Analysts say China lags far behind Western countries in AI development in many areas. But there is one area where Beijing is ahead of Washington, and that is by imposing regulations on the AI industry.
Chinese authorities are aggressively regulating some uses of AI, especially those that allow the general public to create their own content, making compliance a major hurdle for companies in the country.
Some Chinese companies have even gone so far as to create their own rules. Douyin, the Chinese version of TikTok, requires that AI-generated content be labeled and that anyone posting on the platform registers their true identity.
As the use of AI explodes, regulators in Washington and around the world are looking at how to address potential threats to privacy, jobs, intellectual property, and even human existence itself. increase.
Senate Majority Leader Charles E. Schumer of New York last month called on Congress to introduce “comprehensive” regulation of the AI industry.
But there are also concerns that putting up guardrails on the technology in the U.S. would cede leadership in the field to Chinese companies.
Here’s an overview of China’s AI priorities and how the Chinese government is regulating this new technology.
First of all, what is the situation of China’s AI industry?
Chinese companies have been investing in AI for years, with things like installing surveillance cameras on city streets and using facial recognition to monitor paper use in public restrooms. We are at the forefront of surveillance technology.
But when it comes to other types of AI, Chinese companies are years behind their international peers.
This is partly because the Chinese Communist Party-run government tightly controls information and communications.
Rather than focusing on AI technologies such as chatbots and image generators that allow ordinary people to create their own content, Chinese companies are focusing on technologies with clear commercial uses, such as surveillance technology.
As a result, in other areas of AI, such as the large-scale language models that underpin chatbots, Chinese companies have largely had to build on models developed outside of China.
“Many Chinese companies have neither the money nor the leeway to invest in basic AI development, and are often forced to follow the trajectory blazed by American companies,” said Jeff Ding, an assistant professor of political science at George Washington University. said. , focuses on the AI race between the United States and China.
If Washington regulates AI, will it help Chinese companies leapfrog?
In a recent study, Ding found that most large-scale language models developed in China lag behind those developed in the United States by nearly two years, even if US companies had to adapt to regulations. , found it difficult to bridge the gap.
This gap also makes it difficult for Chinese companies to attract world-class engineering talent. Many people want to work for companies that have the resources and flexibility to experiment with cutting-edge research areas.
Restricted access to the cutting-edge chips required to run AI models further exacerbates these problems.
Experts say lawmakers in Washington are debating ways to regulate AI in a way that limits its potential harm, but they have done something the Chinese government has not been able to do: constrain future development. He said there is an opportunity to find ways to put guardrails in AI technology, rather than
How have US restrictions on access to cutting-edge chips affected China’s AI development?
Late last year, the Biden administration restricted the sale to China of certain chips made around the world at U.S. facilities. The US government said the move was necessary to prevent US technology from being used for military purposes in China or leaking to Russia.
This has made it difficult for Chinese tech companies to access cutting-edge chips that run complex AI frameworks. The Chinese government has called the measure, which is intended to strengthen US “technological hegemony,” an abuse.
A recent study identified 17 large-scale language models in China that relied on Nvidia chips, with only three using Chinese chips.
While the Chinese government is pushing to manufacture equivalent chips domestically, Chinese AI companies will need to source chips in every possible way. That includes sourcing from the black market that has sprung up in Shenzhen. A state-of-the-art NVIDIA chip there sells for nearly $20,000, according to Reuters. , more than double than buying elsewhere.
What are China’s priorities and biggest advances in AI?
Despite the obstacles, Chinese AI companies have made great strides in certain AI technologies, including facial recognition, gait recognition, artificial reality and virtual reality.
These technologies will also spur the development of China’s vast surveillance industry, giving Chinese tech giants a global sales edge, such as Huawei’s contract for smart city surveillance from Belgrade, Serbia to Nairobi. ing.
Helen Toner, director of strategy at Georgetown University’s Center for Security and Emerging Technologies, said tech companies in the U.S. and other democracies haven’t invested much in developing this aspect of AI.
“Meanwhile, in China, the government is opening the door and telling companies, ‘Access our huge facial recognition datasets.
While Chinese tech companies are pushing hard to develop this kind of AI, American companies are turning to chatbots and other technologies aimed at entertaining people and providing creative tools to ordinary people. was developing.
How is China approaching regulation of the AI industry?
Analysts and industry insiders say the Chinese government’s approach to AI regulation limits the ability of Chinese companies to innovate.
Companies developing AI in China must comply with certain laws regarding intellectual property rights, privacy, preferred algorithms, and synthetic content, also known as deepfakes. In April, the regulator also unveiled a set of draft rules for generative AI, the technology behind image-generating software Stable Diffusion, and chatbots such as OpenAI’s ChatGPT and Google’s Bard.
They also need to ensure that AI-generated content complies with the Chinese government’s strict censorship regime. Chinese tech companies such as Baidu are adept at filtering content that violates these rules. But that hinders our ability to test the limits of AI.
“We can say a lot about Chinese AI developers, but one thing we can’t say is that they can build whatever they want,” Toner said.
According to Ding, a professor at George University, Chinese tech companies have yet to release language models as large as OpenAI’s ChatGPT to the public, in which the company will try and test its generative AI models on the public. It is said that they are asking for University of Washington.
“That level of freedom is not allowed in China. One reason is that the Chinese government is very concerned about people creating politically sensitive content,” Ding said. .
Can the US government learn anything from how the Chinese government is approaching AI regulation?
Chinese government regulations have placed a heavy burden on China’s AI companies, but analysts say the regulations include protecting personal information, labeling AI-generated content, and if AI develops dangerous features. It contains several important principles that the U.S. government should learn, including a warning to the government of
Johanna Costigan, a researcher at the Asia Society Policy Institute, said AI regulation in the United States would prevent discrimination, protect people’s rights, and comply with existing laws while easing the Chinese government’s heavy-handed approach. said it could fall below
“There can be alignment between regulation and innovation,” Costigan said. “But the question is confronting the context of what this moment means.Are we doing enough to protect the people using this technology? Because people are using it whether the government regulates it or not.”
