Digital platforms – Data Big and Small

Data work in Egypt: Who builds AI behind the scenes?

We know this now: artificial intelligence is not only a Silicon Valley product. When trying to look further, research and the media have found AI’s “hidden workforce” (the data workers who label images, transcribe audio, and evaluate content to train machine learning models) in countries like Kenya, the Philippines, Venezuela, and Madagascar.

In a new study, led by Myriam Raymond and with the collaboration of Antonio A. Casilli and Lucy Neveux, we lift the veil on data work in Egypt. Over 600 questionnaires, 15 focus groups, and an online ethnography reveal the substantial contribution of this country to AI technologies produced and marketed overseas. Egypt’s position in the global AI supply chain is unique, as it serves technology companies both in the Western world (Europe and North America) and in China, often through intermediaries based in the Gulf.

As already observed in other countries, these workers are mostly young: three quarters are below 34 years of age. They live mostly in urban areas. They are also highly educated: in particular, three out of five have an undergraduate degree in science or technical fields. Another notable similarity concerns low pay and lack of protections. We find that four out of five data workers undertake this activity out of financial need, and they spend the income earned in this way immediately on rent, food, and clothes. On average, though, data work pays less than half the country’s monthly minimum wage, and earnings are highly volatile.

The gender gap is more acute than observed elsewhere. Three workers out of four are men. The few female data workers are more dependent on this activity: data work is the only job for two out of ten of them (against one in ten men). Women face unique barriers, reflecting locally-grounded cultural constraints and concerns about online safety.

More generally, culture and morals play an important role in the perceptions that Egyptian workers have of their activity – in ways that had not emerged so forcefully in other countries before. Data tasks sometimes conflict with their principles and beliefs, prompting them to continually question and, at times, reshape their digital identities.

Read the full report here.

Organising AI data workers: barriers, alternatives, and ways forward

The work that fuels AI – from data labelling to content moderation, output checking, red teaming, and so on – is typically outsourced. Digital platforms that operate as online marketplaces play a pivotal role in making this possible. They extend outsourcing to individuals, removing the informational bottlenecks that previously limited it to (multi-person) firms. Platforms treat workers as independent contractors and do not guarantee labour rights. Job insecurity, income volatility, wage theft, and in some cases mental health issues, are common. However, cases of worker mobilisation remain rare. Why, and how can this be changed?

Barriers

A specific challenge that arises on platforms is the asymmetric distribution of work, with a relatively small number of users doing most tasks, and a long tail of minimally-active (or even inactive) people. The reason is that registration is (more or less) open but demand is variable, so that a worker must beat the competition to find tasks to do. This has two main implications. One is that from start, there is an incentive to see other workers as competitors rather than colleagues. The other is that it is difficult to motivate people in the long tail to take action: they are more likely to exit than to voice their grievances.

Lack of a shared worker identity is another crucial gap. Data work was initially portrayed as simple and straightforward, and even sometimes considered as a form of consumption or leisure. Many platforms carefully avoid even using the word ‘worker’, instead preferring terms like ‘Turkers’ (Amazon Mechanical Turk) or ‘Tolokers’ (Toloka). The very fact that workers themselves often take the rhetoric of simple tasks at face value, and struggle to see themselves as such, is indicative of their experience of disrespect, due to widespread misrecognition.

Max Gruber / https://betterimagesofai.org / https://creativecommons.org/licenses/by/4.0/

Juliet Schor writes that “platform earners are not only independent in a legal sense; they also typically do their work independently of other workers”. Technology enables extreme fragmentation of labour and rules out teamwork. Neither do workers ever meet their clients (technology producers), due to platform intermediation. In sum, platform data work isolates workers both from their peers and from other stakeholders (as the above picture cleverly represents). How to organise if you are alone?

Alternatives

In this context, it is useful to broaden our understanding of worker organisation. Beyond collective acts undertaken within an institutionalised framework, we should also embrace informal, unorganised and subtle actions, which can nevertheless lead to positive outcomes.

In crisis-stricken Venezuela, very large numbers of people started data work on online platforms to earn much-needed hard currency. Here, workers have leveraged their personal networks of family and friends to surmount the multiple obstacles posed by platform work. They never created any official organisation, and their actions would rarely qualify as forms of resistance. Some were mere attempts to limit losses in a harshly competitive environment. As researchers, we need to be mindful of cases in which, owing to an unfavourable context, workers prioritise the (short-run) need to counter local scarcity through online earnings, rather than any (longer-run) fight against unfavourable platform management. (More on this case here.)

