You’re sitting in class listening to today’s lesson when your mind starts to wander. At first, your teacher doesn’t notice you’ve stopped paying attention—but the cameras watching you do. They scan the faces of you and your classmates every second to determine how engaged you are. The devices alert your teacher, who tells you to quit daydreaming. You sit up straight and glance warily at the artificial eyes watching you.

This scenario may sound like science fiction, but facial recognition technology like this is actually being used in a school in China. It can detect when students are listening, answering questions, writing, interacting with one another, or asleep. A similar system tested in another Chinese school even analyzes students’ expressions to track their emotions: if they’re angry, happy, disappointed, or sad. The data collected by both systems is used to evaluate students’ class performance. Each person receives a score, which gets displayed on a screen at school and can be viewed on an app by parents, pupils, and teachers.

You’re sitting in class listening to today’s lesson. Then your mind starts to wander. At first, your teacher doesn’t notice you’re not paying attention. But the cameras in the room do. They scan the faces of you and your classmates every second to see how interested you are in the lesson being taught. The cameras alert your teacher, who tells you to quit daydreaming. You sit up straight and glance carefully at the artificial eyes watching you.

This may sound like science fiction. But facial recognition technology like this is really being used in a school in China. It can tell when students are listening, answering questions, writing, talking with one another, or asleep. A similar system is being tested in another Chinese school. It even watches students’ faces to track their emotions. It notices if they’re angry, happy, disappointed, or sad. The data from both systems is used to determine how students are doing in class. Each person gets a score, which is shown on a screen at school. The score can be viewed on an app by parents, pupils, and teachers.

Right now, schools in the U.S. aren’t using this kind of technology to keep tabs on students’ behavior. And it may never be used for that purpose here. But some American schools are considering using facial recognition for other things, like scanning school grounds to spot people who may pose a danger to students. In fact, facial recognition is currently used mainly for security purposes. It can help police identify suspects from video footage. Stores have tested the tech to catch repeat shoplifters. Some makers of electronic devices, like doorbells with built-in cameras that identify visitors, tout facial recognition as an added safety feature.

Some people are embracing this high-tech trend. But it’s raising red flags for others. They believe facial recognition, which is often used without individuals’ knowledge, could violate people’s privacy. The technology has also come under fire for being inaccurate, particularly when identifying people of color. Despite these objections, the use of facial recognition is becoming more widespread, showing up in places from schools and airports to concert venues.

Right now, schools in the U.S. aren’t using this kind of technology to watch students’ behavior. And it may never be used that way here. But some American schools are thinking about using facial recognition for other things. For example, they could scan school grounds for people who may be dangerous to students. Facial recognition is now used mainly for security reasons. It can help police identify suspects from video. Stores have tested the tech to catch repeat shoplifters. Some doorbells have built-in cameras to identify visitors. The makers of such electronic devices say that facial recognition is an added safety feature.

Some people welcome this high-tech trend. But it’s raising red flags for others. They believe facial recognition could invade people’s privacy. That’s because people often don’t know it’s being used. Another worry is that the technology can be incorrect, especially when it identifies people of color. Even with these concerns, the use of facial recognition is becoming more common. It’s showing up in places from schools and airports to concert sites.

Facial recognition is just one of many types of technology today that utilize biometrics to identify people. Biometrics are measurements of physical characteristics unique to each individual. If you’ve ever used your fingerprint to unlock your cell phone, then you’ve been recognized with the help of biometrics (see Fingerprints and Beyond).

Anil Jain is a computer scientist and biometrics expert at Michigan State University. He investigates how facial recognition technology can benefit society. For example, Jain thinks the technology could aid in solving crimes by allowing detectives to track down suspects far more quickly than traditional methods. If a detective has a picture of a suspect, say from a surveillance camera, a facial recognition program could compare it with a database of known faces in hopes of finding a possible match (see How Facial Recognition Works). “It can do a comparison a million times a second,” says Jain. Some experts think the technology could be used in a similar way to help search for missing persons.

Another advantage of the technology, some say, is to simplify security screenings in high-traffic places. For instance, travelers no longer have to show their passports multiple times at the Delta Air Lines international terminal in the largest airport in Atlanta. They just look at a camera for identification and move along. Delta says this gets people through the terminal and onto planes faster.

Today, many types of technology use biometrics to identify people. Facial recognition is just one of them. Biometrics are measurements of physical traits that are different for each person. Did you ever unlock your cell phone with your fingerprint? Then you’ve been identified with the help of biometrics (see Fingerprints and Beyond).

Anil Jain is a computer scientist and biometrics expert at Michigan State University. He studies how facial recognition technology can help society. For example, Jain thinks the technology could help solve crimes. It could allow detectives to track down suspects far more quickly than older methods. Say a detective has a suspect’s picture from a surveillance camera. A facial recognition program could compare it with a database of known faces. The program would look for a possible match (see How Facial Recognition Works). “It can do a comparison a million times a second,” says Jain. Some experts think the technology could be used like this to help search for missing persons.

Some say the technology has another benefit. It could make security screenings easier in busy places. The Delta Air Lines international terminal in Atlanta’s largest airport is an example. There, travelers no longer have to show their passports over and over. They just look at a camera that identifies them. Then they move along. Delta says this gets people through the terminal and onto planes faster.

