Chinese leaders prioritize AI safety at major economic policy summit

Artificial intelligence has been mentioned several times in resolution Last week, China’s third plenum set the agenda for the Chinese leaders’ meeting, which focused on economic policy. The resolution addressed how AI should be harnessed for economic growth, cooperation with developing countries, and protecting minors online. But the third plenum focused primarily on AI safety and the implications of the technology for China’s national security.

Xi Jinping full explanation The results of the third plenum include a proposal to create an AI safety monitoring mechanism (establishing an AI regulatory system is a closer translation (from the original). In language familiar to anyone who has read official American proclamations on the subject, such as that of President Biden Executive Decree Regarding safe, secure and trustworthy AI, Xi Jinping called on China to better harness science and technology to safeguard national security.

Around the world, governance-minded leaders are increasingly convinced that it is their responsibility to ensure AI safety, even if their efforts come at a cost to the industry. But they still cannot agree on what the tolerable cost is or, more fundamentally, on the details of what constitutes safety.

This fear of stifling innovation is still relevant in America, but not as much as it might have been when discussing the downsides of AI for users and the benefits for tech billionaires like Mark Zuckerburg, who published a paper on how AI can stifle innovation. manifest Yesterday, advocates of open source AI models were less opaque. Media coverage of AI is hurting fields such as deepfake porn made the dangers of generative AI more visible to voters.

< position="inread" progressive="" ad-id="article-0-inread" aria-hidden="true" role="presentation"/>

Industry players may be starting to panic. According to Zuckerburg’s manifesto: “I believe it will be better to live in a world where AI is widely deployed so that the largest players can check the power of the smaller bad actors,” he wrote, conveniently imagining the good guys in the majority.

His argument is practical in another way: the open source standard he envisions could absolve he and other tech CEOs against downstream abuses of AI that are neither possible nor likely; they are already happening. As Arati Prabhakar, director of the White House Office of Science and Technology Policy, has said. said on the Decoder The fact is that devastating damage is happening today, and I think we should not pretend that this is just a future problem.

While Washington has been slow to act, some localities seem increasingly aware and concerned about the dangers of unchecked AI. To that end, California passed SB1047 and New York City instituted an AI law to guard against bias. Meanwhile, in March, the European Parliament passed the AI ​​Act AI Lawwhich argues that security and innovation are not mutually exclusive.

In this context, the Third Plenum’s focus on AI security suggests that China’s leaders are not only concerned about winning the AI ​​race against America; they are also concerned about winning internal battles rooted in AI’s potential to disrupt public order or, in the party’s terminology, social stability. This unease is not new. Last year, in the first such instance, Chinese authorities stopped a suspect for using a chatbot (in this case, OpenAI's ChatGPT) to create and spread dangerous synthetic content (a story about a train crash that never happened).

But as AI applications have expanded and diversified, China's calls for national and global safeguards have kept pace. On July 1, the United Nations General Assembly adopted a resolution on AI governance proposed by China. Less than two weeks later, Wang Hongyu, director of the Research Office of the Legal Affairs Commission of the National People's Congress, delivered a speech At the China Rule of Law Conference, the Chinese president delivered a speech on AI governance. His remarks focused on legal responsibility and the profound ethical issues raised by AI. He warned against exaggerating the risks, but also called on China to comprehensively protect national security interests, industrial development, citizens’ rights and interests, and social justice. China’s discourse on how to govern, deploy, and nurture AI carries a tension familiar to other major economies.

Moreover, while Chinese experts appear to recognize the national security threats inherent in current and future AI systems, they have also shown that they have no intention of rising above a more traditional national security concern: competition with the United States. It would be hard to resist the narratives of rivalry, risk, triumph, and victimization that can all coexist in the context of great-power competition for great-power technologies. Neither the United States nor China seem particularly interested in trying.