Ways forward

There are nevertheless signs that successful strategies exist. Kenya is a rare example of organisation: data workers and content moderators in this country initiated actions that attracted international attention and, as in a virtuous circle, support. Of course, not all is easy for them, but Kenya is now a reference, an example for everyone else. This suggests that it is essential to give visibility to workers’ conditions and to any action they undertake to defend their rights.

The other lesson learned from the Kenyan case is that collaborations between multiple stakeholders can achieve a lot in supporting workers and triggering change. Not only established unions, but also researchers, policymakers, and activists in various NGOs (for example, engaged for personal data protection, against discrimination, etc.) can act as multipliers of the resources available to workers.

American AI, made in Venezuela

The political tensions between Venezuela and the United States are at an all-time high, after news of strikes on Caracas and the alleged capture of President Maduro and his wife. The recent escalation follows years of economic sanctions and a deep divide between the two countries.

And yet, Venezuela has massively supplied cheap data work to US-based technology producers throughout all these years. Through digital labour platforms, an educated but impoverished workforce made its way to (the bottom of) the supply chains of US-directed artificial intelligence (AI).

Since about 2017, high inflation, increasing scarcity of even basic goods, and widespread poverty have pushed Venezuelans to work for international platforms that pay in US dollars, albeit at low rates. They have come to constitute a large reservoir for technology producers, mainly (though not only) in the United States. Known for their willingness to accept even the lowest pay rates in the data work market, Venezuelan workers have annotated hundreds of thousands of videos and images for the development of (for example) self-driving vehicles. Ironically, the very policies of Chavist governments – from Chávez himself to Maduro – made this possible. Cheap access to electricity and promotion of digital literacy, including through the widespread distribution of locally produced computers (‘Canaima’) to students and schoolchildren, provided people with the necessary infrastructure to perform data work. Even outdated and malfunctioning, these equipments played a crucial role in enabling widespread Venezuelan participation to the AI pipeline.

Nicolas Gourault (2020). VO: A documentary and sensory investigation about the role of human workers in the training of driverless cars. Source: https://nicolasgourault.fr/films/vo

For Venezuelan workers, platforms labour has constituted a a resilience strategy against adverse local conditions. Participation has never been easy owing to frequent power cuts, slow internet connection, and aging devices, not to mention the difficulty of working almost entirely in English. The high educational levels and computing skills of many workers (including experienced professionals and science/technology students), and embeddedness in densely knit networks of support offered solutions. At the same time, work on platforms is not without challenges, and all Venezuelan data workers have experienced some form of disrespect. Being paid less than peers in neighbouring countries, or even being offered fewer tasks than these foreign peers, are examples of this. At some point, they had to endure a widespread perception that they do not work well. Resisting against international platforms can be more challenging than bypassing local restrictions, and Venezuelan workers limited themselves to occasional acts of minor cheating involving only very few of them.

Venezuelans’ resolve to move out of the crisis and the networked relationships that sustain each of them against hardship have made them massively present, as ‘uninvited protagonists,’ in international data work platforms. Conversely, some AI companies and platforms (from the Global North in general, and from the United States specifically) targeted Venezuela deliberately, not much for the qualities and skills of its highly educated population, but for its low cost at a time of crisis. These platform-mediated encounters enabled short-term solutions, but haven’t raised Venezuela out of poverty, and haven’t ensured a durable provision of high-quality data for AI.

What comes next inside Venezuela is deeply unclear, but unfortunately, nothing (for now) suggests any recognition of the role of these workers in the technology industry, or any opportunity to reshape its outputs in more equitable and respectful ways.

Digital labor in the Middle East and Africa: Emerging trends, challenges, and opportunities

Following the success of the inaugural INDL-MEA Conference in 2024, the second event of the Middle East and Africa chapter of the International Network on Digital Labor (INDL-MEA-2) will take place exclusively online on 25-26 November 2025. The conference will serve as a key regional forum for researchers, policymakers, and practitioners engaged in studying and shaping the future of digital labor, gig work, data work, content moderation, and technology-related jobs in the Middle East and Africa.