In the wake of several deadly shootings at schools in the U.S., some educational institutions are interested in using facial recognition to better protect students. In 2018, officials in Lockport, New York, installed the technology in some of the city’s public schools. As of press time, the system hasn’t yet gone live. But if it does, its cameras would scan people’s faces at school entrances and in hallways. The images would be compared with a database of individuals, like suspended students, not allowed on campus. The software can also detect if someone is holding a gun. If a match is found for a face or a weapon, school officials are notified so they can decide whether taking action, like calling the police, is needed.

Critics, though, say there’s no evidence this technology makes schools safer. Groups that work to protect basic human rights, like privacy, have concerns too. “I’m worried about how schools are conditioning young people to expect that everything they do is going to be monitored and tracked by authorities,” says Kade Crockford. She’s the director of the Technology for Liberty Program at the American Civil Liberties Union of Massachusetts. After pushback from civil liberties groups and parents, New York’s department of education is now reviewing Lockport’s facial recognition system to determine how it should best be used.

Dozens of deadly shootings have happened at schools in the U.S. That’s why some schools want to use facial recognition to protect students better. In 2018, officials installed the technology in some public schools in Lockport, New York. As of press time, the system hasn’t gone live. But if it does, its cameras would scan people’s faces at school entrances and in hallways. The images would be compared with a database of people. The database includes suspended students and others not allowed on campus. The software can also tell if someone is holding a gun. If a match is found for a face or a weapon, school officials are warned. Then they can decide if they need to take action, like calling the police.

But critics say there’s no proof this technology makes schools safer. Some groups work to protect basic human rights, like privacy. They have concerns too. “I’m worried about how schools are conditioning young people to expect that everything they do is going to be monitored and tracked by authorities,” says Kade Crockford. She’s the director of the Technology for Liberty Program at the American Civil Liberties Union of Massachusetts. Civil liberties groups and parents have objected to Lockport’s facial recognition system. So New York’s department of education is now reviewing the system to see how it should best be used.

Another problem with facial recognition is how often it makes errors. The technology is far more accurate at recognizing people than it used to be—but not everyone. It has a harder time correctly identifying women of color and sometimes fails to notice them at all. Researchers at the Massachusetts Institute of Technology Media Lab have helped highlight these flaws. One of their studies used software to analyze photos of famous African American women, including former first lady Michelle Obama and tennis star Serena Williams. The result: The program incorrectly labeled them as men.

These issues exist because all people—including facial recognition software developers—have unconscious biases, or ingrained stereotypes, that affect their behavior, according to Meredith Broussard, an artificial intelligence expert at New York University. Facial recognition software relies on artificial intelligence—a computer’s ability to perform tasks normally associated with human intelligence—to learn to recognize faces. But the databases of images used by the software have typically contained more white people than those of color. “People who created this didn’t notice, because they were mostly white men,” Broussard says.

One way to improve the situation is to make the teams working on facial recognition software more diverse. Until that happens, Broussard worries, biases in the technology could have serious real-world consequences, such as leading police to pursue the wrong suspects or mistakenly arrest someone who’s innocent.

Facial recognition has another problem. It often makes mistakes. The technology is much better at identifying people than it used to be—but not everyone. It has a harder time correctly identifying women of color. Sometimes it fails to notice them at all. Researchers at the Massachusetts Institute of Technology Media Lab have helped highlight these flaws. One of their studies used software to look at photos of famous African American women. They included former first lady Michelle Obama and tennis star Serena Williams. The result: The program thought these women were men.

Meredith Broussard is an artificial intelligence expert at New York University. She says these problems exist because all people have biases without even realizing it. These deep-rooted stereotypes affect their behavior. To learn to recognize faces, facial recognition software uses artificial intelligence. That’s a computer’s ability to perform tasks normally done with human intelligence. The software uses databases of images. But the databases have usually contained more white people than those of color. “People who created this didn’t notice, because they were mostly white men,” Broussard says.

If the teams working on facial recognition software become more diverse, that could help. Broussard worries that until that happens, biases in the technology could cause serious problems in real life. They could lead police to follow the wrong suspects or arrest an innocent person by mistake.

Civil liberties groups have strongly opposed the use of facial recognition by law enforcement agencies, as well as in places like schools and airports. In addition to concerns over privacy and accuracy, these groups worry about the U.S. becoming a surveillance state—a country that closely monitors its citizens.

That’s already happening in China. Its government uses facial recognition cameras to create a giant surveillance network that not only curtails crime but also keeps citizens in line. Police, for example, use the technology to publicly shame people caught committing minor crimes. If people cross the street illegally, police can identify them using facial recognition and post their information on giant electronic billboards to embarrass them.

Civil liberties groups are strongly against the use of facial recognition by law enforcement agencies. They also oppose its use in places like schools and airports. That’s not just because of concerns over privacy and accuracy. These groups also worry about the U.S. becoming a surveillance state. That’s a country that closely watches its citizens.

That’s already happening in China. Its government uses facial recognition cameras to create a giant surveillance network. It doesn’t only fight crime. It also keeps citizens in line. For example, police use the technology to publicly shame people who commit small crimes. If people cross the street illegally, police can identify them using facial recognition. Then they post their information on giant electronic billboards to embarrass them.