The largest U.S. police camera manufacturer says no to facial recognition, saying it will lead to serious bias

Tong Ling from Aofei Temple
Produced by Quantum Bit | Public Account QbitAI

Family members cutting off their own family members’ retreat?

Axon, the largest manufacturer of police body cameras in the United States, has officially announced:

It will not use facial recognition technology in its police body cameras until the technology improves.

This company, which provides camera equipment and software to 47 of the largest police agencies in the United States, has this time given up on facial recognition, a technology with great potential in the security field.

The committee Axon went to said facial recognition technology was not reliable enough to justify its use.

what happened?

Bias + Privacy

The incident was triggered by a 28-page research report.

This is a report from Axon’s internal Artificial Intelligence Ethics Committee, titled “First Report of the Axon AI & Policing Technology Ethics Board”.

The report said current facial recognition technology has "evidence showing unequal and unreliable performance across racial, ethnic, gender, and other identity groups."

The report further stated that the technology is less friendly to people of color than to white people, more friendly to women than to men, and more friendly to young people than to older people.

Barry Friedman, a member of Axon's ethics committee and director of the Policing Project at New York University School of Law, said the technology was not accurate enough and said it would not be considered again until those problems were fixed.

Even if it may take many years.

In addition to the limitations of the technology itself, the report also mentioned the privacy issues brought about by facial recognition .

Axon said in the report:

Technology is evolving at a pace far faster than legislatures and courts can react. Therefore, we believe it is critical for technology leaders to work to understand the ethical, legal, and community implications of new technologies that are too new to be effectively covered by existing law.

As a large security company, Axon certainly wants to embrace emerging AI technologies.

For example, the AI ​​law enforcement recorders owned by police in many parts of the United States can analyze objects captured by the camera, classify them, and automatically archive law enforcement information.

A blog post by MuckRock in the United States shows that at the beginning of this year, Axon began to deploy artificial intelligence technology on a large scale in its equipment. Axon hopes to use AI to replace police clerical and data editing work.

Although Axon strongly advocates AI, they are very cautious about facial recognition technology .

Even before the technology was actually deployed in its products, it was announced that it had been shelved.

"Liberated"

This news made foreign netizens feel great, and they suddenly became excited.

Netizens lamented that facial recognition has taken away the last bit of privacy provided by urban life:

Now, that’s the feeling of freedom!

@ProfFerguson believes that Axon's ban on facial recognition represents a victory, first of all, for the ethics committee, and secondly, for those who care about facial recognition, this is also a good thing.

@Miles_Brundage, a research scientist in the policy direction of OpenAI, said that he was very excited to see the first report of the Axon AI Ethics Committee, and he was also a member of the research team involved in the report.

He believes that facial recognition technology in wearable cameras is still very immature and is expected to have a long way to go.

Some netizens believe that Axon announced the temporary ban in order to control its own reputation. Using technology knowing that it is unreliable may have a bad impact, so they should act with caution.

Some experts analyzed that even if this ban prohibits the use of facial recognition technology in police cameras, it cannot be completely prevented if someone really wants to find loopholes.

An Axon spokesperson said law enforcement officials can download third-party applications, such as Amazon's Rekognition. But Piza, an associate professor at John Jay College of Criminal Justice, believes that this is not easy and requires spending money to purchase another technology, which is difficult to operate in the end.

Facial recognition ban

On the one hand, the recognition efficiency is far higher than that of the human eye, but on the other hand, it is mired in privacy and prejudice. The application of facial recognition is caught in a dilemma.

For the United States, Axon is not the only one taking a conservative wait-and-see approach.

Last month, the first facial recognition ban in history was issued in San Francisco, the center of Silicon Valley. The ban passed by the San Francisco Board of Supervisors, called the Stop Secret Surveillance Ordinance, prohibits government agencies in San Francisco from using facial recognition technology, even police officers who need to investigate cases and arrest people.

The regulations state that surveillance technology represented by facial recognition is always used to oppress special groups of certain races, religions, and sexual orientations, and that the harm that facial recognition technology does to citizens' rights and freedoms far outweighs the benefits it brings.

Therefore, the ordinance stipulates that it is illegal for San Francisco government departments to directly use facial recognition technology or rely on facial recognition technology to obtain certain information.

In other words, facial recognition cameras cannot be directly installed in public places, nor can facial recognition technology be used secretly to automatically match suspects.

In comparison, the deployment of facial recognition technology in China has progressed much faster.

During the national tour of the enthusiastic singer Jacky Cheung, the facial recognition AI won the achievement of catching five fugitives in a row.

Later, the suspect in the Peking University matricide case was successfully matched by the latest facial recognition system at Chongqing Airport after three years on the run.

Technology is advancing, and regulations are improving. The journey of implementing facial recognition technology has just begun.