A Stanford University professor thinks that facial recognition software may become so advanced that it could tell your political beliefs and your IQ by just looking at your face. Prof. Michal Kosinski made the headlines last week for suggesting that computer algorithms can tell if a person is gay or straight by analyzing their photos.
But sexual orientation is not the only thing facial recognition programs can detect based on photos, Kosinski claims. Artificial Intelligence (AI) could get so complex that it could spot people’s political beliefs, IQ levels, criminal tendencies, and many other personal details.
The data could lead to major social consequences as people will be profiled before they can even speak up. Kosinski warns that a too advanced facial detection technology can raise delicate ethical issues like the potential misuse of the tech to target vulnerable individuals.
The face is an observable proxy for a wide range of factors, like your life history, your development factors, whether you’re healthy,
The Stanford professor said.
Governments Already Have the Tech
The professor’s computer program can correctly tell sexual orientation of men 91% of the time and women in 83% of cases, just by looking at a few photos.
The LGBT community slammed Kosinski for his research, arguing that the software could be used by governments to spot and oppress the LGBT community. Kosinski replied that governments and corporations already have the tech, so it is crucial to expose the possible risks in order to convince lawmakers to set in place more privacy protections.
Facebook is already using the controversial tech across its platform. Its facial recognition algorithms can spot people in photos with a 98% accuracy, which makes them more accurate even than the FBI’s home-made facial recognition tech, which has an 85% accuracy.
The FBI’s tech is less accurate because it bases its analyses on mug shots of suspects while Facebook has a wide array of photos of people from different angles. What’s more, if a user tags the face of another user in a photo he greatly helps the system to identify that person in all photos across the entire platform.
Image Source: Flickr