Leading the way

After decades of rapid growth and progress fuelled by foreign investments and know-how, China is now entering a new phase of development characterised by what Wired’s reporter called as a move from ‘imitation to innovation’ (Larson, 2018).

In July 2017 State Council has announced its ambition to make China a leader in artificial intelligence (AI) and creating a 150 billion dollar AI industry by 2030, which according to some experts is a somewhat understated estimate (Mozur, 2017; Yang & Yang, 2018). The ‘New Generation Artificial Intelligence Development Plan’ is seen as “a key impetus for economic transformation” aimed to build China’s advantage in the development of AI and to establish China as an innovative nation and global power in science and technology (Mozur, 2017). Much of this development focuses on enhancing domestic security and building a vast surveillance assemblage to monitor and control the country’s population.

As part of its push for innovation, China has chosen over 300 cities for its national smart city pilot (Li et al., 2015; Wan et al., 2015). The nation views smart city as a “key strategy to promote industrialization, informatization, and urbanization” (Li, Lin, & Geertman, 2015, p. 291). This paper argues that the intensifying surveillance capabilities, facilitated by the new technological practices and AI-enabled networked infrastructures, are often deployed under the guise of urban innovation through the use of ICT and the smart city narrative (Greenfield, 2013; Zuboff, 2015). By using surveillance as a conceptual lens for analysing adoption of emerging technologies in China, this paper aims to shed light on some of the more worrisome aspects of the widely promoted smart city narrative.

China’s ‘Sharp Eyes’

In 2015, the Chinese Ministry of Public Security, national police force, and other agencies called for the creation of an “omnipresent, completely connected, always on and fully controllable nationwide video-surveillance network as a public-safety imperative” (Denyer, 2018; Lin, & Chin, 2017; National Reform and Development Commission, 2015). There are currently an estimated 100 million CCTV cameras installed throughout China (Yu, 2017). By comparison, the U.S. has about 50 million (Lin, & Chin, 2017). The number of CCTV cameras is estimated to grow to over 600 million by 2020 as part of the 13th Five Year Plan that requires 100 percent surveillance and facial-recognition coverage and total unification of the existing databases across the country (Long, 2018; Vincent, 2018)

To achieve this, an ambitious plan known as ‘Xue Liang’, which can be translated as ‘Sharp Eyes’, has been established as part of China’s AI program. The plan’s name takes inspiration from the ruling Chinese Communist Party’s slogan “the people have sharp eyes” which is based on Mao Zedong’s attempt to get every citizen spying on one another (Denyer, 2018). The plan aims to connect public security cameras with private cameras and those installed in smart devices in homes and integrate them into nation-wide surveillance and data mining platform that enables real-time viewing and sophisticated video and image analyses by authorities (Denyer, 2018; Long 2018). An essential component of the infrastructure is an AI-enabled facial-recognition system that is capable of instant identification of any citizen.

Such an all-encompassing undertaking necessitates the collection of vast amounts of data which is enabled by a totalitarian regime and revolving doors between the private and public sector (Lin & Chin, 2017; Wade, 2018). China, unfettered by privacy concerns or public debate, has access to immense amounts of data - photos uploaded by country’s almost 800 million internet users and a centralised image database of citizens, all of which must have a government-issued photo ID by the age of 16 (Lin, & Chin, 2017).

This agenda is spearheaded by a trio of established internet giants Baidu, Alibaba and Tencent (sometimes called ‘BAT’) which are leading the way in the development of AI, self-driving cars and other smart city technologies. These companies are competing to market and develop surveillance systems for government use and provide Chine’s authorities with access to unprecedented amounts fine-grained data on minute details of people’s lives generated through mobile phones and sensors embedded in the various Internet of Things (IoT) devices (Lin, & Chin, 2017). The business ecosystem that sustains China’s booming surveillance infrastructure is further supported by a thriving AI startup sector where companies openly boast that “their best customers are local police bureaus” (Lin, & Chin, 2017).

The next section illustrates the pragmatic aspects of the Smart Eyes plan in the context of China’s capital.

Smart Beijing

Beijing, the China’s ‘smartest city’ (Xinhua, 2017), has achieved 100 percentage video surveillance coverage of the city as part of the preparations for the National Day holiday in 2015 (Yin, 2018). Installation of smart cameras constitutes a crucial aspect of Beijing’s smart city plan. Cameras are already being used for routine activities such as gaining entrance to a workplace, withdrawing cash from an ATM, security checks or for checking in into a hotel (Denyer, 2018; Lin, & Chin, 2017). All of the footage and data is centralised and accessible by the government through an ‘intelligent operation centre’ and can be used to track suspects, spot suspicious behaviour, predict crime and coordinate the work of emergency services (Denyer, 2018).

