Dystopia Come True: China’s Artificial Intelligence Problem
By Anahita Srinivasan
In Xiangyang, China, a billboard has been constructed on the intersection south of Changhong Bridge.
But this isn’t your typical billboard. It displays the photos, government I.D. numbers, and other personal information of jaywalkers at the intersection.
While this may seem like it infringes on privacy, it’s perfectly legal in China.
The display on the billboard is run by artificial intelligence, or AI; more specifically, facial recognition technology. Cameras set up at the intersection can track and identify jaywalkers, search through government databases for their information, and publicly shame them.
“If you are captured by the system and you don’t see it, your neighbors or colleagues will, and they will gossip about it,” spokeswoman Guan Yue said in an interview with the New York Times. “That’s too embarrassing for people to take.”
This may sound like a scene straight from a dystopian movie, but it’s a situation that is becoming all too familiar in the cities of China. Chinese president Xi Jinping’s determination to increase his control over his people has violated the privacy of millions of Chinese citizens and unveiled a terrifying new trend in AI: the use of sophisticated systems for government surveillance.
AI is having a major moment in China, and companies like iFlyTek and Yitu are funneling their resources into creating devices to monitor Chinese citizens. One of the projects that the Chinese government hopes to use is voice recognition and monitoring, which iFlyTek specializes in. The company has collected the voice patterns of over seventy thousand citizens, which it can then use to monitor situations like phone calls in an attempt to identify the voices of people like criminals or even anti-government activists.
“This is potentially a totally new way for the government to manage the economy and society,” said Peterson Institute for International Economics fellow Martin Chorzempa in an interview with the New York Times. “The goal is algorithmic governance.”
The lack of stringent privacy laws in China makes it very easy for companies like iFlyTek to collect the data it needs to improve its artificial intelligence systems. Chinese citizens are told that the collection of their voice patterns is no different than the collection of their fingerprints. In addition, one of iFlyTek’s owners, China Mobile, provides the company with the voice patterns of its eight hundred million consumers.
“We Chinese don’t really realize whether it’s bad or not because it doesn’t seem to affect our daily life,” said Fenny Qi, a fifteen-year-old student in Shanghai. “However, the reason of this situation might be that they control the media well so we just simply don’t know.”
The Shanghai company Yitu is another example of the extremes to which China is taking its AI. The offices of the company are filled with facial recognition cameras that track its employees in an attempt to monitor their productivity throughout the day. And Yitu also won a 2017 contest run by the United States government that challenged competitors to design the best facial recognition system. Many other Chinese AI surveillance companies also did well in the competition.
Facial recognition is arguably the major issue in Chinese surveillance at the moment, and it’s more noticeable than ever. In Wuhu, a fugitive was spotted by a camera as he purchased food. In Qingdao, two dozen criminal suspects were spotted by facial recognition technology and arrested at a beer festival. Drones flying above huge crowds can pick out wanted individuals and alert the police to their presence.
And another trend has emerged amongst the throng of surveillance cameras and tracking drones: facial recognition glasses.
Worn by Chinese police, facial recognition glasses have the ability to identify a person if he or she stands still for a few seconds. Their use? Verifying the identification of travelers.
But nowhere in China is the use of these systems more terrifying than in Xinjiang, home to China’s most persecuted ethnic minority: the Uighurs.
In Xinjiang, facial recognition cameras track the movements of Uighurs as they go about their daily lives. In addition, police officers look through phones and take iris scans. All of this information is placed in a database that contains the biometric data and identification information of the eleven million Uighurs native to Xinjiang.
Special artificial intelligence systems scan through the data and label each person as either “safe,” “normal,” or “unsafe.” This ranking determines everything about an Uighur resident’s life, from where they are allowed to eat to whether or not they can travel.
This systemic discrimination is in part because of Xinjiang’s reputation as a territory prone to riots and separatist discourse. The Chinese government justifies its dystopian-like surveillance on the Uighurs by stating that they follow radical Islamic principles, which are a threat to the safety of the nation.
The artificial intelligence systems in place have forced Uighurs into silence as they are fearful of retaliation against family or friends if they speak out.
“If you are someone that values privacy and you don’t like the idea of being tracked or potentially misidentified by law enforcement, then you’re more likely to try to reduce engagement,” says Rashida Richardson, director of policy research at the AI Now Institute. “One of the problems is that it makes us less safe in society.”
This fear is not without reason. It has been estimated that around five hundred thousand Uighurs are currently detained in “political training centers,” which are re-education systems that attempt to homogenize the Uighur population and assimilate them into mainstream Chinese life. This includes forbidding Uighurs from practicing Islam.
Even though the surveillance in Xinjiang is ubiquitous, the surveillance in the rest of China is not. Large holes in the data system mean that human officers still have to parse the data to ensure that the facial recognition cameras are accurate or that the voice recognition systems have actually identified a threat. Not all the files are digitized, and some of the data spreadsheets are incompatible with each other.
Despite this, the Chinese government is on a fast track to implement seriously restrictive AI throughout the country. As artificial intelligence becomes smarter and faster, the extent of control Jinping has over his citizens increases.
“There’s no privacy and information security these days,” said Li Shufu, the chairman of Geely Holding Group and Volvo Cars, in an interview with SFGate. “When you walk on the road, there are surveillance cameras everywhere.”