Digital labor continues to evolve as a defining feature of global and regional economies, shaping employment opportunities, economic structures, and policy debates. The Middle East and Africa present unique dynamics in digital labor, characterized by platformization, algorithmic management, labor informality, and digital entrepreneurship, alongside issues of regulation, fair work practices, and digital workers’ agency.

With INDL-MEA’s second edition, we aim to enhance interdisciplinary and policy-relevant insights into platform work, automation, labor protections, and digital rights in the region. The programme is available here, and it is still possible to register here.

Where do restaurants come from?

How do digital platforms affect the concrete functioning of markets that pre-existed them? Platforms are intermediaries and it was initially thought that they could solve any mismatches between supply and demand. In the restaurant sector, the hope was that they would seamlessly connect diners with available tables and help restaurants fill their rooms. Yet traditional booking methods remain, and many restaurants restrict the number of seats offered through platforms. A recent study, which I have just co-published with Elise Penalva Icher and Fabien Eloire, examines why.

We borrow Harrison White’s famous producer market model, based on the idea that the key problem of a firm is to position itself in a market that consists of differentiated niches. Restaurants are not homogeneous, and they continuously scan the market to fine-tune their offer – from fine dining to bistro and pizzeria. They evaluate two main indicators: volume, which is relatively straightforward, and quality, which is harder to gauge as it depends on subjective customer perceptions. Platforms break through this limitation by publishing consumers’ reviews and aggregating them into ratings. They provide “digital glasses” that reveal quality alongside volume.

The study investigates dine-in services in Lille, France, in the case of a widely adopted booking and review platform. Methods include participant observation, interviews, web-scraping, and quantitative analysis of business data.

Lille restaurants in Harrison White’s model plot. Note: Horizontal axis: volume, vertical axis: quality. The sub-axes distinguish a non-viable (“Failure”) region from a viable one, in turn subdivided into three different regions (“Ordinary”, “Advanced” and “Paradox”). Zone A = Paradox, zone C = Ordinary, zone D = Advanced, all other zones = Failure. N = 105.

Findings highlight three key effects. First, an amplification effect: platforms enable restaurants to see “like a market,” not just through their own customers but also through competitors’ clients. Second, a normalization effect: platform use pushes firms to standardize their offers, fostering similarity without complete homogenization. Third, a duration effect: sustained platform participation depends on quality positioning, although many restaurants exit after a few years, partly in response to platform dominance. These dynamics suggest a broader rationalization process in which platforms make market observation more systematic and efficient.

This perspective nuances common claims about platforms as market “revolutions.” The study finds no evidence that platforms improve consumer–producer matching. None of the interviewed restaurateurs feared empty tables, and some deliberately withheld capacity from the platform to accommodate walk-ins or phone bookings. Overemphasizing intermediation, earlier research may have overlooked subtler effects. The key function of platforms does not always have to be matching. They can play diverse and even unbalanced roles on a single side of the market, without striving toward a competitive supply-demand equilibrium.

The analysis also reaffirms the validity of White’s model. Originally designed for settings where firms observed only volumes, the model still applies when platforms disclose quality through reviews. Its insights hold across different technological contexts.

Finally, the study underscores the limits of using platforms as sources of research data. We relied on platform data, but we faced gaps: available data are partial because platform objectives differ from research needs, and algorithms remain proprietary. This raises concerns, as platforms exert broad societal influence while controlling critical information.

Overall, the research advances understanding of how platforms affect business practices, in this case restaurants. It contributes to critical scholarship that recognizes the novelty of platform intermediation while tempering claims about its benefits.

The study is available in open access here.

A successful INDL-8 conference in Bologna

When we created ENDL (the European Network on Digital Labour), back in 2017, we booked a room with 17 places. A few days ago, the last conference of the network (which in the meantime has become INDL, replacing ‘European’ with ‘International’) hosted about 200 participants. Internationalisation has not only meant numerical growth, but also inclusion of a diverse range of voices: every year, we see more participants from countries that are often under-represented on the scientific scene, from India and South Africa to Argentina and Brazil. Participants have also diversified in another sense, too: if the majority have always been academics, it is a pleasure to see more and more workers, as well as labour organisers. This year, we could for example benefit from the presence of associations of data workers from Kenya, freelancers from France, and content moderators from Spain.

