Who Measures the Value of Work? Artificial Intelligence and the New Transnational Job Precarity
By Nahima Dávalos-Vázquez
“Let’s say I’m a number of tasks per hour”
Ricardo, data labeling, 26 years old, Mexico
On the margins of globalized labor, the internet connects, but it also isolates. Over the last decade, the expansion of outsourcing companies, commonly known as digital labor platforms or talent clouds, has transformed how work is distributed and managed in the Global South. Platforms like Appen, Accenture, Remotasks or Teleperformance rely on thousands of remote workers to perform the invisible work of the internet: data annotation, chatbot training, content moderation, image labeling, process improvement, and customer support. These are the tasks that make AI systems appear “intelligent”; yet the workers behind them remain mostly absent from mainstream debates on digital innovation. Their performance is also often monitored through productivity software like SuccessFactors, Hubstaff or Time Doctor, which quantify their activity through constant metrics: clicks per minute, response time, accuracy, and availability.
This article focuses on the everyday reality of this type of work through an ethnography that gathers the experiences of three remote workers located in Mexico. All of them work for transnational companies, performing essential tasks for the maintenance of digital infrastructures in the Global North. These workers operate under flexible arrangements that, while often presented as opportunities for autonomy, in practice generate isolation, lack of protection, and a deep sense of dispensability. They do not know the final client and interact mainly with automated systems that monitor their speed, accuracy, and availability.
The experiences of these workers reflect broader power structures resulting from an inflexible logic of productivity. These workers do not just click, they participate in a regime that cancels their judgment, while also extracting their interpretive labor, such as when they must label images as offensive content without knowing the context or being able to discuss the criteria. This resonates with what Milagros Miceli, Schuessler and Yang (2020) call a vertical imposition of meaning, in which workers must internalize tasks with logics based on values and goals that are far removed from their own realities. A practice that strips them of interpretive agency and, at the same time, contributes to the consolidation of a global cognitive infrastructure that naturalizes these values as universal.
Martin Krzywdzinski, Daniel Schneiß and Sperling (2025) go deeper into this issue by highlighting that algorithmic management is not only about control, but also about displacing responsibility, delegating decisions to systems that hide who is accountable for what. In my own experiences, this displacement felt almost intentional, as workers were blamed for mistakes, but no one could explain the rules to them. Meanwhile, Salvador Regilme (2024) offers a more precise framework by suggesting that what we are witnessing is a new or renewed form of colonialism, one that not only extracts minerals and land to sustain the digital industry, but also extracts attention, cognition, and hours of life. These workers in Mexico may be connected to global systems, but they are radically disconnected from rights, context, and recognition. Through their stories, this article asks not only what this tells us about labor today, but also seeks to contribute to understanding how we are allowing systems to reorganize the limits of human dignity.
Does an Algorithm have Agency?
If we answer from an outdated perspective that sees technology as an inert material, it is tempting to say no, especially because compared to a living being, it does not feel or decide and, therefore, has no intention. Nevertheless, if this system actively determines the weekly productivity score of a worker, if it defines their incentives, if it can and might take away their job, then its agency becomes more concrete and operative. Digital Anthropology, and its widely praised reflexive flexibility allows us to ask about this kind of agency without falling into the trap of anthropomorphizing machines and systems, but also without accepting blindly that technologies are “neutral.”
At this moment, it is urgent to stress that technologies are not neutral, nor are the social relations that design and finance them, and those that encourage growing inequality. The history of technology has also been a history of its use for exclusion (Ricaurte, 2024). We can see it from surveillance systems to facial recognition algorithms that AI clearly amplifies inequalities. What is more worrying, however, is that these innovations are designed with this particular purpose. Critical research emerging alongside the accelerated development of AI is revealing various problems and violations of basic human rights, something that is exacerbated for those who are marginalized by these developments.
The re-vulnerabilization experienced by workers in the Global South is a tangible example of this. These invisible bodies sustain the AI infrastructure: they produce the data, train the models, or moderate the content. In this context, their re-vulnerabilization is not a side effect, but a structural condition of the model. People with exhausting workloads shape a renewed notion of productivity where pause and rest are becoming a privilege.
