
I watched that YouTube video from a Swiss media called RTS and I wanted to share that as I found it very interesting to look at what is going on behind the scene. The video is in French originally but an English version is also available if you go the video settings.
The Illusion of Automation While AI is presented as a magical, autonomous technology, the documentary reveals it is heavily dependent on human intervention. Behind every algorithm—whether for sorting recycling or generating text—are millions of “invisible” workers who label data and correct the machines.
Data Labeling and Outsourcing
- Industrial Applications: In Switzerland, a recycling plant uses AI to sort waste, but the system must first be trained by humans who manually annotate thousands of images of trash. This tedious work is largely outsourced to countries with lower labor costs, such as Venezuela [01:23].
- The Hub in Kenya: Nairobi has become a major center for this digital labor. Workers there perform “micro-tasks” like identifying objects for self-driving cars or medical AI. Despite the high-tech nature of the product, wages are low, and employment is precarious, often forcing workers to hold multiple jobs to survive [06:26].
The Human Toll: Mental Health and Exploitation
- Content Moderation: A darker side of the industry involves filtering toxic content for platforms like TikTok and Meta. Moderators describe traumatizing experiences, such as spending 8 hours a day tagging pornographic or extremely violent videos to protect regular users. One worker recounts how this exposure destroyed his mental health and marriage [11:36].
- Legal Action: Mercy Mutemi, a lawyer in Kenya, is suing Meta (Facebook’s parent company), accusing them of exploitation akin to human trafficking and forced labor due to the deceptive and inescapable nature of the work conditions [15:21].
A Global Issue The exploitation is not limited to the Global South. The documentary features a Swiss academic who worked for a platform called Outlier (linked to Meta). He was hired to write creative text and poetry to train AI on local cultural nuances. Despite his qualifications, his effective earnings averaged around 8 CHF per hour due to unpaid mandatory training and research time [19:02].
Conclusion The film argues that the “magic” of AI is a marketing narrative that hides a reality of human exploitation. As sociologist Antonio Casilli explains, AI does not run on its own; it feeds on human knowledge and labor that is often undervalued and invisible [20:33].