Seeking a Panacea: The Party-State’s Plans for Artificial Intelligence (Part 2)
While seeking to leverage AI to enhance its performance legitimacy, the Chinese Communist Party (CCP) is also taking advantage of the potential of AI and big data to bolster its capacity for control. China’s New Generation AI Development Plan (新一代人工智能发展规划) highlights that AI can “play an irreplaceable role in effectively maintaining social stability,” including through enabling prediction and early warning of risks to social security through rapidly detecting changes in mass public opinion and psychology. Since preserving social stability is considered an existential imperative for the Party, the CCP is actively increasing the use of big data and AI to enhance social governance (社会治理). Indeed, Xi Jinping’s report during the 19th Party Congress included a call for the “intelligentization” (智能化) of social governance. As Samantha Hoffman has written, this concept, which has deep roots in traditional CCP ideology, involves improvement of “governance capacity to shape, manage, and respond to social demands,” with the use of tactics to coerce and coopt individuals. Looking forward, the Party-State’s expanding employment of big data and AI to enhance the precision and pervasiveness of policing, censorship, and surveillance, in collaboration with private enterprises, could reinforce regime stability but may also create new risks and challenges.
At present, China is building a 21st-century Panopticon beyond anything that Bentham could have imagined. The Chinese government has installed millions of surveillance cameras and increasingly utilizes facial recognition to identify individuals of interest, while even expanding into the collection of voice prints and DNA samples. Reportedly, China’s Ministry of Public Security (MPS) is building the world’s largest facial recognition database, which will have the capability to identify any citizen “within seconds.” Not unlike the design of the original Panopticon, there is thus a constant potential for surveillance, which can result in control even in the absence of outright coercion. Concurrently, the new social credit system leverages big data to extend state influence to the level of individual behavior with a range of benefits and penalties, while also serving as a unique identification system for individuals and enterprises.
The sophistication of the AI techniques in use will likely continue to increase. In September 2017, Meng Jianzhu (孟建柱), then Politburo member and the Secretary of the Central Political and Legal Affairs Commission and of the Central Public Security Comprehensive Management Commission, called for taking the initiative in the use of AI to address risks to public security. At the time, Meng Jianzhu emphasized, AI “can complete tasks with a precision and speed unmatchable by humans, and will drastically improve the predictability, accuracy and efficiency of social management.” Indeed, MPS is pursuing the research, development, and implementation of a range of AI-enabled policing and surveillance technologies. For instance, its Third Research Institute (第三研究所), which has focused on “smart surveillance,” supported an AI team, Trimps-Soushen, that placed at the top of the 2016 ImageNet competition in object recognition and detection. The increased accuracy of such algorithms will enhance their precision in the automation (and acceleration) of censorship and surveillance.
Often, partnerships with private AI enterprises are integral to these efforts. For instance, the world’s largest video surveillance company, known as Hikvision (海康威视), is a subsidiary of the 52nd Research Institute of the China Electronics Technology Group (CETC). Hikvision is also a member of a MPS key laboratory, the Key Laboratory for Public Security Informatization Applications Based on Big Data Architectures (基于大数据架构的公安信息化应用重点实验室). The company has collaborated closely with MPS on a number of projects, contributing to the development of technologies for video surveillance and big data platforms for smart cities (智慧城市), while also actively expanding sales in the U.S. and internationally. Similarly, iFlytek (科大讯飞), a prominent Chinese AI start-up focused on intelligent voice recognition and speech-to-text products, jointly established the MPS Key Laboratory of Intelligent Voice Technology (智能语音技术公安部重点实验室) and currently is collaborating closely with MPS in the development of a national voice print database, which has provoked concerns and questions over potential privacy and human rights violations from Human Rights Watch. These are just a few of numerous examples that reflect the emergence of a Chinese surveillance-industrial complex that is rapidly expanding domestically and could be exported internationally.
Within China, a number of emerging techniques and technologies will be first applied in Xinjiang itself, which has become a frontier for such efforts, against the backdrop of the rapid expansion of the security state. Of note, in July 2017, the National Engineering Laboratory of Social Security Risk Sensing and Prevention and Control Big Data Applications (社会安全风险感知与防控大数据应用国家工程实验室), was established in Urumqi, Xinjiang. This new laboratory is led by the CETC Electronics Science Research Institute (电子科学研究院), in collaboration with the Xinjiang Uighur Autonomous Region Public Security Office Special Reconnaissance Team (公安厅特别侦察队), and the MPS First Research Institute, among others. As its name indicates, the new lab is intended to promote an integrated approach to the use of big data to sense, prevent, and control social security risks, enabling “early warning” of potential incidents. The initial experimentation and implementation occuring in Xinjiang, often justified as necessary to counter the threat of terrorism, will be a bellwether for the trajectory of these techniques within China as a whole.
Even as AI could increase state capacity, China’s AI revolution will also become a key test of the Party-State’s capability to leverage the potential of this emerging technology to advance development, while mitigating the risks of disruption. For instance, the Party is likely concerned about the potential impact on employment, since massive economic dislocation could create social unrest, as reflected in the focus in China’s new AI plan on educational programs and retraining workers. In practice, the political issues that such a technology might raise could be varied and difficult to anticipate. For instance, Tencent recently shut down chatbots developed by Turing Robot and Microsoft after both appeared to ‘go rogue,’ with comments criticizing the Party as “corrupt and incompetent.” Thus, at a technological level, the CCP will seek to ensure that AI remains “secure, reliable, and controllable” (安全，可靠，可控). Concurrently, the power and potential of AI systems could render their control a point of contention in elite power politics.
Going forward, China’s quest to “lead” in AI will also constitute a test of the continued relevance of the CCP’s traditional style of central planning in a world in which the locus of innovation has shifted to the private sector. Indeed, the private sector has, in many respects, been the leader in China’s “rise” in AI to date. To some degree, the new AI plan itself may reflect private sector impetus and influence. (For instance, Li Yanhong, Baidu’s CEO, started to advocate for the concept of a “China Brain” project that would direct massive state investment to advance Chinese AI research and enterprises AI as early as 2015.) In actuality, the prospects for successful implementation of China’s ambitious new agenda in AI, which could involve billions in funding, could fall short of perhaps inflated expectations, yet the focus on funding cutting-edge research, supporting AI enterprises, and cultivating human capital could have long-term dividends. At the same time, as the Party seeks to rely on the private sector to spur innovation, the successes of major Chinese tech companies might be seen as a threat to its monopoly on power, necessitating stronger assertion of control, as perhaps seen in the trend towards the creation of Party committees within tech companies. Despite efforts to incorporate business leaders into the Party, there could be further frictions between CCP leadership and China’s new entrepreneurial class. Ultimately, it remains to be seen whether the Party-State will be capable of managing the cross-cutting changes resulting from the societal, economic, and political transformations affected by the AI revolution.
Image Source: South China Morning Post