Participants to the INDL-8 conference, Saint-Cristina cloister, Bologna, IT, 10 September 2025.

A conference like this one is meant to give hope – hope of mutual understanding across countries and cultures, hope of dialogue across disciplines and fields, hope of connections between academic research and action. We worked together to ensure a welcoming environment for all, for instance by encouraging constructive comments, rather than sheer criticism, after each paper presentation. We also strived to keep costs down in order to make the conference free of charge, and with the DiPLab research programme, we could give a few small scholarships to promising presenters who might not have been able to travel otherwise.

Two speakers (M Francesco Sinopoli, Fondazione Di Vittorio, and Ms Kauna Malgwi, Uniglobal) at the plenary panel ‘Plenary panel: New Unionism, towards global alliances’, part of the INDL-8 Conference, DAMA Tecnopolo, Bologna, IT, 11 September 2025

Surely, problems remain. A couple potential participants had visa issues, while others had to cancel due to lack of funding. These problems weigh especially hard on people from emerging and lower-income countries outside Europe and North America. The future is also uncertain, as funding sources become increasingly dryer, and visa restrictions tighter. For this reason, the main INDL-9 conference next year (Geneva, ILO, 9-11 September 2026) will be accompanied by the growth of local chapters. The Middle-East and Africa area is preparing its second conference, this time online only, on 25-26 November. In the US, a one-day event will take place at Yale University on 29 April 2026. Colleagues in Chile and Argentina are launching a series of online events.

Closing keynote (Prof. Sandro Mezzadra, chair: Prof. Marco Marrone), Saint-Cristina Aula Magna, Bologna, IT, 12 September 2025

More information on the INDL-8 conference (including the full programme) is available here.

Women in the loop: the gendered contribution of data workers to AI

I presented today, at the WORK2025 conference in Turku, Finland, a paper on the human-in-the-loop systems that integrate human labor into the production of Artificial Intelligence (AI). Beyond engineers who design models, myriad “data workers” prepare training data, verify outputs, and correct errors. Their role is crucial but undervalued, with low pay and poor working conditions. Shaped by outsourcing and offshoring practices, the market for such services has grown steadily over time, with digital platforms acting as the main intermediaries between AI producers and workers. In their communication with clients, these platforms often emphasize that human workers provide nuanced judgment in complex tasks.

The three main functions of micro-work in the development of data-intensive, machine-learning based AI solutions. Source: https://doi.org/10.1177/2053951720919776

But who are the humans in the loop, and whose contributions count? Here, I focus on women’s participation and its evolution as the market expanded. Data work is theoretically well-suited for women, since it can be performed remotely from home. Besides, platforms generally do not share gender information, thereby limiting direct discrimination. One might thus expect women’s representation to be high. However, the statistical evidence is mixed. Across studies, the proportion of women data workers exceeds 50% only in four cases. Besides, reports sometimes differ for the same country, across platforms or at different moments in time. Looking at the lowest reported shares, then in no country except the US do women represent more than 40% of all data workers. Even in the US, recent data indicate that women constitute about half of the data workforce, down from 57-58% some years ago. Why are women underrepresented, and why does the pattern vary across countries?

Highest proportion of women data workers reported in existing studies (incl. own datasets). Source: author’s elaboration, created with MapChart.

Lowest proportion of women data workers reported in existing studies (incl. own datasets). Source: author’s elaboration, created with MapChart.

The earliest explanation comes from P. Ipeirotis (2010), who analyzed Amazon Mechanical Turk, then the dominant platform. Most workers were from the US and India. In the US, data work paid too little to sustain a household and was often taken up by un- and under-employed women seeking supplementary income. In India, dollar-based pay was more attractive and often a main household income, drawing more men into the activity. Later, as the market expanded, this explanation appeared insufficient: the above maps show that not all rich countries have many female data workers, and some lower-income countries do. Yet, my data show a negative correlation: the larger the share of workers for whom data work is the main income source, the smaller the proportion of women. Ipeirotis’s hypothesis still holds but requires updating to today’s more competitive and globalized platform economy.

Proportion of workers for whom data work is the main source of income vs. proportion of women, by country. Source: own survey data (from projects TRIA and ENCORE, 2020-24).

