The deputies adopted, on March 23, the use of algorithmic video surveillance, which the executive wants to test on the occasion of the Paris Olympic Games in 2024. According to the companies which sell these technologies, the cameras combined with intelligence artificial intelligence (AI) could detect crowd movements, abandoned luggage or “suspicious” behavior.
However, this technology has never demonstrated the slightest effectiveness, and many observers, including the NGO Amnesty International or the association La Quadrature du Net, warn of the dangers of this “unjustified and disproportionate” surveillance.
Psychologist and documentary filmmaker, Dorothe Dörholt, in a powerful documentary, demonstrates the embarrassing and dangerous shortcomings of these technologies. First, the list of services involved is as long as your arm: security in public places, admission to higher education, job search, justice and crime prevention, transport, bank loans, insurance, consumption habits… Nothing escapes these sequences of mathematical operations so well informed about you that the engineers who make them want to predict your tastes or your behavior.
Welcome to the “surveillance capitalism” that the Gafam (Google, Apple, Facebook, Amazon and Microsoft), in particular, exploit, while blocking, with millions of dollars spent on lobbying, regulatory laws.
fantasy of objectivity
The film begins and ends with dead ends: in Spain, the case of a mother forced into hiding to escape an abusive partner whom the program used by the police does not deem determined enough; in the Netherlands, entire families have been thrown into poverty because they have been unfairly accused of childcare benefit fraud. In question, an algorithm sniffer of free riders from data on the nationality or the place of residence of the families… Between the two, the documentary filmmaker tells how these programs are fed with data by workers paid by the task and the slingshot.
“Perception is cultural, political…”, says American artist Trevor Paglen, author of the installation Faces of ImageNet, which highlights the absurdity of AI: “You can imagine what kinds of biases and prejudices are embedded to these classifications,” he says.
A finding shared by Timnit Gebru, researcher in AI ethics and fired by Google in 2020 for having published a scientific article critical of the “search” function of the engine. Designed by humans, the tool reproduces the stereotypes of the social sphere that produces them.
“The field of AI is populated by white people, men and people who have had higher education in the big schools, says researcher Sasha Costanza-Chock. The developers of these technologies are not representative of the world’s population. Like the American mathematician Cathy O’Neil, Dorothe Dörholt shows how these technologies for monitoring and categorizing human beings have not finished luring us into the transhumanist fantasy of objectivity in order to decide better.
