The most serious attempt to regulate AI in China
As of March 1, China has officially implemented new rules that will limit data collection and AI capabilities. These will, in part, affect algorithms that set prices, manage search results, recommend videos and filter content, and will primarily affect large tech companies. This is due to a phenomenon known as algorithm bias, which comes in a range of different formats and ways of of digital discrimination.
Algorithm bias has increasingly started to be noticed by a growing number of Chinese citizens. For instance, when using apps to order a cab, some have noticed that those who are less likely to summon expensive cars are assigned lower prices for the same trip. They blame the company’s algorithms and believe it is a matter of profiteering. The complaints have been issued toward Chinese carriers.
But these carriers claim that prices vary because of traffic fluctuations. But this doesn’t tell the full story. According to a number of studies and publications, apps can set different prices depending on a number of factors, including travel history and the phone a person uses. In relation to completely regulating this incognito practice, there has been little movement in the country. Au contraire, Western commentators have speculated that this practice has been furthered by the government itself in relation to the country’s social credit system.
However, as of March 1, China has completely banned such algorithmic discrimination and is probably the most serious attempt to regulate AI so far. The rules will affect algorithms that set prices, manage search results, recommend videos and filter content. Restrictions will affect large businesses that engage in advertising, e-commerce, streaming, and social media.
Xi Jinping steps in
As per the Chinese President, Xi Jinping, this is only the next round of measures aimed at regulating against China’s most popular and valuable companies. They have previously received heavy fines and had to postpone public offerings. As he noted during an October 2021 speech, “some unhealthy and disorderly signals and trends have occurred in the rapid development of our country’s digital economy.”
Some provisions aim to address problems with online services. In particular, companies will be prohibited from using users’ personal characteristics to offer them different prices for a product. They will also be required to notify users and allow them to opt out when algorithms are used to make recommendations.
The rules were developed by the CAC (Cyberspace Administration of China). The administration enforces cybersecurity, Internet censorship and e-commerce regulations, prohibits fake accounts, traffic manipulation and promotion of addictive content, and protects delivery workers, drivers and other gig economy workers.
Violating companies could face fines, banning new users from signing up, revoking licenses, and shutting down websites or apps.
Not easy to enforce
Some elements of the new rules may be difficult or impossible to enforce. For example, it may be technically difficult to control the behavior of an algorithm that is constantly changing because of new inputs. But the Chinese public seems to be largely supportive of measures aimed at limiting the influence of the large technology platforms whose services they use every day, even if it concerns large companies.
The past 18 months have shown that Chinese regulators are not shy about punishing these well-known companies. The app Didi was removed from Chinese app stores shortly after its U.S. IPO because the government had questions about its data policies. The giant Alibaba was also forced to pay millions in fines for antitrust violations.
Some of the rules are rather vaguely worded and can be interpreted in a multitude of ways. The introduction of rules also touches upon other areas.
Another section of the proposal requires Chinese companies to avoid policies that make users addicted. Last year, similar concerns led the country to restrict the ability to collaborate with Influencers and the time minors could spend playing video games.
Chinese companies have already made changes
ByteDance, the developer of the video service Douyin, in October began showing five-second videos urging users to log out after a long viewing period. The move was designed to combat the addiction to “algorithmically selected videos”. The company has also implemented features in Douyin and other apps that allow users to opt out of providing personal information to recommendation algorithms.
An investigation by a Chinese media outlet found that 26 out of 28 popular apps last year gave users the option to opt out of personalized recommendations due to another law concerning the protection of personal information.
Chinese technology companies have also begun closely monitoring content to comply with rules on sharing information the government deems “harmful”.
Earlier this month, the social media platform Weibo deleted 71,000 posts criticizing Olympic athletes, including Zhu Yi, an American-born figure skater who competes for China.
Tencent Video published a version of the 1999 movie “Fight Club” with the ending altered by a Chinese distributor (the original was quickly restored).
It is also reported that scenes depicting LGBT themes in the sitcom Friends, which is hugely popular with Chinese viewers, have reportedly been cut or altered.
A separate but related proposal addresses the problem of so-called “synthetic content”. This term includes fake news, synthetic audio, and deepfakes. The provisions also require that producers of deepfakes software verify the real names of authors and “explicitly mark” any of these fakes.
Software to make these is popular in China and has sparked heated discussions about privacy and ownership of personal data.
Evolvera – Tech, Startups and Futurism in the New Era.