Guest Blog: Fair synthetic data is not about fairness

by Mykhaylo Bogachov

Bogachov, M. (2026). Fair synthetic data is not about fairness. Big Data & Society, 13(3). https://doi.org/10.1177/20539517261473353

"Data" usually means evidence of what the world is like. Modern AI is trained on unfathomably large datasets, and this makes its impressive capabilities easy to explain: when AI works well, it is because it has picked up meaningful patterns in records of the real world. There are, however, many skills that AI cannot learn from existing data. This is because either the data describing them does not exist, or the skills are so complex that humanity cannot produce enough data for AI to learn them successfully. In response, the AI industry increasingly turns to so-called synthetic training data, one that does not represent actual events or phenomena, but depicts how the world could or should look. Such data is generated using AI itself and is usually very similar to "real" data. Training on it often allows AI companies to cut costs and improve benchmark results at the same time. Still, there are also big risks. Over-reliance on synthetic training data can make AI systems perform poorly in the real world, leak private information, and amplify biases.

Bias in AI systems is studied in a research field, known as algorithmic fairness. There, researchers focus on developing measures that help AI systems, especially those used for decision-making, adhere to the political value of fairness. Some researchers believe that synthetic training data, if used right, can make AI systems less biased than they would be if trained on real data. In my paper, I analyse several proposed approaches to improving fairness in AI through synthetic training data. It turns out that synthetic data engineers tend to misunderstand the nature of the political problem they aim to solve, and end up achieving something else instead. Fair synthetic data promises benefits to society, but in practice it serves the interests of AI developers and deployers, allowing for cheaper training or deployment of AI, just as other synthetic data techniques that do not aim for fairness. So, many of the approaches I've looked into should not be expected to make AI meaningfully fairer. Still, some of the underlying conceptual and technical innovations are genuinely promising, and I discuss how they might be used to make AI substantively fairer in the future.

Guest Blog: Routinized Data Activism: Citizen Data Practices and Everyday Data Citizenship in South Korea

by Danbi Yoo

Yoo, D. (2026). Routinized data activism: Citizen data practices and everyday data citizenship in South Korea. Big Data & Society, 13(3).

You still believe that code and data can make the world better. You’ve found like-minded developers who gather after work to build civic tech projects. But after long workdays, unstable freelance jobs, and Seoul’s rising costs, civic hacking can feel like one more unpaid task added to an already exhausted life.

So when, and how, can data activism happen? And how can such engagement become durable—recurring rather than one-off—without reproducing the very pressures it seeks to resist?

My article grows out of fifteen months of ethnographic research with Nullfull, South Korea’s longest-running grassroots civic hacking community. What drew me to Nullfull was not only what its members built, but how they kept showing up every Monday. A misleading news chart could become a GitHub correction; entering receipt data during an idle moment could help map lawmakers’ public spending; and a shared meal at a matjip tour could turn socializing into critique.

I call this routinized data activism. It is not simply the idea that mundane practices can be political. Rather, it explains how small, situated contributions become durable when linked through repetition, peer learning, and lightweight coordination. In Nullfull, Monday meetings and online exchanges folded individual attentiveness back into shared projects, turning ordinary acts into interventions addressed to politicians, media organizations, and corporations.

This shifts where we look for data citizenship: not in technical competence or access to data alone, but in repeated practices of noticing, caring for, interpreting, correcting, and acting on data together, within the labor and political conditions of everyday life.

For activists struggling to open even small cracks in data capitalism, and for ethnographers studying the overlooked textures of everyday life in a fast-moving academy, this article is written in solidarity and gratitude.

Nullfull's catchphrase — "French fries matter more than the work!"

A matjip tour tracing lawmakers' expense-account dining

Guest Blog: From Rule of Law to Rule of Algorithm: Generative AI’s Threat to Democracy

by A.T. Kingsmith

Kingsmith, A. T. (2026). From rule of law to rule of algorithm: Generative Artificial Intelligence’s threat to democracy. Big Data & Society, 13(2).

What drew me to this research was a fundamental problem: the same technology people use to write emails or plan holidays is now deciding who receives welfare, whose visa application succeeds, and what citizens are told their government’s policies mean for them.

Generative AI systems like Microsoft’s Azure OpenAI are already embedded in government services across many countries. These technologies are drafting policies, handling benefit inquiries, and generating citizen communications. My main concern is that this represents a shift in governance procedure and logic from traceable rules and human deliberation toward outputs that nobody fully controls and few can meaningfully scrutinize.

The algorithmic systems of the 2010s (Australia’s Robodebt, the Netherlands’ SyRI) were troubling because they removed human judgment from key decisions. But citizens were still formal legal subjects. There was something to appeal, someone to confront. What changes with generative AI is that the system does not merely classify citizens, it shapes the world they encounter: the explanation of a policy, the framing of a consultation, the letter that nudges someone away from a benefit claim they were entitled to make.