Platforms fragment work into tasks and assign them to individuals framed as independent contractors competing for access. Unlike traditional firms, workers do not collaborate but face intense competition. Outcomes vary by national context. In countries facing stagnation or crisis, such as Venezuela, international platforms offer a rare source of income for highly qualified workers. Competition becomes fierce, and “elite” workers – often young men with STEM backgrounds – dominate. Women are disadvantaged, either due to fewer technical qualifications or because care responsibilities limit their ability to invest in building strong platform profiles and reputations. By contrast, in more dynamic economies such as Brazil, local job markets absorb highly skilled professionals, leaving platform work to more disadvantaged groups. Here, women with family duties are more visible. Thus, platform demographics reflect national conditions: in poorer or crisis-stricken countries, men from the educational elite seek career advancement, while in richer countries, women (especially mothers) take on such work primarily to supplement household income. Women may be equally educated, but they often lack the time to cultivate advanced STEM skills. As platforms demand longer and more specialized tasks, men increasingly gain the upper hand, crowding women out—even in countries where they were once the majority.

Platform design ignores these dynamics. Workers are treated as abstract entities, stripped of the socio-economic and cultural contexts that shape real inequalities. Competition, combined with local conditions, deepens gender gaps. Interventions must therefore consider gender disparities. Otherwise, they risk reinforcing inequalities. Supporting women’s access to data work—particularly those constrained by family responsibilities—can contribute to more balanced labor participation and ensure that AI benefits from a broader diversity of human input.

AI, labour and natural resources in Santiago

Last week in Santiago, Chile, I had the tremendous opportunity to give a keynote speech at the 4th annual workshop of the Millennium Nucleus on the Evolution of Work (M-NEW), of which I am also a senior international member. This interdisciplinary workshop brought together labour scholars from various parts of Latin America and beyond. I really liked the inspiring talks and the friendly and stimulating interactions with colleagues.

My own talk drew on my multi-year research programme on the crucial yet invisible human labor behind the global production of artificial intelligence. I first examined the evolution of this form of work over the last two decades, demonstrating that while its core functions in the development of smart systems have remained consistent, the scope and volume of such tasks have expanded significantly. I then analyzed the organization of this labour at the intersection of three trends in recent globalization: outsourcing, offshoring, and digitalization. These dynamics account for the marginalization of these workers within the tech industry and the relocation of their labor to lower-wage countries. Based on these insights, I described four cases—Venezuela, Argentina, Brazil, and Chile—highlighting the diverse effects of local conditions. I concluded by identifying emerging scientific and policy challenges, particularly concerning the recognition of skills, and the place of the informal economy.

The following week, still in Santiago, I was excited to participate in the kick-off meeting of the new research project SEED (“Social and Environmental Effects of Data connectivity: Hybrid ecologies of transoceanic cables and data centers in Chile and France”), a collaboration between my research group DiPLab and another Millennium Nucleus, FAIR (“Futures of Artificial Intelligence Research”). SEED received joint funding from the ECOS-SUD programme (France) and ANID (Chile) to analyse the AI value chain, from its production and development to its impact on employment, use and environmental consequences, by studying the Valparaíso-Santiago de Chile and Marseille-Paris axes.

My presentation introduced the concept of the ‘dual footprint’ as a heuristic device to capture the commonalities and interdependencies between the different impacts of AI on the natural and social surroundings that supply resources for its production and use. I framed the AI industry as a value chain that spans national boundaries and perpetuates inherited global inequalities. The countries that drive AI development generate a massive demand for inputs and trigger social costs that, through the value chain, largely fall on more peripheral actors. The arrangements in place distribute the costs and benefits of AI unequally, resulting in unsustainable practices and preventing the upward mobility of more disadvantaged countries. The dual footprint grasps how the environmental and social dimensions of AI emanate from similar underlying socio-economic processes and geographical trajectories.

Desinformación, trabajo y plataformas en Buenos Aires

Acabo de regresar de un viaje muy lindo a Argentina, donde fui invitada por el Instituto francés para participar en varios eventos.

