AI-Powered Propaganda and the CCP’s Plans for Next-Generation “Thought Management”

Introduction

The propaganda apparatus of the People’s Republic of China (PRC) is in full swing: to ward off the negative international repercussions of the COVID-19 pandemic, the PRC’s diplomatic corps and state media are actively asserting Beijing’s perspectives online. For example, in April the Chinese Communist Party (CCP)  International Liaison Department tweeted links to a series of “public prevention” tip videos in 5 languages, seeking to spotlight positive elements in China’s epidemic management strategies (Twitter, April 17). The PRC’s propagandists also aim to redirect anger and blame toward other actors, such as when China’s state television network CGTN and Foreign Ministry spokesman Zhao Lijian (赵立坚) both promoted the theory that the SARS-CoV-2 virus originated in the United States (Twitter, March 12; YouTube, March 17).

However, such attempts to divert public attention and manipulate narratives to protect China’s image are rudimentary compared to the CCP’s latest public relations project: propaganda powered by artificial intelligence (AI). Last year, CCP General Secretary Xi Jinping urged China to “explore the application of AI for news collection, production, [and] dissemination… to comprehensively increase [our] ability to lead [public] opinion” (National Academy of Governance, January 25, 2019). Intended for both domestic and international application, the system envisioned by CCP analysts and communications theorists will identify early warning indicators of social unrest, assist state journalists in producing effective content, and disseminate approved narratives to target audiences. If successful, AI will amplify the Party’s voice and boost its influence over public opinion.

The CCP’s Quest for Next-Generation “Thought Management”

As demonstrated by developments in the past year, the CCP is placing a renewed focus on propaganda and ideological indoctrination throughout Chinese society (China Brief, April 24, 2019; China Brief, December 10, 2019; China Brief, December 31, 2019; China Brief, April 13). A review of Chinese government statements, commentaries carried by government publications and state media, and academic analyses all reveal the trajectory of China’s propaganda and “thought management” (思想管理, sixiang guanli) apparatus, and the intent of the CCP to dominate electronic media in particular. The Communist Party sees the internet as “the frontline in the struggle over [people’s] opinions” (Cyberspace Administration of China, December 17, 2019). The CCP perceives a continuing weakness in its control over the issues that people focus on, and an inability to respond to those issues in a timely and compelling manner—and, thus, impaired influence over public opinion on the internet (Cyberspace Administration of China, December 17, 2019; People’s Daily Online, November 19, 2010).

The plurality of voices and speed at which public opinion evolves online, especially in times of crisis, are at the heart of these challenges (Cyberspace Administration of China, December 29, 2016). The CCP’s communications theorists especially emphasize the need to reform the type (i.e., text, video, etc.) and sophistication of Party content, as well as to improve the speed of its dissemination. Doing so, they argue, will increase the timeliness and efficacy of state propaganda in leading public opinion online (QS Theory, April 21; Cyberspace Administration of China, December 17, 2019; People’s Daily Online, August 15, 2019).

To rectify its perceived inability to control opinions online, the CCP has identified AI as the next weapon it will deploy to this frontline. To this end, in August 2019 the CCP’s Central Propaganda Department (CPD), along with several government ministries, issued a document titled “Guiding Opinions on the Promotion of Deeper Integration of Culture and Technology.” This document calls for exploring “the use of artificial intelligence for news gathering, production, distribution, reception, and feedback; for comprehensively improving [the state’s] ability to guide public opinion; [and] for making personal customization… and intelligent push notification services to serve positive publicity” (Beijing News, August 26, 2019).

Beijing’s approach to next-generation “thought management” will rest on three core pillars: early warning, effective content, and targeted distribution, as further detailed below.

Early Warning

The first goal of China’s AI systems will be to take the pulse of the communities the CCP hopes to influence. In order to create effective propaganda content, authorities and state media must map the issues around which ideologically incorrect thinking exists, and identify imminent crises. Thus, CCP analysts see AI as a means to “continuously monitor websites, forums, blogs, Weibo, print media, WeChat and other information, [to reach a] timely, comprehensive, and accurate understanding of… trends in public opinion and public attitudes and sentiment” (Cyberspace Administration of China, December 29, 2016). AI would build and accurately interpret audience “comprehensive profiles of ideological behavior” (思想行为的整体画像, sixiang xingwei de zhengti huaxiang) from big data, identifying “ideological confusion” (思想困惑, sixiang kunhuo) and supporting the development of “personalized countermeasures” (University of Electronic Science and Technology, April 27, 2018). Researchers hope that natural language processing and machine learning will allow authorities to identify potentially controversial domestic and international stories before unapproved narratives go viral (Cyberspace Administration of China, December 29, 2016; China Social Sciences Net, January 4, 2017).

Effective Content

The second goal of China’s AI systems will be to assist propagandists in generating influential and ideologically correct content. Once AI has helped authorities and state media identify impending unrest, it will improve the quality and production speed of content for managing public opinion. AI will assist content planning, lead identification, data collection, data visualization, writing, and video production (People’s Daily Online, April 18, 2019; Cyberspace Administration of China, August 2, 2018). AI would further assist editors of state media in assessing the impact of their content, in order to further refine their production algorithms (Reference Net, September 16, 2019). Particularly for international audiences, AI would help China’s journalists identify the keywords around a topic of concern so that they can use the correct terms when creating external-facing propaganda, thus maximizing viewership and resonance (China Social Sciences Net, January 4, 2017). Likewise, machine translation will expand the reach of China’s messaging around the world (People’s Daily Online, April 18, 2019).