Canada’s Chinook immigration system illustrates the accountability problem. When the tool produced disproportionate rejection rates for applicants from African countries, no one could explain it. The department blamed the tool; the developers said it only assisted human decisions. Applicants had nowhere to go. This is not a peripheral outcome. It is what happens when opaque commercial systems are dropped into public administration without anyone being made responsible for what comes out.

Courts have pushed back, and the EU AI Act sets useful limits. But regulation that arrives after deployment is playing catch-up. What I take from this research is a concern less about any particular AI system than about the general direction of this shift. When the technology that governs people’s lives cannot be closely examined, and responsibility for its decisions dissolves between vendor and state, something essential to democratic life is being lost. Most societies have not yet stopped to decide whether that is a trade they want to make.

Guest Blog: The Data Fix: AI, Farmland, and the Future of Agricultural Transformation

by Lin Zhang and Tu Lan

Zhang, L., & Lan, T. (2026). The data fix: Smart farming and the sociotechnical politics of datafication and assetization. Big Data & Society, 13(2).

As AI becomes the defining infrastructure of our moment, its footprint increasingly extends beyond cities and server rooms into production landscapes. Around the world, farmland is under growing pressure—not only from climate change and urban expansion, but also from the rapid buildout of data centers, digital infrastructure, and new forms of platform coordination. At the same time, agriculture itself is being transformed into a site of data extraction. Farming is no longer simply about producing food; it increasingly involves producing streams of data.

Much of the existing research on digital agriculture has focused on North America, where agricultural datafication unfolds through histories of settler colonialism, private property, and financialized land markets. In these accounts, data often operates as a mechanism of enclosure, speculation, and accumulation.

But what happens when land cannot be fully assetized, and where the financial market is heavily regulated?

In our recent article, we turn to China to explore this question. Drawing on fieldwork in the mushroom industry of Gutian, Fujian Province, we introduce the concept of the data fix: the use of data infrastructures to manage economic, social, and ecological contradictions under digital capitalism. Through an ethnographic case study of the expansion of the national agricultural digital brain project with Gutian’s mushroom farming as a pilot site, we explore the rise of data-human assemblage in agricultural labor and the changing land politics as the local state remobilizes resources to pivot towards the new data and AI economy.

Mushroom Digital Brain Platform on Display

Inside the Tremella Farming Factory

Rather than transforming farmland directly into speculative assets, China’s model—shaped by collective land ownership, state coordination, and platform experimentation—redirects datafication toward governance, visibility, and territorial management. This hybrid data governance cannot be reduced to either market transition or technological catch-up. We argue that contemporary Chinese agricultural datafication unfolds through tensions between residual socialist commitments to social and regional equity and renewed pressures for technological developmentalism and cybernetic futurity. In this sense, projects such as the Agricultural Digital Brain are not simply new infrastructures; they are also continuations and mutations of earlier post-reform experiments in governing through information, feedback, and technocratic coordination.

The result is not an alternative to digital capitalism, but a different pathway through it—one in which data becomes both an asset and a way of governing land, labor, and rural futures.

Young Generation of Farmers Selling Mushrooms via Social Media

 

Guest Blog: Digital passing: “I died once, so I could live. Perhaps that is my real story”

by Beata Paragi, Corvinus University of Budapest

Paragi, B. (2026). Digital passing: “I died once, so I could live. Perhaps that is my real story”. Big Data & Society, 13(2).

Cover stories can save lives. In her recently published “Digital passing: “I died once, so I could live. Perhaps that is my real story,” Beata Paragi approaches digital identity fraud from the perspective of vulnerable individuals accused of committing identity fraud as a criminal offence. Inspired by literary works and using insights from Holocaust-studies, this paper recalls the matter of identity performance during the Holocaust – known as passing with assumed identities – and conceptualizes digital passing by exploring similarities and differences regarding imagination, resistance and survival.

Persecuted Jews and Roma could survive the Holocaust and Porajmos by passing – assuming others’ identities, hiding and/or preserving anonymity – a desperate strategy that not only relied on past practices, but is also echoed by contemporary stories. Today’s migrants whose citizenship does not guarantee the adequate level of legal protection often have to make similar choices to escape violence, abuse, hunger or early death.

The places where people find 'refuge' – crowded refugee camps, detention centres, and border regions from the Mediterranean to Australia – are often brutal, the inhuman conditions of which are documented by Forensic Architecture and Border Forensics, among others. As a result, millions trapped between entry denial at the border and danger at home are expected to prove persecution they may not be able to document, not least to narrate for the very nature of any trauma. They are not necessarily persecuted by law, nor do they wear visible markers like the yellow stars, but diverse digital and biometric technologies can reveal, verify, or even infer origins from biometric data, making origin and anonymity harder to preserve, and making the 'narrating self' redundant for decisions on eligibility.

 At normative level everyone should enjoy human rights protection, but in practice citizenship (or its very lack) can block not only safe travel but the very chance to a life without the risk of premature death. The tension between lived experiences and human subjectivities on the one hand and obsession with tech-accelerated identification on the other hand sits at the heart of several urgent conversations today, such as data justice, critical criminology, autonomy of migration, and the technologies of migration and border control. The conceptualisation of digital passing helps us see how human lives become disposable/worthless as a result of beliefs in security, financial incentives and tech-solutionism and how survival strategies are adapted to the digital/biometric conditions.

