This project exposes the hidden human cost behind the AI boom, a cost disproportionately borne by workers from and in the Global South.
As Big Tech races to build “smarter” systems, millions of digital workers, known as AI trainers or tutors, are recruited across the world to train large language models like ChatGPT. Marketed as highly skilled, flexible work, the reality is far more exploitative: no promise of work, low pay, and little agency over their livelihoods.
By collecting and analyzing thousands of job postings from Mindrift, a micro-tasking platform owned by Toloka, now part of the Netherlands-based Nebius Group, this reporting project reveals a calculated recruitment strategy: Hire en masse despite limited work, creating the illusion of scale to secure lucrative contracts with Big Tech clients.
Through intimate profiles of African digital workers, this investigation shows how human labor is used as collateral in a high-stakes gamble to attract investment and fuel the hype around artificial general intelligence (AGI) — something experts warn may never exist.
By centering African voices, this reporting project exposes how Big Tech’s AI ambitions deepen global inequities and asks a critical question: Who profits from this "bubble," and who is left behind to bear its costs?
Image caption: Image by Janet Turra / https://betterimagesofai.org / https://creativecommons.org/licenses/by/4.0.
