providing the services that Data Labs could offer. Apart from cleaning and enriching data, adopting technical data standards and validation processes, there is a need for ongoing governance of data quality throughout its lifecycle. For the Data Union Strategy to be truly effective it will need comprehensive data stewardship protocols and governance frameworks for data quality and integrity. Subsequent versions of the strategy could dig a bit deeper on how these two aspects need to be deployed. In doing so, it might be useful to combine top-down tools (Data Labs, Simpl) contracted by the European Commission with bottom-up, decentralised initiatives by local and sectoral stakeholders.

How the Data Union Strategy implements the security and privacy safeguards it promises, in the context of large-scale data sharing across diverse sectors, will also be critical. Simply stating the need for these safeguards is insufficient, and the next versions of the strategy will have to provide detailed information on the specific technical and organisational measures that will be put in place to guarantee alignment with the GDPR (in its current or future form) the Data Act, and the Data Governance Act. A particular challenge will arise in handling mixed datasets, which contain a combination of personal and non-personal data. The Strategy will need to clearly articulate how it intends to navigate the complexities of these mixed datasets to enable effective data reuse while simultaneously ensuring full compliance with all relevant privacy regulations. Again, there will be a need not only for anonymisation or pseudonymisation requirements, but also for robust data governance frameworks. In this regard, the Data union could consider how to embed social license devices for data reuse that include participatory mechanisms. Linked data requires linked democracy.

The Strategy’s emphasis on data pooling, while beneficial for AI development, could also raise antitrust risks if dominant firms end monopolising access to Data Labs. The Data Union must ensure that its explicit reference to safeguards like the ‘trusted intermediaries’ of the Data Governance Act actively prevent that no one engages in competition-restrictive practices (articles 101 and 102 of the TFEU). The strategy should therefore consider incorporating specific measures designed to prevent the emergence of new data monopolies or the unfair leveraging of existing data advantages by dominant market players.

Finally, balancing openness and security is always challenging. The Strategy’s goal to “attract more valuable data” while protecting sensitive information echoes tensions in the GDPR’s international data transfers. Without clear guidelines, businesses may face uncertainty, stifling data exchanges and cross-border collaboration.

A polycentric approach to the Data Union

The efforts of the Data Union Strategy to foster data sharing and innovation risk being undermined by top-down, rigid governance structures, echoing the challenges faced by Gaia-X. The Strategy is silent on any lessons learnt from Gaia-X, which raises doubts on how its achievements (e.g. its open source technical stack) and shortcomings (among the many noted in this EuroStack piece, the ‘lack of functional data spaces’) will inform the Data Union.

The success of the emergent European Health Data Space will also depend on the consistent compliance of member states, alignment of national contact points and Health Data Access Bodies. The application timeline of the regulation (extending into 2031) gives an idea of the overall complexity of this process, and it is not clear how implementation will align with the deployment of the Data Union Strategy.

A top-down approach to the Data Union governance could also inadvertently prioritise the creation of broad sectoral federations at the expense of localised innovation and niche but valuable use cases. For instance, the development of regional language models, crucial for linguistic diversity and inclusion, might be sidelined if the focus remains solely on pan-European data spaces. This inherent prioritisation could stifle the very grassroots innovation the Data Labs aim to impulse.

Furthermore, the limited membership of initiatives like ALT-EDIC (currently 17 member states) raises concerns about inclusivity and the true scope of the Data Union Strategy’s impact. By potentially missing the full linguistic diversity and specific data needs of non-participating countries, the very goal of ‘cultural preservation’ risks being compromised.

A more effective path forward for the Data Union Strategy lies in adopting a hybrid, polycentric approach to governance, empowering regional and sectoral bodies with greater decision-making capacity. This would strike the right balance between ensuring EU-wide coherence and being flexible enough to incentivise active participation from cities, regions and private sector actors. To avoid becoming a repurposed echo chamber of previous initiatives, a Data Union Strategy has to evolve into a decentralised Data Poleis Strategy, one that integrates the diversity of its local and regional centres across the continent.

About the author

Marta Poblet is Senior Research Lead at The Data Tank and Adjunct Professor at RMIT University. She co-founded the Institute of Law and Technology at the Autonomous University of Barcelona and was a researcher at ICREA (Catalonia). She holds a JSD in Law and a Master in International Legal Studies from Stanford University. Her work focuses on the intersection of law, political science, and technology, with expertise in data, AI, distributed systems, and democracy theories. She has led numerous research projects and published over 80 academic works on these topics.

The Data Union Strategy: Game Changer or Déjà Vu? was originally published in Data & Policy Blog on Medium, where people are continuing the conversation by highlighting and responding to this story.

Publisher’s note: This post concerns the main theme of Data for Policy 2025 (DfP’25) — Europe Edition, which will take place in The Hague, The Netherlands, on 12–13 June 2025. The submission deadline for abstracts, full papers and panel proposals is this week: 21st February, 2025.

The twin transitions — digital and green — are not just buzzwords; they represent the intertwined evolution of technological innovation and environmental responsibility, reshaping how societies pursue progress and sustainability. The digital transition, facilitated by advancements in AI, data analytics, and smart technologies, opens up unprecedented possibilities for efficiency and innovation. At the same time, the green transition, driven by ambitious climate targets and sustainable practices, aims to address pressing ecological challenges and fosters long-term resilience. At their interface, big data, more computing power and advances in complex models allow for better capture of complex ecological systems and thereby provide an evidence base for sustainability policies. Furthermore, the capability for better monitoring and incorporating more inputs (in frequency and diversity), allows for better monitoring and the (near) real time identification and analysis of potential anomalies.

