Source-linked AI summary
Digital Labor Platforms as Infrastructure of Extractivism
Julian Posada
TL;DR
The chapter addresses how AI depends on data and labour whose production and maintenance are shaped by social relations. It proposes platform extractivism and draws on mixed-methods research with data workers in Venezuela and Colombia. It concludes that platforms organize data and labour appropriation, concentrate value in core regions, and reproduce epistemic and territorial hierarchies.
Problem
Machine-learning AI is commonly attributed to computational power and large datasets, while the extensive human labour and social relations behind data production and maintenance also underpin its development.
Source: §1 · p. 1Method
The chapter develops platform extractivism from digital labour studies and Latin American critical thought, alongside mixed-methods research with data workers in Venezuela and Colombia.
Source: §1 · p. 2Results
The chapter finds that platforms organize corporations’ appropriation of data and labour, concentrate value in core regions such as Silicon Valley, and amplify some knowledges while marginalizing others.
Source: §5 · pp. 7–8Takeaways & Limitations
Platform extractivism offers a framework for examining core–periphery value exchange, labour conditions, knowledge and ideology, and the social and non-social relations shaping production.
Source: §5 · p. 8
Abstract
from arXiv · showhide
This chapter examines how digital platforms, as infrastructures of extractivism in contemporary artificial intelligence, enable corporations to appropriate data and labour while concentrating value in core regions such as Silicon Valley. It proposes platform extractivism as a framework linking commodification, epistemic dominance, and reproduction through territory. Drawing on digital labour studies, Latin American critical thought, and mixed-methods research with Venezuelan and Colombian data workers, the chapter discusses how AI depends on precarious platform labour to produce, clean, annotate, and maintain data. It further shows how annotation systems normalize culturally specific categories, especially US norms, as universal knowledge, reproducing colonial hierarchies. Finally, it highlights how data work is sustained by unpaid reproductive labour, community networks, and material conditions that make workers available for platform labour. Platform extractivism thus reveals AI as a labour-intensive and territorially embedded political-economic system under capitalism that perpetuates legacies of coloniality.
1. INTRODUCTION
The chapter argues that machine-learning AI depends on extensive human labour and the social relations through which data are produced and maintained. It proposes platform extractivism as a framework linking commodification, epistemic power, and territorial reproduction.
- Machine-learning AI depends on extensive human labour, not only computational power and large datasets, to produce, collect, and maintain data.Source: §1 · p. 1
- Platform extractivism situates platforms within historical core–periphery relations and connects data extraction, data production, labour exploitation, and power concentration.Source: §1 · p. 2
- The proposed framework links data and labour commodification, imposed forms of knowledge, and extractive dynamics sustained and spatially organized through territory.Source: §1 · p. 2
- The chapter draws on digital labour studies, Latin American critical thought, and mixed-methods research with data workers in Venezuela and Colombia.The research combined interviews, a survey of more than 600 workers, platform-interface traffic analysis, and analysis of annotation instructions.Source: §1 · p. 2
2. EXTRACTIVISM AND LABOUR COMMODIFICATION
The chapter treats extractivism as a system that organizes and sustains appropriation, rather than as extraction alone. It connects platform-mediated data work to core–periphery relations, labour, and the infrastructure that moves data and value.
- Venezuelan economic crisis following declining oil prices and other factors led many people to seek online work paid in more stable US dollars.Source: §2 · p. 3
- Data is not a commodity in the strict sense, yet corporations routinely acquire, exchange, and monetize it through markets organized by platforms.Source: §2 · pp. 3–4
- Platforms organize data extraction and connect firms to an expanded, deregulated labour force whose work produces, annotates, and refines machine-learning data.Source: §2 · p. 4
- Framing platforms as extractivist infrastructure extends analysis to labour conditions and epistemological and territorial dimensions beyond individual platforms.Source: §2 · p. 4
3. EPISTEMIC DOMINANCE
The chapter argues that data extraction also circulates and imposes ideas, normalizing some forms of knowledge while erasing others. Its analysis shows how annotation practices reproduce culturally specific norms as universal categories.
- Extractivism entails imposing epistemic orders alongside material appropriation, channeling ideas toward centres of power and normalizing some knowledge while erasing other forms.Source: §3 · p. 4
- Annotation guidelines across three platforms and one Latin American outsourcing firm encouraged workers to encode US-derived cultural norms as dataset “ground truth.”The examples included object-naming conventions and social classifications such as racial categories; workers faced surveillance, discipline, and possible account deactivation for deviating.Source: §3 · p. 5
- Platform extractivism extends power asymmetries into cognition, knowledge production, and meaning-making, reproducing colonial legacies while rendering workers and contributions largely invisible.Source: §3 · p. 5
- The promise of techno-libertarian freedom is uneven: it sustains commodity extraction and epistemic imposition rather than universal emancipation.Source: §3 · p. 5
- The chapter connects labour and production to knowledge circulation across territories, arguing that platform-mediated effects reach beyond production sites into territories and epistemic orders.Source: §3 · p. 5
4. SOCIAL REPRODUCTION AND TERRITORY
The chapter shows that platform work is sustained by unpaid care and domestic labour, community networks, and material conditions. It uses territory to frame workers as embedded in relations that make both life and production possible.
- Unpaid domestic and care work by family members sustained Venezuelan data workers’ availability for paid platform work; all interviewed or surveyed workers lived with family.Source: §4 · p. 6
- The outsourcing of data production depends on unpaid reproductive labour that keeps workers continuously available, making this labour especially inexpensive.Source: §4 · p. 6
- Amid crisis and reduced institutional support, workers relied on neighbours and local networks to access food and water, making environmental resources visible as conditions of labour.Source: §4 · p. 6
- Territory frames production as dependent on relations among workers and their social and material surroundings, including land and water.Source: §4 · pp. 6–7
- When value extraction is privileged over life’s reproduction, territories are degraded; data-extractive costs are reported on workers’ bodies, physically and psychologically.Source: §4 · p. 7
5. CONCLUSION
The chapter frames AI as an extractivist system that appropriates data and labour while concentrating value in core regions. It proposes platform extractivism to connect these dynamics to knowledge hierarchies, working conditions, and the social and territorial relations that sustain workers and production.
- Those who provide data and labour for contemporary AI are rarely compensated in proportion to their contributions.Source: §5 · p. 7
- Data production, cleaning, labelling, and maintenance are foundational forms of labour without which machine learning systems could not function.Source: §5 · p. 7
- Platform-mediated data work makes raw information legible and valuable to computing firms, while concentrating extracted value in core regions such as Silicon Valley.The chapter says datafied knowledge and skill are drawn from data workers, users, artists, and writers.Source: §5 · pp. 7–8
- Platform extractivism links AI development to knowledge hierarchies, as platform decisions amplify some knowledges and marginalize or render others invisible.Source: §5 · p. 8
- Data work depends on workers embedded in social and material ecologies, including families, colleagues, neighbours, and other human and non-human actors.Source: §5 · p. 8
- The framework connects core–periphery value exchange and platform working conditions with knowledge, ideology, meaning, and actors beyond the individual worker.Source: §5 · p. 8