Shining a Light on the Hidden Labor of Data Work: The Costs and Opportunities of Tech Work Abroad
The world of data work, including tasks such as moderation and annotation, often remains hidden from those who benefit from the results of this labor. However, a groundbreaking project is shedding light on the lived experiences of data workers worldwide, revealing the true costs and opportunities associated with tech work abroad.
Many monotonous, unappreciated, and psychologically damaging tasks have been outsourced to poorer countries, where workers are willing to accept jobs for a fraction of what their American or European counterparts would earn. This labor market falls under the category of “dull, dirty, or dangerous” jobs, similar to electronics recycling and shipbreaking. While moderation and annotation work may not pose physical dangers like losing a limb or contracting cancer, it does not mean these jobs are safe, enjoyable, or rewarding.
The Data Workers’ Inquiry, a collaboration between AI ethics research group DAIR and TU Berlin, takes inspiration from Marx’s late 19th-century work on labor conditions. The reports produced by this project are collectively created and politically actionable.
During an online event, the project’s creators unveiled all the freely available reports. These reports provide firsthand accounts from data workers and offer a glimpse into their realities.
As the scope of AI applications continues to expand, human expertise remains crucial to its development. However, companies often seek this expertise at the lowest possible cost without attracting negative attention. When users report content online, they are unaware that their reports may be handled by someone in a different country earning a mere 3 cents for their work. The sheer volume of reports necessitates the mass outsourcing of moderation and annotation tasks to cheap labor markets.
While the reports in the Data Workers’ Inquiry primarily consist of anecdotes, this was a deliberate choice. Rather than relying on quantitative analyses alone, these reports serve as systematic anthropological observations that capture the essence of the data workers’ experiences.
Quantifying these experiences often fails to capture the true costs involved. Companies tend to focus on statistics that paint a positive picture, such as higher wages, job creation, and savings for clients. Rarely do they mention or measure the negative aspects, such as moderation workers suffering from nightmares or developing chemical dependencies.
One report by Fasica Berhane Gebrekidan highlights the mental health and drug-related challenges faced by Kenyan data workers. Despite working for a company that claims to provide a more ethical data work pipeline, the workers describe their job as relentless misery, with little support from the local office. Exposed to an endless stream of violent, graphic, and hateful content, they must quickly review and take action on each item to meet performance expectations. Failure to do so results in reduced pay. Some workers handle over 500 such items daily, with potentially damaging consequences.
Rahel Gebrekirkos, one of the contractors interviewed, described the job as soul-crushing and expressed concerns about being scarred for life. Support personnel were ill-equipped and unprofessional, leading many moderators to turn to drugs to cope with the psychological toll.
While some may dismiss these reports due to their anecdotal nature, it is crucial to acknowledge their value. Anecdotes always contain some truth, unlike statistics, which can be manipulated. These stories come directly from the source — the data workers themselves. Even if these reports represent just a small fraction of moderators in countries like Kenya, Syria, or Venezuela, their experiences should concern everyone who relies on their work.
The Data Workers’ Inquiry plans to continue its work with a second cohort of data workers from countries like Brazil, Finland, China, and India. While there have been no official responses or changes from the implicated companies yet, the strength of these initial results suggests that further investigation is warranted.
Ultimately, this project aims to expose the hidden realities of data work and advocate for better working conditions and support for data workers worldwide. By amplifying their voices, we can create a more equitable and respectful tech industry.