Guest Blog: How authoritarian ideas about AI governance shape global debates

by Gregory Asmolov

Asmolov, G. (2026). The rise of AI sovereignty: Authoritarian technological imaginaries as a form of reflexive control. Big Data & Society, 13(2), 20539517261426

Much of the current debate about the contribution of artificial intelligence to the rise of digital authoritarianism focuses on its role in disinformation, propaganda, and surveillance. But authoritarian influence may also operate at a deeper level, by shaping how AI itself is governed. This question motivated my recent article in Big Data & Society.

The purpose of the article is to explore how authoritarian ideas travel across borders and influence global AI governance. Rather than focusing on the export of technologies and AI-enabled practices, it examines mechanisms of authoritarian diffusion through informational influence, showing how narratives and ideas shape the forms AI governance takes.

The analysis focuses on the rise of “AI sovereignty” as a central concept in global debates. Drawing on the notion of sociotechnical imaginaries and the concept of reflexive control from Soviet strategic thought, the article shows how political actors can influence others by shaping perceptions of technological futures.

Using Russia as a case study, the article analyses authoritarian imaginaries that inform approaches to AI governance. It identifies three dominant narratives: AI as a tool of global power and domination; “Western AI” as a cultural threat; and AI as a driver of state efficiency. It then examines how these narratives may extend beyond Russia, influencing global debates and contributing to the spread of sovereignty-centred approaches to AI governance.

The main argument is that these imaginaries influence the cognitive environment of global policymaking. As they circulate, they promote sovereignty-centred and security-driven approaches to AI governance, even beyond authoritarian contexts.

Recognising this diffusion mechanism highlights the need for greater sensitivity to the potential impact of authoritarian imaginaries on global models of AI governance. Developing such awareness is essential for mitigating the subtle spread of authoritarian logics through policy debates and regulatory frameworks. The article contributes to this effort by advancing a theoretical understanding of how diffusion operates through informational influence, and by showing specifically how it can shape the governance of AI across different political contexts.

Guest Blog: AI Failure Loops in devalued work: The confluence of overconfidence in AI and underconfidence in worker expertise

by Anna Kawakami

Kawakami, A., Taylor, J., Fox, S., Zhu, H., & Holstein, K. (2026). AI failure loops in devalued work: The confluence of overconfidence in AI and underconfidence in worker expertise. Big Data & Society, 13(1), 20539517261424164.

From headlines to prediction markets, AI’s impact on work is increasingly framed as a question of which jobs will disappear. Yet, in many workplaces, the more immediate issue is not displacement but failure. In historically devalued fields such as K-12 teaching, social work, and home health care, AI systems are often built on reductionist understandings of human work, leading to deployments that fall short in patterned ways. We describe these recurring breakdowns as AI Failure Loops: dynamics in which the devaluation of worker expertise shapes flawed AI systems—and those systems, in turn, further erode recognition of that expertise. 

We conduct a focused review of academic and grey literature on AI deployments across three domains of devalued, feminized labor 1 in the United States (social work, home health, K-12 teaching), grounded in our team’s direct experiences studying and designing AI alongside workers in these fields. Through three case studies—AI-based risk assessments in child welfare, AI for home healthcare, and AI tutoring systems in K-12 teaching—we illustrate how AI Failure Loops arise in practice. We identify a set of six interconnected failure modes in the design, development, and deployment of AI systems that underlie AI Failure Loops: Expertise Misunderstanding, Managerial Over Worker Needs, Design Exclusion, Inappropriate Evaluation, Forced Use, and Unwarranted Blame (Figure 1). 

Figure 1: A visual overview of AI Failure Loops. The six failure modes (in circles) that contribute to the dynamics of the AI Failure Loops exist within a web to illustrate the inter-connected relationship amongst the failure modes. 

These failures are not isolated problems, but arise from a broader tension: AI Failure Loops emerge from the confluence of overconfidence in the capabilities of AI systems and underconfidence in the skills and expertise of workers. Today, with recent waves of both AI hype and actual increases in AI capabilities, developers are targeting a rapidly expanding range of socially complex tasks across feminized and other devalued occupations. Drawing lessons from past AI deployments in these contexts, we argue that this may further accelerate gross underestimations of worker expertise and overestimations of AI capabilities. 

Importantly, nothing about the future of work is inevitable. By genuinely centering and uplifting workers in the design, evaluation, and governance of AI systems, we can begin to reverse these loops and develop AI that actually supports and dignifies, rather than diminishes, worker capabilities. Our paper ends with implications for shifting towards this future of pro-worker AI—a future where worker expertise is celebrated and uplifted through AI practice, and by society more broadly. 

1 Feminized labor is a particularly extreme example of devalued labor; historically, it has been mislabeled “women’s work,’’ and today, these jobs continue to be predominantly performed by women and people of color. Workers in these fields frequently report feeling undervalued and overworked, with many roles experiencing high retention rates.