El 12 de mayo, participé en la conferencia “Manipulación Informativa e Injerencia Extranjera: Desafíos Globales y Respuestas Democráticas”, organizada por la Delegación de la Unión Europea en Argentina y por varias embajadas (como la de Francia). En el panel “Cómo contrarrestar la desinformación respetando la libertad de expresión y el derecho a la información”, hablé de cómo la desinformación se financia a través del mercado publicitario que sustenta toda la web, y de la necesidad de regular este mercado. También destaqué la importancia de fortalecer la investigación científica sobre este tema – como en el proyecto AI4TRUST, financiado por la propia UE.

Al día siguiente, tuve dos encuentros con estudiantes de periodismo, en la Universidad Nacional de Avellaneda y en la Universidad Abierta Interamericana, también sobre temas relacionados con la desinformación en internet. Fue un honor y un placer ver, en cada caso, la sala llena y mucho interés. Los/as estudiantes hicieron muchas preguntas y demostraron una gran disposición a aprender y progresar.

Los días 16 y 17 de mayo participé en tres paneles organizados en el marco de la Noche de las Ideas, una iniciativa del Instituto francés que se organiza todos los años y esta vez tuvo lugar en el famoso Teatro Colón de Buenos Aires. Fui en la sesión inaugural sobre el tema de este año, “El poder de actuar”. Además, participé como ponente en un debate muy interesante sobre el trabajo en plataformas digitales que se titulaba provocativamente ¿Nuevas servidumbres voluntarias? Jóvenes y precariedad y en otro sobre la inteligencia artificial: “In A.I. we trust?”. Actuar con y en contra de las nuevas tecnologías.

El 20 de mayo, di una charla sobre “El futuro del trabajo y la IA” como parte del ciclo UBA Digital, en la Universidad de Buenos Aires. Presenté algunos resultados de mi investigación sobre el trabajo digital y su papel en la producción de IA, desarrollada en el marco del programa de investigación DiPLab. Una vez más, me alegró ver a muchos participantes con preguntas muy interesantes. Fuimos acogidos por la facultad de odontología y también tuvimos la extraordinaria oportunidad de visitar la clínica.

Uninvited protagonists: the social networks of Venezuelan AI data workers

After years of work, the long-awaited good news: my article ‘Uninvited Protagonists: The Networked Agency of Venezuelan Platform Data Workers‘, co-authored with Juana Torres-Cierpe, has just been published in New Technology, Work and Employment!

Workers in Venezuela are powering AI production, often under tough conditions. Sanctions and a deep political-economic crisis have pushed them to work for platforms that pay in US dollars, albeit at low rates. They constitute a large reservoir for technology producers from rich countries. But they are not passive players.

They build resilience, rework their environment, and sometimes engage in acts of resistance, with support from different segments of their personal networks. From strong local ties to loose online connections, these informal webs help them cope, adapt, and occasionally push back. Their diversified relationships comprise an unofficial and often hidden, albeit largely digitised relational infrastructure that sustains their work and shapes collective action.

These findings invite to rethink agency as embedded in workers’ personal networks. To respond to adversities, one must liaise with equally affected peers, with family and friends who offer support, etc. Social ties ultimately determine who is enabled to respond, and who is not; whether any benefits and costs are shared, and with whom; whether any solution will be conflictual or peaceful. Social networks are not accessory but constitute the very channel through which Venezuelan data workers cope with hardship.

Not all relationships play the same role, though. Venezuelans discover online data work through their strong ties with family, close friends, and neighbours. To convert their online earnings into local currency, they rely on their broader social networks of relatives and friends living abroad and indirect relationships with intermediaries. For managing their day-to-day activities, Venezuelans expand their social networks through online services like Facebook, WhatsApp, and Telegram, connecting with diverse and less-close peers within and outside the country. Different social ties affect the various stages of the data working experience.

Overall, no Venezuelan could work alone – and the networked interactions that sustain each of them against hardship have made them massively present, as ‘uninvited protagonists,’ in international platforms. Their massive presence in the planetary data-tasking market is a supply rather than demand-driven phenomenon.

This analysis also sheds light on the reasons why mobilisation is uncommon among platform data workers. Other studies noted diverging orientations of workers, unclear goals, lack of focus, and insufficient leadership. Another powerful reason hinges upon the predominance of weak ties in building up online group membership: indeed, distant acquaintances are insufficient to prompt people to action if their intrinsic motivations are low.

The article is available in open access here.