In the case of remote workers from this region, AI not only decides who gets a job, but it also optimizes the dynamics of precarization. Microtasking has become a symbol shaping this labor market. Small tasks, fragmented, monitored every second. Under this model, workers stop being subjects and become extensions of the machine. In these environments, their creativity and experience are only valuable if they are used to finish faster and improve the assignments. Time is the guideline for platforms like SuccessFactors, Hubstaff o Time Doctor. These concentrate power in the Global North, reproducing new forms of digital colonialism (Couldry & Mejias, 2019), where labor is not extracted from territories, but from connected and remotely monitored bodies.
Bodies Turned Data
In this new labor cartography, bodies disappear in productivity indexes, but their fatigue remains. I met and accompanied the faces, bodies, and exhaustion of three remote workers during fieldwork for a digital and sensory ethnography I conducted between 2023 and 2024. I followed their everyday life, registered their bodily and emotional states, analyzed their productivity interfaces, and the ways they inhabit, come to terms with, or resist these environments.
It is a misconception to believe that the body’s place in remote working is only in its physical absence. In these environments, the body is not erased; it overflows. As Alejandra Gabrielidis (2020) proposes, in our digital interactions, the body does not disappear; it unfolds and transforms into a numeric extension of itself. Thanks to the breadcrumbs of data we leave in every interaction, this data-body grows and deepens its form. A process that, paradoxically, happens unseen. What is captured is no longer the organic density of the body, the somatic body, but the traces it leaves in response speed, clicks per minute, inactivity time, or cursor position. Fragments of presence, let us say.
In this new corporality, not recognizing the value or the burden of these data bodies leads us to the inhuman problem of a somatic body that must hold an infinite load of stimuli it cannot handle, designed for a 24/7 rhythm. This somatic body cannot keep up and signals this through insomnia, gastritis, panic attacks, or unclear pain. Gilbert Simondon (1989) mentions that the relationship between humans and technology is always a process of co-individuation, that is, we change with what we use, and at the same time, what we use changes with us. However, in the environments I observed, this reciprocity seems broken and more like a deep asymmetry. At first sight, instead of symbiosis, there is only space for subordination. Companies do not expect the worker to shape the tool, but that the tool, aligned to their profit plans dictates the rhythm. Making somatic bodies invisible and not giving value or recognition to the data body is not just a practical strategy; it is also a political gesture. Making them invisible is more profitable than taking care of them.
Carolina was a junior consultant in change management, living in Mexico City. She worked for a consultancy that offered services to international companies with projects in Latin America. Her workdays were complicated because she had to adjust to European schedules, even if it was not stated in her contract. This often implied very early meetings that added to the workload, which extended past midnight in Mexico. Every ten minutes, a software that monitored her activity could capture her screen. Her company considered this as a tool “to better organize productivity”, but in practice, it gave her constant anxiety. Her role was to manage the implementation of tech resources for different operational teams, although, ironically, that did not happen for her.

Ricardo lived in Querétaro, Mexico. When I met him, he was a data entry and labeling worker for a U.S.-based AI project for six months. At first, he thought it would be simple: classify images. But he soon realized the system not only scored the amount of labels he added, but also how long he took, how many clicks he made, and if he left the mouse still for too long. “It was supposed to be flexible, but they monitor me all the time.”
Laura was a Virtual Assistant. She lived in Oaxaca and worked for a retail company. Her job was created during the boom of virtual assistant chatbots, her company, trying to stay ahead of the innovation and to generate and collect the data needed to train the chatbot, hired her and others. Her interaction was only through messaging, and she had a limit of messages and topics to solve. Those on top of this limit were redirected to a call center, although the goal was to solve most issues with the minimum number of responses. When she told me about her work, she described it like this: “My job is virtual assistance as if I were a robot, a chatbot, the idea is that the client feels like they are being helped by a program and feels better, or well, helped more quickly.” When something did not work, part of her job was to report the incident so the design and programming team could fix it.
