Social credit scores aren’t enough. All-pervading censorship doesn’t touch the sides. So now Chairman Xi Jinping wants his country to push ahead with constructing a “national security risk monitoring and early warning system” to find dissenters before they dissent.
Addressing a meeting of the National Security Commission on Tuesday, Xi said China faced a “complex and grave” situation.
He warned officials that they must “adhere to bottom-line thinking and worst-case-scenario thinking, and get ready to undergo the major tests of high winds and rough waves, and even perilous, stormy seas”.
But people aren’t his only problem.
He’s also worried about the solution he’s banking on — artificial intelligence.
Chairman Xi wants AI to say only nice things about him. It must toe the party line. So he’s ordered Chinese programmers to find a way to turn his political dogma into an all-consuming reality.
Shall we play a game?
Chairman Xi’s proposed national security solution is an AI-enhanced “real-time” monitoring and early warning system.
“China should fully implement the holistic approach to national security and be well-prepared to deal with long-term, complex and severe internal and external challenges to maintain long-term stability and security,” the Beijing-controlled China Daily news service quotes a National Security Academy official as saying.
“It is necessary and urgent to perfect the national security risk monitoring and early warning systems to prevent and defuse major risks.”
To Xi, national security is everything. Literally.
The definition he has been enforcing since coming to power in 2013 encompasses political thought, economic theory, cultural expression, cybersecurity and defence posture.
“We must accelerate the modernisation of the national security system and capabilities, highlight the clear guidance of actual combat and practical use,” Xi told the commission. “(We must) pay more attention to efficient co-ordination, rule of law thinking, technology empowerment, and grassroots foundations, to promote the construction of all aspects of organic articulation, linkage and integration …”
And when it comes to AI, it’s the same as any other intelligence: It’s either a benefit or a hazard. If it’s a benefit, it’s not a problem.
More human than humans
“China’s aspirations to become a world-leading AI superpower are fast approaching a head-on collision with none other than its own censorship regime,” China analysts Nicholas Welch and Jordan Schneider argue in Foreign Policy.
This week, China’s national Cyberspace Administration issued rules targeting generative AI — a new type of algorithm. This can comb vast amounts of information to understand, summarise and generate answers to specific questions. It’s the technology behind the “large language model (LLM)” system ChatGPT that has taken the digital world by storm since it was released last November.
But it appears Beijing only wants a particular type of intelligence.
One that conforms entirely to “Xi Thought”.
“The fundamental problem is that plenty of speech is forbidden in China—and the political penalties for straying over the line are harsh,” the analysts write. “A chatbot that produces racist content or threatens to stalk a user makes for an embarrassing story in the United States; a chatbot that implies Taiwan is an independent country or says Tiananmen Square was a massacre can bring down the wrath of the CCP on its parent company.”
Watching the watcher
The new Chinese legislation insists AI models must be proven to only generate answers that embody the Chinese Communist Party’s (CCP’s) “fundamental socialist values” before being released to the general public.
“(It must) not contain any content that advocates the overthrow of the socialist system, incites division of the country or undermines national unity,” the Cyberspace Administration regulations demand.
That’s likely to be a steep hurdle for the likes of Chinese tech giants Alibaba, Baidu, Bytedance, SenseTime and Tencent to leap.
“Everything in Xi’s PRC is national security, and there is an intensifying focus on better-co-ordinating security and development, with the security side winning out over the economics side, it appears,” China analyst Bill Bishop wrote in his Sinocism newsletter this week.
“There is also an intensifying hardening of the system against all sorts of threats foreign and domestic.”
The National Security Commission reportedly approved two new official policy documents this week — “Opinions on Accelerating the Construction of the National Security Risk Monitoring and Early Warning System” and “Opinions on Comprehensively Strengthening National Security Education”. However, no details were released on what they contained.
Deep thought
Analysts say large language algorithms have one serious flaw: nobody knows exactly how they think. And that means these AI systems apply a degree of “imagination” — or just straight-out “lie” — when answering some user requests.
OpenAI’s ChatGPT system, which has triggered an industry-wide rush to produce similar algorithms, isn’t allowed in China. Even the likes of Google and Facebook are banned — although Apple appears to have conformed sufficiently for the Communist Party to allow the use of its products there.
Now China’s tech industry is stuck between a rock and a hard place.
President Xi Jinping declared in 2018 declared artificial intelligence was a strategically important goal and has since invested billions of dollars in fast-tracking its development.
In February, the Ministry of Science and Technology stated that it was “committed to supporting AI as a strategic emerging industry and a key driver of economic growth” and that one of the “important directions for the development of AI is human-machine dialogue based on natural language understanding.”
Baidu unveiled its chatbot — Ernie — in March.
It was a risky thing to do.
Just weeks earlier, Google ended up with egg on its face after the unveiling of its Bard chatbot. In a live demonstration, the wayward AI invented an answer to a question it didn’t know.
And that may be why Baidu only offered prerecorded question-and-answer sessions as evidence of its system’s performance.
The only way is down
Chairman Xi’s “cultural engineering” efforts extend far beyond making AI Communist Party compliant.
His Cyberspace Administration announced in May that it would clamp down on anyone posting anything on social media that “deliberately manipulates sadness, incite polarisation, create harmful information that damages the image of the Party and the government, and disrupts economic and social development.”
That includes any mention of being poor.
Chairman Xi famously declared a “comprehensive victory in the battle against poverty” in 2021. That now means no reporting on the subject will be tolerated.
A recent viral video showed a retiree demonstrating the meagre basket of groceries her monthly pension could buy. It was deleted by Beijing authorities.
Now Xi has ordered Beijing’s State Art Museum to uphold “a politically correct direction”. It comes after high-profile painter Yue Minjun was pilloried for one of his self-portraits appearing to lampoon the People’s Liberation Army.
Xi’s instructions were outlined in a letter to the institution on the 60th anniversary of its founding. It must “champion the core values of socialist culture under Xi Jinping Thought,” reports the China Daily.
“It should persist in upholding political correctness and practising the core values of socialism in order to make greater contributions in the new era.”