Yet, these simultaneous transitions also bring significant risks and trade-offs that need careful navigation. Digitalization can drive efficiency, optimize production processes, and enable the dematerialization of goods and services, contributing to a more resource-conscious economy. However, these benefits are counterbalanced by risks such as rebound effects, where efficiency gains lead to increased consumption, and significant environmental costs. Furthermore, the high energy demands of digital technologies, the growing accumulation of e-waste, and the carbon footprint of emerging innovations like cryptocurrencies and artificial intelligence highlight the tensions between technological progress and ecological sustainability. Thereby, the public sector plays a crucial role in guiding both the digital and green transitions.

We want to highlight three trade-offs specifically: energy versus emissions, regulation versus innovation, and inequality versus inclusion.

Starting with energy versus emissions, there is a potential clash with the urgent goal of achieving net-zero emissions and the expansion of digital infrastructures, like data centers that come with immense energy demands. Coming from a fossil-emissions-based economy, the digital economy offers models that can be sustainable. However, it does lead to increasing energy demands, and the digital economy’s markets provide for little incentives to use renewable energy (e.g. there are “market-based” emissions reporting, the official figures companies disclose, while ignoring “location-based” emissions — the actual carbon footprint of data processing at specific sites. This leads to the ironic situation that one of today’s key transitions may thwart the other.

Second, there is at times tension between regulation and innovation. This means that policies have to strike the right balance between encouraging innovation and upholding essential rights such as privacy, transparency, and fairness. Digital transitions show that with correct incentives, markets can achieve rapid transitions. This also means that governments need to continuously monitor and regulate certain energy markets in particular to create the correct incentives in the context of twin transitions. Past developments have shown that ‘when incentives are introduced, they tend to stimulate innovation and transitions’.

In terms of inequality versus inclusion, there is concern that the costs and benefits of the digital and green transitions are unevenly distributed. As the global economy increasingly relies on digital skills, 2.6 billion people still lack reliable internet access, while 760 million remain without electricity. At the same time, the shift to a green economy threatens the livelihoods of communities dependent on fossil fuel industries, highlighting the socio-economic disparities created by these transitions.

While these challenges are global, effective solutions must be tailored to regional contexts. From the EU’s Green Deal, AI Act and Data Act to policies in Asia, Africa, and the Americas, responses to the twin transitions require both international collaboration and tailored approaches to local needs.

Data and Policy at the Nexus of Transformation

Data lies at the heart of the twin transitions, driving real-time insights, evidence-based policymaking, and cross-sector collaboration. It has the potential to bridge digital innovation and sustainability, empowering stakeholders to craft solutions that are both effective and inclusive. However, this potential also comes with risks: without robust governance, data can perpetuate biases, deepen inequities, and create inefficiencies that hinder progress.

Government policies can encourage innovation and the adoption of digital technologies that support environmental sustainability while simultaneously implementing regulations to mitigate their ecological impact. This includes promoting energy efficiency, expanding the use of renewable energy sources, and ensuring the responsible management of e-waste to balance technological progress with sustainability goals. The European Commission has prioritized the twin transitions, defining climate and environmental challenges as a ‘generation’s defining task’ while recognizing digital transformation as a ‘key enabler for reaching the Green Deal Objectives’.

Data for Policy in Action: European Data for Policy Regional Conference 2025

To reflect the increasing urgency of integrating digital and green transitions into policymaking, this year’s European Data for Policy Regional Conference 2025 (June 12–13, 2025) will explore “Twin Transitions in Data and Policy for a Sustainable and Inclusive Future”. Convening in The Hague, The Netherlands, this conference aims to provide a space where researchers, policymakers, and practitioners can address the pressing questions posed by these transitions. The conference will serve as a platform for debate, collaboration, and innovation — focusing not only on current possibilities but also on shaping a sustainable and inclusive future. Bringing together experts from diverse fields, the conference seeks to examine the role of data and policy in driving a future that is not only sustainable but also just and inclusive.

About the authors

Laura Zoboli is Assistant Professor of Commercial Law in the Department of Economics and Management at the University of Brescia. She is also an Area Editor (Focus Area 3: Focus on Policy & Literacy for Data) of the open-access journal Data & Policy at Cambridge University Press.

Sarah Giest is a Professor of Public Policy with a focus on Innovation and Sustainability at the Public Administration Institute of Leiden University. She has been developing an interdisciplinary research agenda and teaching on using a policy lens to look at technical, environmental and social solutions that shape sustainable societies. She is an Editorial Board Member and Area Lead (Focus Area 1: Digital & Data-Driven Transformations in Governance) of the open-access journal Data & Policy at Cambridge University Press.

Bram Klievink is Professor of Public Administration at Leiden University, with a special focus on Digitalisation and Public Policy. He has a strong fascination with the interface between digitalisation and government. His research is about how digital innovations challenge the incumbent practices and institutions of public governance, and how digitalisation might be used in novel governance arrangements for the digital age. His current research include several projects on the politics and use of big data and algorithms in a public governance context. He is an Area Editor (Focus Area 5: Algorithmic Governance) of the open-access journal Data & Policy at Cambridge University Press.

***

This is the blog for Data & Policy (cambridge.org/dap), a peer-reviewed open access journal published by Cambridge University Press in association with the Data for Policy Conference and Community Interest Company.

Beyond the Buzzwords: Why Twin Transitions Demand Our Attention was originally published in Data & Policy Blog on Medium, where people are continuing the conversation by highlighting and responding to this story.