She knew very well that her role was training her future replacement, but what stressed her most was the need for her uninterrupted attention. In a conversation, jokingly, she told me: “A chatbot does not take breaks.” It may sound paradoxical or even dystopian, but her simulating a future assistant technology came with vital sacrifices, like having to always be near her keyboard and screen to answer. Laura is very skilled — she usually had to cook or eat while staying alert to messages. That meant she constantly sacrificed moments of dignity in her daily life. On top of that, she lived with the uncertainty of having to look for a new job soon.
I believe the problem is not only that the body becomes data, but that data becomes the norm. In that transition of embodied metrics (body to data), everything else is lost: the tremble, the pause, the doubt, the need. Thus, the margin for the body to say “enough” is amplified. It feeds the idea that the most productive body is the one that is least felt. But bodies feel. Even if they are erased from the interface, they become present through insomnia or unresolved hunger. As I mentioned, the body does not disappear, it overflows.
The Digital Maquila
It is not the first time the Global South has been tasked with sustaining the North through invisible labor. European and North American companies outsource digital services through outsourcing platforms, under the banner that remote work is a sign of flexibility and modernity. However, in the fieldwork, I saw another side: unstable contracts, low wages, and especially the total absence of labor representation. Just like in the 1990s (and still today), when physical maquilas were full of women sewing for international brands, today’s digital maquilas once again feed on the time, energy, and attention of workers who no longer need to cross borders, yet their production continues without the right to demand rights.
The difference is that this exploitation now happens in the intimacy of the home: in Ricardo’s bedroom where he set up his desk, during Laura’s lunch hour, or in Carolina’s early mornings ‒ silently, in front of a screen. This flexibility also brings loneliness. As Carolina noted, “At the office you see your coworkers, you go out to eat, you run into them in the hallway ‒ here, you never meet anyone.” While virtuality offers many benefits, in exploitative environments, it becomes a strategy to eliminate community, organization, and the safety needed to demand change. Add to that the already eroded existence of labor unions. If work is individual, then complaints are too ‒ and networks of solidarity become nearly impossible.
The digital maquila, like its physical predecessor, continues to impose exhausting rhythms, but under different logics. Now, piecework is embedded in deliverables: tagged items, reviewed documents, improvement changes, answered messages, and closed tickets. However, what distinguishes it is that this productivity is chronometrically monitored and recorded. Conversations repeatedly emerged around this new order and language of time and immediacy: who delivered at what minute, who reviewed minutes later, who delayed, who received a delay warning. Every activity fragment is measured in metrics that generate new environments and meanings to make sense of productivity, time, and energy ‒ even outside of work tasks. Ricardo told me: “You spend so much time here (working) that you start saying stuff like ‘Why aren’t they replying quickly on WhatsApp?’ Or ‘Why is the Didi (delivery guy) taking so long, it’s not that far.’”
In this context, what emotions and bodily sensations emerge when your workday becomes a continuous scan? It was a guiding question for this analysis. In the answers, psychosocial issues came first. According to data from Mexico’s Labor Observatory (2024), more than 60% of those working in outsourced digital jobs report high levels of stress linked to algorithmic surveillance.
During the fieldwork, Carolina, Ricardo, and Laura kept sensory journals. It was a cathartic experience for all. Seeing the management of their time, energy, and sensations during their workday through a log also served as an emotional release of complaints and fatigue. This daily record helped them recognize how invisible their labor struggles had become. One might think they had normalized their exhaustion or powerlessness in the face of job insecurity. However, the story is more complex: a web of needs and deprivations lived as ongoing crises.
Carolina: “Today I worked with a headache for the third time this week.” Ricardo: “Yeah, it’s common that when you’re in the last hour, your hand goes numb. You have to massage it or change positions constantly, or you just can’t finish. […] My back and neck… sometimes I don’t even notice until I get up.” Laura: “Since I’m sitting all the time, after a while my feet get super cold, so I always use my blanket or wrap myself up.”
