Saturday , September 19 2020
Home / ARTIFICIAL INTELLIGENCE / New world. The worrying democratization of facial recognition

New world. The worrying democratization of facial recognition

Pedestrians on a street in an American city. (JAMES LEYNSE / CORBIS HISTORICAL)

A technology that identifies a person from a simple photo, by comparing them with some three billion images from social networks like Facebook, YouTube, etc. It is not an application created by a large technology group or by a government agency, no, it is a cheap solution that comes from a simple start-up.

According to the New York Times, which reveals the existence of this solution, it is even already ready to work in an augmented reality headset.

Developed by Clearview AI, this solution is starting to be controversial in the United States, because it is based on the exploitation without authorization of photos from Facebook in particular. Above all, however, this raises the question of the democratization of a technology as sensitive as identification by facial recognition. It's too simple, too cheap. The New york times reveals that the app is already in use by several local police agencies, in the United States and elsewhere. Tomorrow, by anyone? Clearview was founded by a close friend of former New York City mayor Rudolph Giuliani and is financially supported by investor Peter Thiel, co-founder of Palantir, a company that specializes in harnessing big data for governments.

The bright side of facial recognition is the ability to find terrorists or children lost in the midst of a crowd. The bad thing is the risk of tracking down ordinary citizens or, in some countries, political opponents. In addition, there is the question of reliability. Clearview announces 75% of success, which leaves a big risk of “false positives”, that is to say people wrongly identified. We know that facial recognition algorithms in particular find it difficult to be effective with people with black skin.

Facial recognition, technically, is a reality today. We can bet that this type of tool will develop, as will those used to create deepfakes (fake news from videos). So it all depends on the use that is made of this technology. The answer is not technological, it is legal.

Cities like San Francisco have made the decision to ban facial identification by the police. In France, experiments are carried out (Nice, Marseille). At the European level, prudence prevails. The European Commission is considering a three- to five-year moratorium, during which facial identification would be prohibited in public space. Like everything related to the balance between security and privacy, it is therefore an eminently political question.

Leave a Reply

Your email address will not be published. Required fields are marked *