Some new capabilities of artificial intelligence, or AI, have increased concern about China’s national security. Beijing sees AI systems that can operate without human supervision as a major risk in particular.
There is more discussion in the United States about the future risks of AI. Many researchers there fear that extremely powerful AI which slips out of control could become a threat to humanity.
China has until now been more concerned with the comparatively nearer risks of AI. These included the creation of fake images and videos, child safety and the protection of personal data.
Some recent developments are changing China’s outlook, however. The matter has become tied to national security since evidence emerged that powerful AI models can carry out cyberattacks.
A group of programmers in California in the United States recently said they had used AI to create a cyberattack “worm”. They claimed that if it were released it could be possible to attack millions of accounts on WeChat, the popular social media platform in China.
Kevin Xu, founder of the United States technology investment firm Interconnected Capital, said there are more similarities than differences between concerns about AI safety in China and the United States.
He said, however, that these matters are discussed widely in public in the United States, whereas such discussion is generally much less visible in public in China.
Experts believe that if AI produces cyberattack agents able to operate without human supervision, that could become one of the Chinese government’s greatest concerns.
Sun Chenghao, a researcher at the Center for International Security and Strategy at Tsinghua University in China, said China and the United States are now forming a more realistic idea of the risks of AI. That is also increasing the need for cooperation between the two countries on the subject.
Intense competition between the two countries over AI may stand in the way of cooperation, however. Coordination between them could be difficult in particular on safety testing of AI models and on arrangements for giving warning when a major crisis arises.
The United States president, Donald Trump, has meanwhile stressed maintaining United States leadership in developing AI. He has said the United States is ahead of China in AI and that he wants to keep that position.
The Chinese specialist Sun Chenghao said China fears that attempts may be made to limit the development of AI and other technologies in the name of “security”.
China has already made some rules on the use of AI. Creating images with AI using someone’s face or identity without their permission has been banned, for example.
Brian Tse, head of the Beijing-based AI safety organisation Concordia AI, said China’s AI policy gives more weight to present harms in people’s everyday lives.
In his view China sees AI as a technology that, like electricity or the motor car, can create great benefits for society and industry. But Beijing believes it can be controlled.
The capabilities of AI have been growing rapidly of late. At the same time a number of incidents have brought its potential risks to the fore. National security is therefore becoming more important to China’s policymakers.
Concern and a pragmatic attitude towards AI can be seen among ordinary people in Shanghai as well as in the capital, Beijing. Many think that some problems are bound to come with a new technology. But if problems are identified and solved, they say, the safety of the technology will improve further.
One Shanghai resident said the emergence of the possibility of an AI cyberattack on WeChat is not a bad thing. It has instead created an opportunity to identify weaknesses and fix them.
Many have also argued for strict rules on AI. Their argument is that if responsibility for safety is left to commercial companies alone, they may not impose sufficient restrictions on themselves.
Xia Zhijian, founder of the Singapore-based firm Inspired AI, said his main concern is the breaking of complex passwords and important security codes with powerful AI models.
In his view both China and the United States could be at risk in such a situation. If one country’s AI capability becomes very strong, the other may consider itself weak in terms of security.
বাংলায় মূল প্রতিবেদন পড়ুন · Read the original Bengali report