China's emergence as a global digital technology leader is arguably the most important factor in the increasing acrimony and levels of competition with the United States. It is also a matter of considerable importance for the Party's domestic programme, both for the sake of finding new sources of economic prosperity and for enhancing the party-state's ability to govern. Artificial intelligence (AI) lies at the core of both these domestic and international trends. At home, a raft of policy documents outline a vision in which AI can be used to increase domestic productivity and efficiency, provide better healthcare and education, but also keep closer tabs on the population's activities. Abroad, those documents speak of the Chinese ambitions to become a worldwide AI leader. Certainly, many in the US, too, believe they are locked in an AI race with China.
Yet if this is a race, it is only just getting under way. While the applications of AI technologies are certainly burgeoning, many of them are merely embryonic or demonstrate much need to grow before they can be reliably used. In other words, we are probably merely at the dawn of an era in which the development of AI technology, as well as the breadth and efficacy of its applications has a long way to go. For China scholars, this means AI may become a factor that cannot be ignored of the social, economic or political phenomena they study. For students of international relations, it will be necessary to gain a detailed understanding of how Chinese AI policy and actual technical capabilities and processes evolve and interact, before gauging their impact elsewhere in the world. In short, the emergence of AI is an empirical fact that will probably necessitate theoretical and methodological adaptation.
Zeng Jinghan's slim volume is a pioneering contribution to this new debate. It argues that understanding the “Chinese characteristics” with which Chinese-developed AI will be endowed requires engagement with three factors: the way national AI strategy is formed and executed; the process of securitization of AI-related questions; and the role that AI plays in facilitating and supporting the surveillance practices required to maintain regime integrity and more broadly, the socio-economic applications of AI intended to increase regime legitimacy. The three substantive chapters flesh out these points in greater depth.
On the first point, Zeng argues that China's AI strategy is primarily a bottom-up, decentralized process, where local governments take cues from general central instructions and develop their own strategies, in this case in collaboration with the private sector businesses, research institutions and technical experts that drive AI forward. This argument forms a response to the widespread international assumption that China's AI policy is a centralized and top-down affair. Moreover, Zeng argues this strategy is predominantly aimed at economic prosperity rather than geopolitical power. In making this point, which is argued convincingly, Zeng is perhaps a bit too tempted to present the choice between bottom-up and top-down as an either/or binary. Often, large-scale policy developments tend to take the form of “call and response” between the centre and the localities, when considered across time. AI policy only emerged as a top political priority in the second half of the 2010s. Moreover, current indications are that digital policy in general is becoming more centralized. It thus remains to be seen how future-proof this particular argument will be.
On the second point, Zeng claims the central leadership has “securitized” AI, where AI is mobilized as a security concern to engender popular support. This securitization bid involves invoking both historical traumas, such as the century of humiliation, and current-day anxieties about growing great power competition with the US. However, this process of securitization has triggered mirroring effects abroad (where the US and its allies have increasingly come to securitize the previously trade-dominated relationship with China), leading to growing constraints on private business, as investors, entrepreneurs and talents are discouraged from entering Chinese AI activities.
On the third point, Zeng proposes that the leadership sees AI as part of a broad-spectrum set of measures to digitize public services, enhance surveillance and stability maintenance practices, and generate economic growth opportunities. These factors are often at odds which each other, requiring Beijing to carefully balance the coercive, legitimacy-building and economy-growing components of AI. Here, Zeng could have perhaps devoted a little bit more attention to the role of private companies, who generate and hold the vast majority of data on Chinese citizens that can be used in AI applications and to train AI algorithms. Central authorities have faced significant difficulties in getting access to those data in the past, and it remains to be seen whether AI can realize the leadership's towering expectations at a technical level, even before any discussion on politics or policy.
Readers of this book will look in vain for a thorough discussion of these technical considerations, or for very detailed case studies of AI applications in various practical settings – this is clearly not what the book sets out to do. Instead, its goal – which it meets successfully – is to establish a fundamental set of parameters and paradoxes to study, in order to structure the major academic debate on AI in China that is sure to flourish in the coming years.
