
MFW11 BEST DOCUMENTARY SHORT
THEIR EYES
[d] Nicolas Gourault
France / 2025 / Documentary / 22:36
A screen-based investigation about the invisible workers who teach AI how to read the world.

Biography
Nicolas Gourault is an artist and filmmaker based in Paris (FR) with a background in visual arts and visual studies. He has worked with Forensic Architecture before graduating from Le Fresnoy, Studio national des arts. His work navigates between online open-source investigations and the critical use of new media as documentary tools.
Filmography
This Means More – student film – 2019
VO – student film – 2020
Their Eyes – 2025
Director’s Statement
Their Eyes is a loose prologue of my previous film VO (2020), in which I investigated the first deadly accident between a pedestrian and a self-driving car. One cause of the accident, which I didn’t investigate further at the time, was that the AI of the car was unable to detect any human walking on the road outside a crosswalk. This missing category with tragic consequences thus triggered a follow-up research on how are those self-driving cars trained to make sensé of the world we live in. I discovered that „the dirty secret of AI is that it is all supervised learning“, as Adrien Gaidon, a machine learning researcher at Toyota puts it. It basically means that all the data used to train AI has to be manually processed. It involves an immense amount of tedious manual work that can’t be automated yet and is thus mainly outsourced to countries of the Global South where the workforce is cheaper.
Their Eyes explores the daily reality of online micro-workers from Venezuela, Kenya and The Philippines, who annotate images for self-driving cars. It investigates the power asymmetries and neocolonialist exploitation involved in the human labor necessary to train AI systems. The film reveals the invisible work that help shape how machines read our world. Yet, far from solely focusing on the exploitation of this system, the film emphasizes the agency and the know-how of the workers, as well as the different micro-strategies by which they make a bit more sense of this alienating work and try to organize collectively to alleviate some of their working conditions. Their Eyes reveals the blind spot in current AI systems which ignore the extractivism they are built upon. Through this, it thus also raises the question of how to reappropriate these technological images which depict and categorize the world we live in without our prior knowledge or consent.