Our Methodology
What is data work?
Data work is the human labor involved in producing, annotating and preparing data for machine learning and artificial intelligence systems (Miceli and Posada 2022). The tasks performed by data workers are fundamentally about making sense of data to make it categorizable, sortable, and interpretable for computers, constituting a structurally necessary step in AI development and maintenance.
Data work encompasses a wide range of activities that flow into the different stages of AI production from the collection and annotation of data, to the moderation and verification of outputs.
Concrete examples of data production and collection include transcription, generating descriptions of images, or producing new data (text, audio, images) according to specific instructions. Data annotation or labeling aims to enrich data with meaning, by identifying categories, topics, emotions or tone. Another form of annotation is content moderation, which classifies user-generated materials as permissible or not according to the policies of social media platforms and generates materials to train hate speech detection models (Abdelkadir et al. 2025, Gillespie 2020). Quality checks and verification of outputs provide new data to optimize or re-train an existing algorithm and AI impersonation, when workers directly mimic model behavior, supplements AI systems when they fail and provides further training data (Tubaro et al. 2020).
What is Workers' Inquiry as a Research Methodology (WIRM)?
In this project, the data workers serve as community researchers. Every community researcher works for two to four months on their inquiry and is compensated for all their working hours. We collaborate with data workers globally. A decisive sampling criterion is that these are data workers who are already organized in workers councils, unions, communities, or advocacy organizations.
Workers’ Inquiry as Research Methodology (WIRM) is the methodological approach guiding our collaborative work. It builds on the tradition of community-based participatory research (Stoeker 2013; Jason & Glenwick, 2016) and on the assertion at the core of Marx’s Workers’ Inquiry , that workers themselves are the epistemic authority concerning historically specific manifestations of capitalist exploitation and the necessary agents of social change (Marx, 1880).
Each inquiry is the result of a collaboration with workers, who guide the direction of the research, such that it is oriented towards their needs and goals of building workplace power.
What we did?
In this project, the data workers serve as community researchers. Each community researcher works for two to four months on their inquiry and is compensated for all their working hours. We collaborate with data workers globally. A decisive sampling criterion is that these are data workers who are already organized in workers councils, unions, communities, or advocacy organizations.
To guide this ambitious endeavor, we developed the Workers’ Inquiry as Research Methodology (WIRM) framework.
Building and maintaining trust with the community researchers is an ongoing task as is reflecting upon our power and privilege. Our positionality is constantly renegotiated throughout the process and in the specific context of each collaboration (Stoecker, 2003; LeDantec & Fox, 2015; Marshall & Rotimi, 2015).
Across iterative and ongoing cycles of constant exchange between us and the community researchers, they systematize the findings, uncover patterns and crystalize political demands.
The inquiries’ findings, which are very diverse and adapted to each workplace’s material and political conditions, range from research reports over podcasts to documentaries showcasing workers’ daily lives. This breadth of knowledge mobilization allows the research to move beyond the confines of academic publishing to have a larger impact in the public sphere.
What is a Workers’ Inquiry?
Based on an understanding of capital’s development as dynamically changing, the Data Workers’ Inquiry is an attempt to adapt Marx’s 1880 Workers’ Inquiry to the phenomenon of data workers who are both essential for contemporary AI applications yet precariously employed—if at all—and politically dispersed. In this spirit, we invite data workers to provide “exact and positive knowledge” (Marx, 1880) that is collectively produced and politically actionable.
The rich history of the workers’ inquiry guides our methodological considerations. For Marx, the workers themselves are the epistemic authority concerning historically specific manifestations of capitalist exploitation and the necessary agents of social change:
“We hope to meet in this work with the support of all workers in town and country who understand that they alone can describe with full knowledge the misfortunes from which they suffer and that only they, and not saviors sent by providence, can energetically apply the healing remedies for the social ills which they are prey” (Marx, 1880).
This way, the workers’ inquiry is intended as a format for political intervention, a tool for mobilization and organization of workers’ struggles. This approach to knowledge acquisition was picked up both by Trotskyists and the Operaismo movement (Woodcock, 2014; Haider & Mohandesi, 2013) throughout the 20th century. Whereas the former rejected academic inquiries as too detached from workers’ actual concerns and argued for data from engaged people as a necessary tool for emancipation, the latter argued that the working-class viewpoint is imperative to decipher the precise class composition of a historical moment, comprising political, technological, and, crucially, social components. In recent years the workers’ inquiry was utilized by collectives such as Kolinko and Notes from Below to uncover the political logics of global, technologically mediated workers’ exploitation.
The politics of each workers’ inquiry stems from an understanding of the ever-changing composition of labor in capitalist modes of production, and its potential for revolutionary social transformation (Figiel et al., 2014). Our Data Workers’ Inquiry aligns with Operaismo’s proposal of a “participatory action research” (Woodcock, 2014) agenda to engage those who might otherwise be the objects of research to actively shape research questions and methods (Howard & Irani, 2019).
How does the Data Workers’ Inquiry work?
In this project, the data workers serve as community researchers. Every community researcher works for two to four months on their inquiry and is compensated for all their working hours. We collaborate with data workers globally. A decisive sampling criterion is that these are data workers who are already organized in workers councils, unions, communities, or advocacy organizations.
Each community researcher develops their own research questions, designs and conducts their inquiry in collaboration with us and their colleagues, and prepares a presentation format for their findings. These cycles are ongoing, leading to constant exchang between us and the community researchers to systematize the findings and outputs. Each inquiry is unique both in its thematic focus as well as its geographical and organizational context. The inquiries’ findings, which are very diverse and adapted to each workplace’s material and political conditions, range from research reports over podcasts to documentaries showcasing workers’ daily lives. This breadth of knowledge mobilization allows the research to move beyond the confines of academic publishing to have a larger impact in the public sphere.
What is our role?
We, as the organizing team and “informed outsiders” (Headland et al.,1990; Naaeke et al., 2012), conjoin the workers’ experiences of exploitation into a repository that highlights structurally conditioned issues. To this end, we meticulously support every phase of the research journey. We provide guidance including adapting and condensing Marx’s original questions, training in specific data collection and analysis methods, and constantly evaluating the legal and ethical boundaries of the inquiries. We also procure financial means to make each inquiry possible and coordinate the translation of findings into the diverse presentation formats available in the project’s repository.
Building and maintaining trust with the community researchers is an ongoing task as is reflecting upon our power and privilege. Our positionality is constantly renegotiated throughout the process and in the specific context of each collaboration (Stoecker, 2003; LeDantec & Fox, 2015; Marshall & Rotimi, 2015). We believe in the workers’ inquiry as a mode of collective sensemaking, where the active participation of workers from below bolsters organizing efforts based upon first-hand, in-depth perspectives on the many facets of workers’ exploitation. The goal is to be proactive supporters while deferring to the workers’ epistemic authority (McAllister, 2022; Muñoz García et al., 2022), as only they are fully knowledgeable of the multiple dimensions of data work and their struggles as data workers (Gallagher et al., 2023).