In May 2017, a facial-recognition system was used at Belt and Road Forum hosted by President Xi Jinping in Beijing to promote old Silk Road trade routes. Wall Street Journal reporters were present when paramilitary police, stationed next to face-detecting consoles at entrances to the event, showcased how the system instantly pulled up names, photos and profiles of people approaching the screening area, verifying them as guests (Lin, & Chin, 2017).

The facial-recognition systems work by breaking down footage of a face into a series of measurements and using deep learning algorithms to assess face’s characteristics and to generate a template that can be compared with others in a database. Many developers of these systems are pitching them as an alternative to keys, credit cards and ID cards (Lin, & Chin, 2017). Besides the claimed security benefits, these systems are also promoted commercially as means of providing convenient and smooth customer experience. For example, a KFC restaurant in Beijing is “scanning customer faces, then making menu suggestions based on gender and age estimates” (Lin, & Chin, 2017).

SenseTime’s showroom in Beijing offers a glimpse of capabilities of such system which can estimate “visitors […] age, gender, mood, attractiveness and closest celebrity resemblance, while also serving up ads based on those characteristics” (Lin, & Chin, 2017). Another product offered by the company can use camera networks to track a person’s movement around a neighbourhood. A system by a Intellifusion can track an “individual’s movements inside a building through facial recognition and alert authorities if that person attempts to access restricted floors” (Lin, & Chin, 2017). Li Xiafeng, director of research and development at Cloudwalk, a company specialising in facial-recognition, says that the overarching aim is “to track routine movement, and after you get this information, to investigate problematic behavior” (as cited in Denyer, 2018).

Figure 1. - A CCTV display using the facial-recognition system Face in Beijing. Reprinted from Denyer, S. (2018).
Figure 1. - A CCTV display using the facial-recognition system Face in Beijing. Reprinted from Denyer, S. (2018).

Full Spectrum Biometric Control

According to Zenz & Leibold (2017), the “combination of low-skilled foot-soldiers stationed in and around convenience police stations and high-tech equipment connected to extensive information processing systems has dramatically increased the Party-state’s surveillance capabilities, providing what local media claims is ‘complete coverage without any chinks, blind spots, or blank spaces’” (p. 26). The Sharp Eyes plan being implemented across the country at the moment is just one the foundational components of China’s massive effort to build an omnipresent AI-enabled surveillance apparatus.

The Sharp Eyes system is developed alongside a ‘social credit’ system that will, according to a document released by State Council “allow the trustworthy to roam everywhere under heaven while making it hard for the discredited to take a single step” (Mistreanu, 2018). The system is planned to be implemented nation-wide by 2020 with an aim to create an ecosystem “made up of schemes of various sizes and reaches, run by cities, government ministries, online payment providers, down to neighborhoods, libraries, and businesses” (Mistreanu, 2018). Under the scheme, currently being tested in the pilot city of Rongcheng, people who are considered troublemakers – such as those who have smoked on public transport, tried fare-dodging, ’spread false information’ online, or are otherwise considered a security threat or unfavourable to the ruling Communist Party – will be prevented from buying tickets for a plane or a high-speed train. By contrast, the ‘well-performing citizens’ will receive perks such as discounts on heating bills and favourable bank loans (Mistreanu, 2018; Long 2018).

Another major project labelled ‘Police Cloud’ aims to “scoop up such data as criminal and medical records, travel bookings, online purchase and […] social media comments — and link it to everyone’s identity card and face” (Denyer, 2018). This system is designed to uncover relationships between events and people ‘hidden’ to the police by enabling longitudinal analysis of spatial and behavioural patterns of individuals – tracking where have they been, who are they with, what have they been doing, as well make predictions about their future activities (Human Rights Watch, 2017a). Additionally, the government is collecting voice pattern samples of individuals with an aim to establish a national voice biometric system which will be able to identify voices in phone conversations (Human Rights Watch, 2017b). There have also been reports that Chinese authorities have been collecting DNA samples, iris scans, fingerprints and blood types in an attempt to build a comprehensive biometric database of all residents of Xinjiang – a place described as a “fortress city with technologies” that serves as a testbed for some of the most invasive and repressive smart city technologies yet (Haas, 2017; Phillips, 2018; Wade, 2018; Wang, 2017).

The practices and modalities described thus far resemble a scenario from an Orwellian science fiction novel. However, the ‘surveillance as a mode of living’ constitutes a day-to-day reality for a growing number citizens in China where people are already being identified and detained based on the systems described in this paper (Denyer, 2018; Tucker, 2012). With greater surveillance being accepted as a cost for the increased sense of security and more convenient services, a growing number of everyday activities are becoming entirely dependent on code for their socio-spatial production (Kitchin & Dodge, 2011; Webster, 2012).

