- The website is part of SHERPA, a European Union-funded "project which analyses how AI and big data analytics impact ethics and human rights."
- The interactive documentary uses your webcam to analyze your face, predicting metrics like age, attractiveness, gender, body mass index and life expectancy.
- Despite the shortcomings of facial recognition, there's currently no set of national laws regulating the use of the technology by governments or private companies.
Facial recognition technology has sparked controversy since law enforcement began using it two decades ago. But as facial recognition continues to creep into our public spaces and private devices, there’s one fact law enforcement, tech companies and privacy advocates can agree on: the technology isn’t perfect.
Sure, facial recognition can identify people with near-perfect accuracy — if the system is fed high-quality images, with good positioning and lighting. But the accuracy rates drop significantly when the systems use photos of people “in the wild,” as the Center for Strategic & International Studies recently noted.
Accuracy rates drop even further when facial recognition tries to identify African-American and Asian faces. In January, this deficit ostensibly led to the wrongful arrest of a Detroit man named Robert Julian-Borchak Williams, who was held in jail for 30 hours.
Want to see for yourself how well these systems work? Check out a new interactive mini-documentary called “How Normal Am I?”, created by Tijmen Schep, a technology critic and privacy designer.
The documentary is part of SHERPA, a European Union-funded “project which analyses how AI and big data analytics impact ethics and human rights.” To experience it, you’ll need to grant the website permission to access your webcam, though “no personal data is collected.” (You can also access the website and then disconnect your computer from the internet; it should still work fine.)
“How Normal Am I?” uses facial recognition to predict your age, attractiveness, body mass index, life expectancy and gender. Don’t get upset if you get a low attractiveness or a high age score: Tilting your head, moving closer to the camera, or just running the program a second time can produce different results.
And that’s sort of the point: If facial recognition technology is unreliable on a broad range of measures, to what extent should governments and the private sector be using it? Even if it does become reliable, to what extent should governments be allowed to use it on citizens?
In a 2019 Pew Research Center survey, a majority of U.S. respondents said it’s acceptable for law enforcement agencies to use facial recognition to scan for threats in public spaces. However, far fewer said it’s acceptable for advertisers to use facial recognition to do things like analyze how people respond to commercials in real time.
What could change how facial recognition operates in the U.S. is a set of national laws, which currently don’t exist. (Although, some states and cities do regulate the technology.) There are currently more than a dozen bills addressing facial recognition technology, ranging from legislation that would outlaw warrantless usage of facial recognition, to banning federal agencies from using it altogether.