Artificial intelligence tools that generate text, images, and audio are underpinned by a vast, often invisible workforce of low-paid contractors who perform repetitive and sometimes disturbing tasks. A recent investigation by Agence France-Presse (AFP) highlights the plight of these workers in countries like Kenya, Colombia, and India, who say they endure long hours, meager pay, and psychological distress while preparing data for AI models.
The work, known as data labeling, involves sorting and tagging raw data so algorithms can learn to recognize patterns. For instance, to enable a chatbot to produce an autopsy report, contractors must sift through thousands of graphic crime scene images. Ephantus Kanyugi, a data labeler in Kenya, told AFP, “You have to spend your whole day looking at dead bodies and crime scenes. Mental health support was not provided.”
While the labor is often performed remotely, the conditions are far from comfortable. Workers describe 20-hour days, six-day weeks, and a lack of basic workplace protections. The toll is both physical and mental: eye strain, back problems, anxiety, and depression are common, according to Kanyugi. He added, “Then despite working so many hours, you only get poor pay, and you might also not get paid.”
The structure of the industry compounds the problem. Major AI companies such as OpenAI and Google do not employ these workers directly. Instead, they contract with third-party firms, which in turn hire labelers in countries where labor laws are weak or nonexistent. Kenya, for example, has no specific regulations governing data annotation work, leaving workers with little recourse.
One of the largest players in this space is Scale AI, a San Francisco-based company with ties to OpenAI, Meta, and the U.S. Pentagon. According to AFP, Scale AI operates through various subsidiaries and shell companies. One of its subsidiaries, Remotasks, reportedly pays labelers roughly one cent per task, even when a single task can take hours to complete. Kanyugi compared the system to “modern slavery.”
The report underscores a broader ethical dilemma: the convenience of AI comes at a hidden cost. While the technology promises efficiency and innovation, its foundation is built on the labor of workers who are often invisible and undervalued. As consumers interact with AI chatbots or use AI-generated content, the human effort behind these tools remains largely unacknowledged.
Why This Matters
The revelations come amid growing scrutiny of the tech industry's supply chain. Just as concerns have been raised about the environmental impact of AI's energy consumption, the social impact on workers is now drawing attention. Advocacy groups and labor rights organizations have called for greater transparency and accountability from AI companies, urging them to ensure fair wages and safe working conditions for all contractors.
For now, the industry's reliance on low-cost labor shows no sign of abating. As AI continues to expand into new sectors, the demand for data labeling is likely to grow, making the need for ethical labor practices more urgent than ever.