China envisions a smart city designed around a spatially distinct individual and multi-modal biometric portraits of its citizens that provide, in combination with a totalitarian regime and utter disregard for privacy, a level of governmental oversight unprecedented in human history (Human Rights Watch, 2017b; Tucker, 2012).

The next section looks at these technological developments in China in the broader context of a global smart city discourse.

Global smart surveillance assemblage

China’s growing surveillance apparatus is invasive and totalitarian to the degree that it may seem unlikely that such an approach to smart cities might establish itself in Western democracies. However, with the rapid advances in capabilities of AI, rise of predictive policing, and the revelations about the extent of NSA’s and GCHQ’s dataveillance in the recent years, it is likely that similar activities are already taking place in western countries - they have just not yet been codified into laws and policies that would allow them to be publicly acknowledged and enforced (Greenwald & MacAskill, 2013; Hopkins, 2013; O’Neil, 2016; Webster, 2012; Zuboff, 2015).

China’s modus operandi for the development of its smart cities sets unsettling precedence for a panoptical ‘all-seeing’ model of disciplinary power in which the surveilled bodies become self-regulating conformist bodies (Foucault, 1991/1975; Han, 2017; Tucker, 2012). China’s push for total surveillance can be viewed from the perspective of a dominant neoliberal discursive regime that praises the use of ICT to improve “safety, security, efficiency, antifraud, empowerment, productivity, reliability, flexibility, economic rationality, and competitive advantage” (Kitchin & Dodge, 2011, p. 19). In this discourse, the inherently political decisions about the future of cities are presented as value-free and neutral through techno-centric rhetoric and a heavily-marketed ideal of the interconnected smart city (Graham & Marvin, 2001; Greenfield, 2013).

Despite sustained criticism, much of the prevailing smart city discourse is shaped by economic interests of global ICT infrastructure monopolies that are promoting solutions based on a technology that is explicitly about observing, monitoring and shaping behaviour (Graham, 2005; Greenfield, 2013; Webster, 2012). These novel AI-enabled technological assemblages are part of the surveillance capitalism ideology that “is constituted by unexpected and often illegible mechanisms of extraction, commodification, and control that effectively exile persons from their own behavior while producing new markets of behavioral prediction and modification” (Zuboff, 2015, p. 75; see also Han, 2017). Already firmly established in the production of digital spaces, this pervasive ideology is gradually permeating spatialities of everyday life that is increasingly “dominated by the cultures and biopolitics of (attempted) separation, a vast infrastructure of continuous, digital filtering and tracking, and the cultural politics of fear” (Graham, 2005, p. 576).

The global smart city agenda enacted through ‘winner-take-all’ mechanisms of surveillance capitalism is characterised by significant information asymmetries and unequal power dynamics that are poised to widen social and economic inequalities further. Due to the ubiquitous and all-encompassing nature of computing, this agenda cannot be analysed in isolation at a level of individual cities. Instead, the smart city needs to viewed as a process of ever-present urbanisation in an all-pervasive, invisible global metropolis that extends urban modalities of living to everyone, whether they live in a city or not.

Conclusion

This paper has presented China’s smart city vision with a distinct focus on the country’s pursuit of total surveillance through an expansive AI-enabled infrastructure. Due to the relentless pace of technological buildout and rapidly evolving socio-economic landscape in the country, it is difficult to conduct a comprehensive analysis without getting bogged down with inconsequential details. Therefore, the author has situated developments in China into a broader context of global surveillance assemblage that is emerging as a constitutive part of smart city initiatives constructed around principles of surveillance capitalism.

In spite of the expanding global surveillance apparatus, the concept of surveillance is still “significantly under-deployed and underdeveloped in mainstream studies of public administration” (Webster, 2012, p. 26). Much of the critical literature seems to be reiterating familiar concepts from surveillance studies using a Foucauldian type of discourse analyses to polemicise about social issues of urban development. This approach is sometimes criticised for its over-reliance on the panoptical model of surveillance which is considered rather deterministic and applies a limited top-down power enaction perspective that fails to account for more nuanced horizontal, networked designs of peer-to-peer surveillance (Tucker, 2012).

There is also a growing body of critical urban studies literature that successfully captures much of the complexities of the contemporary smart city discourse. However, it appears unable to construct a coherent narrative that would sustain alternative visions of the future of urban life that go beyond the idealistic citizen-centric conceptualisations that lack a sufficient economic impetus to merit their adoption within our existing model of neoliberal global governance.

This paper shares some of the analytical pitfalls mentioned in the concluding section. Consequently, future work should focus on analysing possibilities for citizen-driven systemic change and methods for facilitating positive cultural change concerning smart city development.