AI needs to be controllable, experts say
Experts have called for increased efforts to ensure Chinese companies develop next generation artificial intelligence technology that is “safe and controllable” as the country seeks higher-quality development of AI.
“For AI, innovative development and safety governance are equally important. The AI industry won’t develop in a healthy way if there are safety problems,” said Zhang Bo, president of the Institute of Artificial Intelligence at Tsinghua University and an academician at the Chinese Academy of Sciences.
“The existing AI algorithms are very fragile. If the algorithm is applied to a real scenario that is very different from the training scenario, there will be security hazards.”
In autonomous driving, for instance, it is impossible for an AI system to simulate all possible scenarios during training. If faced with new emergencies it cannot deal with, the technology will be misused and such vulnerabilities also make AI systems vulnerable to attack, he said.
“It is especially the case when cutting-edge technology is being used in more and more scenarios with faster-than-expected speed,” Zhang said, adding that technologies like AI deepfakes — the manipulation of videos to yield seemingly realistic, but fabricated, images and sounds — have had a huge impact on society.
Zhang said more efforts are needed to leverage “knowledge, data, algorithms and computing power” to build next generation AI technologies that are safe, reliable and credible.
China’s top leadership called in May for accelerated steps to turn the country into a scientific and technological powerhouse but also warned of potential risks during technological development.
The remarks came at a meeting that combined the general assemblies of the Chinese Academy of Sciences and the Chinese Academy of Engineering, and the national congress of the China Association for Science and Technology.
The country should also achieve higher levels of self-reliance and strength in science and technology to offer strategic support for national development, the meeting was told.
Tian Tian, CEO of RealAI, a Tsinghua University-backed, Beijingbased AI company, said more efforts should be made in developing AI security — especially related to data and algorithms — to solve hidden dangers such as the leaking of private information.
“At the same time, we need to make sure that applications are controllable, especially in core technologies, so that technology can offer strong support for high-quality economic development,” he said.
According to a report by global market research firm International Data Corp, China will become a major driver of global AI development and account for 15.6 percent of the market by 2024.
Over the next three years, sales revenue from the country’s AI sector will maintain a compound annual growth rate of 30.4 percent to reach $17.22 billion by 2024, the report said.
For AI, innovative development and safety governance are equally important. The AI industry won’t develop in a healthy way if there are safety problems.”
Zhang Bo, president of the Institute of Artificial Intelligence at Tsinghua University