Unveiling the Global Workforce of Data Workers
Behind sophisticated systems lies an overlooked army performing repetitive labour
A mid all the talk about
Tasks machines cannot perform well are often offloaded to marginalised workers across the globe struggling in precarious labour markets. They perform ostensibly ‘automated’ work under exploitative conditions.
Data work is essential to building and refining AI systems. Before AI models can ‘learn’, human workers must categorise, label, test and moderate vast volumes of text, images, audio and video to make data usable for training.
This labour is performed by an expanding global digital workforce that prepares datasets not only for big tech, but also for high-stakes industries such as banking, insurance, healthcare and government agencies, including defence.
To understand the
Inequality is baked in
My interviews with 10 people to date show that precarious labour markets and marginalised social status have pushed digitally literate young workers into the data-labelling industry.
As one interviewee said, “We do the manual work so that they get the credit for the intelligence.”
There is considerable inequality across the data labour market, shaped by workers’ qualifications and geographic location.
Those with PhD-level or equivalent qualifications and STEM certifications can typically access more specialised tasks. If based in the Global North, such workers tend to be higher-paid, earning A$400–800 per hour, depending on the task.
But such specialised and highly paid tasks are rare. Most workers I interviewed perform general tasks, such as repetitively drawing bounding boxes around images used in drones, self-driving cars and automated vending machines, or annotating audio.
These workers normally receive as little as A$6 per day, or even less. The pay cannot cover daily expenses, while long hours leave workers with chronic eye strain and back pain.
Part of the
Data work is not unlike other poorly regulated jobs in the gig economy.
Workers have no formal contracts and are not employees. They are classified as ‘users’, and aplatforms call on them when tasks align with their expertise and track record.
User agreements primarily protect the companies behind the outsourced work, including by requiring workers not to disclose information they see.
This is despite datasets already being anonymised. Workers often have no way of knowing which companies they conduct data labelling for. They do not even know whether humans or AI agents assess their completed work, and have minimal rights to appeal performance assessments.
All interviewees reported getting less work as AI advances. What remains is more difficult and time-consuming. Interviewees expressed little concern about their jobs eventually being replaced by AI, but this apparent indifference stemmed from a pessimistic outlook, “If I don’t make this money, someone else will, and I will be replaced [by AI] eventually anyway.”
As one worker noted, what AI actually affects is the working class itself. This group is expanding as more professionals are pushed into data labelling by the precarity of the current job market.
All work, little pay
How a worker gets paid is determined by the platform. US crowdsourcing platforms generally offer higher-paid tasks and pay workers when they submit their work.
Chinese platforms or companies often pay workers only after tasks have been assessed and confirmed to meet preset standards. As a result, workers can spend hours completing tasks without receiving payment. Workers in China also cannot access US platforms; using a VPN to circumvent this risks triggering an account ban.
Companies prefer consistency in their workforce because turnover is costly. Workers require instruction and training before beginning a task, and further time to complete tasks efficiently. As workers typically get faster the longer they stay in the role, companies want to retain experienced workers. But many leave because the pay is so poor. To offset this, companies have turned to recruiting more vulnerable groups. One example is collaborating with local government initiatives supporting disabled people. Such workers are less likely to quit because the job is often their last resort.
Workers reported being unable to find other employment or searching for full-time positions due to disability, pregnancy or being recent graduates.
The bigger picture is grim
The AI economy has created jobs, but many involve workers correcting errors and handling tasks machines cannot resolve. Such work can be cognitively and emotionally demanding.
Workers may not know whether they answer to human managers or AI systems, weakening their bargaining power.
As Australia pursues an AI-driven economy, the focus should be on job quality, not just numbers. Without protections, short-term gains could come at the cost of workers’ wellbeing and quality of life.
-By Fan Yang, The University of Melbourne, The Conversation