On top of that, the environment has its conditions. Carolina, Ricardo, and Laura all wish they could have a proper home office ‒ some space to mark off that part of their day. But they live in small, shared apartments with other dynamics that make that separation impossible. Therefore, they must bend their everyday life toward labor. Laura has a table that serves as dining table, desk, and at one end a supply organizer for the three people living there. Ricardo received an office chair after three months at his job, but had to spend his savings on a small desk, which now sits in the corner of his bedroom. He could not use the living room, as it is too small and shared with two other housemates. Carolina lives with a roommate who spends most of the week at her boyfriend’s place, so she is usually alone. That may seem like a quiet space for work, but it has also meant more social isolation and less daily human contact.
All of this happens in solitude. Even if housemates, partners, family, or friends know some of these struggles, ultimately there are no coworkers with whom to discuss shared challenges. One of the most relevant and urgent findings of this ethnography is the tight relation between the increased digitalization of labor (like AI-driven productivity management) and the rise of psychosocial disorders like burnout. For Carolina, and for her friends and peers with similar remote jobs, the use of antidepressants has become a fast, effective way to deal with work overload, the social isolation that their jobs entail, and the fear of losing employment. Add to that alcoholism or physical and emotional crises, which, despite their weight, are faced daily oscillating between resignation, helplessness, and a constant struggle to find small cracks of change.
In Ricardo and Laura’s daily realities, access to specialized psychological or medical care is nonexistent. Some of their main work tools are cheap medications like paracetamol and antacids. They know their palliative practices will only worsen future health outcomes. Ricardo has already experienced tendinitis more than once. Still, they know they have to work. On top of this, food and the whole structure of self-care is subject to the demand for efficiency. You eat to produce more: fast-filling carbs, coffee to wake up, energy drinks to stay focused but these bring more anxiety and imbalance. In these scenarios, self-care is not neglected out of ignorance; it is neglected because there is no time for it.
Remote work’s flexibility is both a promise and a trap. Anxiety appears almost daily in their sensory diaries. In these environments, the debate over the 8-hour workday has shifted toward the anxiety and exhaustion caused by self-managing micro-tasks. In addition, between micro-tasks, micro-deliverables, and micro-productivity, the lifetime is fragmented to structure the day. Time management, seemingly autonomous and flexible, becomes captured by a need-desire to perform and produce at the cost of dignity. You eat later, you rest later, and you deliver quickly. When we reflected on their true work needs, other experiences and desires surfaced. Ricardo wishes he could take more breaks and that his work was actually flexible. Carolina would like to be paid more for what she does, while Laura would like to live somewhere with space for a real office, to have time to rest.
From a sensory and embodied perspective, workers experience this constant scanning, on the one hand, as physical pressure that accumulates in body parts: tense shoulders, numb fingers and wrists, swollen feet, rigid backs, tired and irritated eyes. On the other hand, they experience it as emotional pressure that fluctuates as they try to make sense of their demands shifting from overwhelm to anger, or from exhaustion to fear. The body is not just a work interface; it is the territory where this violence is inscribed.
Looking at Each Other Again and Rebuilding Community
What could be the solution? There is no single answer but there are proposals. Thinking about ethical artificial intelligence is not only about limiting its excesses, but also about imagining systems designed from the real experiences of those who use them and fuel their operation. When I began to explore the topic of growing burnout, I wondered: What if monitoring softwares were designed and trained not only to detect when a worker’s productivity drops, but also when they show signs of fatigue? There are vital limits to the amount of stimuli and responses a human body can endure.
Sometimes we look for answers within the very problem. Increasing inequalities and the lack of rights do call for innovation as part of the solution but also for new approaches to guide it. In the digitization of work environments, it is crucial that workers themselves participate in defining the metrics that regulate their labor, so we can create contexts where AI not only measures but also cares. Remote workers are real bodies, real time, real emotions, and real needs. In the face of expanding digital outsourcing and algorithmic surveillance, it is urgent to rethink: What does it mean to work, to inhabit, and to resist in this new transregional scenario?