Targeted Distribution

The third goal of China’s AI systems will be to disseminate content for maximum impact. Armed with a trove of data on their audience’s online behavior, state media officials hope to tailor content distribution to meet personalized needs (that is, supply the right “countermeasures” to “guide” a wayward individual back towards the Party line). AI will selectively push out propaganda based on “interest tags” (兴趣标签, xingqu biaoqian) derived from the individual’s “profile” (People’s Daily Online, April 18, 2019; Reference News, June 18, 2019). AI will allow state media to tailor content based on variables including how long a person spends consuming news, what time of day they are online, the type of content they engage with, and myriad other factors (People’s Daily Online, May 24, 2019; University of Electronic Science and Technology, April 27, 2018).

AI is also intended to support real-time distribution, further increasing the timeliness of propaganda to guide how individuals perceive events (Reference Net, September 16, 2019). Outside of specific points of “ideological confusion,” interest-based dissemination is also possible because CCP propaganda is often not overtly political: narratives are economic, cultural, and social in nature, and authorities often strive to package their lessons in entertainment. [1] Similarly, content for international audiences should highlight the positive aspects of the country’s culture, history, economy, and participation in global affairs (People’s Daily Online, February 22, 2019).

Artificial Intelligence as a New Foundation for Propaganda Management

Attempts to harness AI for propaganda will be grafted onto ongoing attempts to unify messaging across China’s disparate channels of communication, both at home and overseas. Since at least 2014, Xi Jinping has urged the “media fusion” (媒体融合, meiti ronghe) of both traditional and emergent forms of communication in order to strengthen the Party’s ability to provide “public opinion guidance” (舆论引导, yilun yindao) across all channels of dissemination (People’s Daily Online, August 18, 2014; National Academy of Governance, August 24, 2019). In response, China’s news services are actively working to expand their reach and messaging across the “media matrix” (媒体矩阵, meiti juzhen), which comprises newspapers, websites, online interactive and mobile apps, official social media, personal social media, and third-party representations (Xinhua, April 7, 2017; People’s Daily Online, September 3, 2018). [2]

China’s modern propaganda foundation will integrate domestic and international thought management. The CCP seeks to “effectively expand [China’s] positive voice [on the world stage], firmly protecting national interests and the national image” (Cybersecurity Administration of China, December 17, 2019). China’s theorists call on state media to recognize and make use of the fact that internal reports have international impact, and that international discourse can affect citizens at home (People’s Daily Online, December 11, 2019). Under these conditions, China’s government, its diplomats, and media personalities are all taking to international social media. [3] In addition to social media, China’s news outlets have greatly increased their reach into traditional communication channels overseas, such as through paid content in foreign publications and the creation of a new international news service (Global China Television Network, undated).

Finally, AI will require copious amounts of data from which to generate insights. This data will be mined from across the “media matrix,” pulled from public sources like Weibo as well as from opt-in services like WeChat or the mobile apps developed by state media outlets (People’s Daily Online, August 15, 2019; Cyberspace Administration of China, December 29, 2016). Internationally, it is likely (but not explicitly stated) that data will be harvested from news websites, Twitter, Facebook, and other platforms. Other services, such as machine translation, provided overseas both directly and indirectly by state-owned companies, may also be used to collect data that facilitates the Party’s “efforts to shape, manage and control its global operating environment” (Australian Strategic Policy Institute, October 14, 2019).

Of course, the pivot to AI is not without challenges. Collecting, structuring, and integrating information on the scale that China’s researchers propose is daunting, likely requiring advanced computer applications. Thus, these efforts must be seen in the context of the PRC’s broader push to build an industrial base prepared to harness big data (Gov.cn, August 31, 2015). With regards to implementation, relying on AI to disseminate content based on individual interests may detract from editors’ discretion to determine what content is consumed, cause echo chambers in society, or even diminish people’s grasp of ideology by propagating non-political subjects  (People’s Daily Online, May 24, 2019; Orator Net, July 16, 2019; Legal Daily, November 8, 2019). With these and other challenges in mind, the vice president of Xinhua reminded attendees to the 2018 China Internet Media Forum that “humans lead, machines assist” (CCTV, September 6, 2018).

Conclusion

While heavy censorship allows the CCP to prohibit certain topics of discussion, it cannot actively promote the Party’s values. Achieving the latter requires a sustained and responsive online presence. CCP propaganda theorists and communications researchers see AI as a means to establish such a presence. If the challenges are overcome, the PRC’s new AI-powered propaganda will strengthen what Dr. Samantha Hoffman has called China’s “Autonomic Nervous System” for protecting threats to the party-state (MERICS, December 12, 2017). AI will preemptively identify and address emerging crises in public opinion, push out government messaging before unapproved narratives go viral, and disseminate personalized content to individual readers and viewers. As in every communications revolution, propaganda will evolve—and as conceived by the CCP, processors and algorithms will form the foundation of next-generation “thought work.”

Devin Thorne is a Senior Analyst at the Center for Advanced Defense Studies (C4ADS) in Washington, DC. Follow his research on Twitter @D_Thorne.

Notes

[1] Anne-Marie Brady, “Conclusion: The Velvet Fist in the Velvet Glove: Political and Social Control in Contemporary China,” in China’s Thought Management, ed. Anne-Marie Brady (Oxford: Routledge, 2012), 193-194.

[2] In PRC state media’s formulation, the “media matrix” consists of “print media + websites + user applications + official Weibo + independent media + representative operations” (纸媒+网站+客户端+官微+自媒体+代运营).

[3] For a thorough listing of Chinese State-linked Twitter accounts see the Hamilton 2.0 Dashboard (Alliance for Securing Democracy, undated).