The work I witnessed in the stories of Carolina, Ricardo, and Laura is not an accident of innovation; it is a snapshot of a development that moves forward disguised as efficiency, promising false flexibility, and especially, dissolving human bonds into metrics and screens. While transcribing their words, I kept thinking about how we have reached a point where people feel they need to ask for permission to breathe, as Laura described it to me. A context in which taking a break, a vital necessity, is increasingly becoming a privilege.

Yet, they resist. With small tricks that reveal the powerful drive for change within these seemingly immovable environments. Like placing a spoon over the mouse to simulate activity and sneak a moment to stretch, or scheduling a video call with oneself to eat away from the desk without being interrupted. The counterpart to micro-tasks are micro-strategies not enough to save the day, but enough to make it more bearable. Even without knowing each other, something these three individuals shared in their narratives was the consolation and motivation of saying, “This won’t last forever.” The desire for something different and more dignified is always there.
This text does not aim to offer technical solutions, but rather to contribute to an open and uncomfortable question: Who has the right to be a body in this new era of work? And what happens when that tired body demands a place in the world beyond performance? Perhaps the most political act today is to look at each other again. To build community where everything pushes us not to do so. To return names and faces to those who keep these invisible systems running and to remember that behind the chatbot, behind the labels, behind the screen that never turns off, there is someone breathing.
References
Couldry, N., & Mejias, U. A. (2019). The Costs of Connection: How Data Is Colonizing Human Life and Appropriating It for Capitalism. Stanford University Press.
Krzywdzinski, M., Schneiß, D., & Sperling, A. (2024). Between Control and Participation: The Politics of Algorithmic Management. New Technology, Work and Employment, 40(1), 60–80. https://doi.org/10.1111/ntwe.12293
López-Gabrielidis, M. A. (2020). Datificación e individuación: Estudio sobre la corporalidad digital en prácticas artísticas contemporáneas [Doctoral dissertation, Universitat de Barcelona]. https://diposit.ub.edu/dspace/bitstream/2445/170989/1/ALG_TESIS.pdf
Miceli, M., Schuessler, M., & Yang, T. (2020). Between Subjectivity and Imposition: Power Dynamics in Data Annotation for Computer Vision. Proceedings of the ACM on Human-Computer Interaction, 4(CSCW2), Article 115, 1–25. /doi.org/10.1145/3415186
Regilme, S. S. F. (2024). Artificial Intelligence Colonialism: Environmental Damage, Labor Exploitation, and Human Rights Crises in the Global South. SAIS Review of International Affairs, 44(2), 75–92. DOI: 10.1353/sais.2024.a950958
Ricaurte, P. (2024). The Digitalization Imperative. Dialogues on Digital Society, 1(1), 12–17. https://doi.org/10.1177/29768640241262263 (Original work published 2025).
Simondon, G. (1989). L’individuation psychique et collective. Paris: Aubier.
About the Author
Nahima Dávalos-Vázquez studied a PhD in Digital Anthropology at Universidad Autónoma Metropolitana (UAM), Mexico, where her research explored how young mestizo and Indigenous women relate to and appropriate technology. This article stems from a research project developed during her residency at the Weizenbaum Institute for the Networked Society in Berlin. Her current work focuses on the imaginaries and everyday uses of technology in the Global South, the role of young women in digital environments, and the labor realities of platform workers across Latin America. She is also a consultant at The Growing Lab, where she contributes to the creation of situated and sustainable methodologies to support the wellbeing of digital workers. Her work is particularly oriented toward building bridges between critical academic research and applied models that respond to local realities.
Citation: Nahima Dávalos-Vázquez, Who Measures the Value of Work? Artificial Intelligence and the New Transnational Job Precarity, in: TRAFO – Blog for Transregional Research, 10.07.2025, https://trafo.hypotheses.org/60094
OpenEdition suggests that you cite this post as follows:
Forum Transregionale Studien (July 10, 2025). Who Measures the Value of Work? Artificial Intelligence and the New Transnational Job Precarity. TRAFO – Blog for Transregional Research. Retrieved February 12, 2026 from https://doi.org/10.58079/14b7v


