In a move towards greater transparency and better understanding of China’s role in global AI development, AidData and the RAND Corporation today released a new report, “China’s AI Exports: Technology Distribution and Data Safety.” It is accompanied by an underlying dataset and web tool and technical documentation. This groundbreaking effort sheds light on AI export projects funded by China’s official sector to low- and middle-income countries around the world, offering a more comprehensive understanding than ever before of Beijing’s role in shaping the global AI landscape.
This study is the first of its kind to combine quantitative and qualitative data to examine China’s AI technology exports. The dataset was developed using advanced data mining tools to identify projects that use or enable AI exports. The authors—Jennifer Bouey, Lynn Hu, Keller Scholl, William Marcellino, and Rafiq Dossani of RAND and Ammar A. Malik, Kyra Solomon, Sheng Zhang, and Andy Shufer of AidData—analyze the quantitative dataset, as well as qualitative country case studies based on subject-matter expert interviews and sentiment analysis, to examine the distribution, technology, financing, and data safety aspects of China’s AI exports.
Join AidData and RAND for an engaging event on Dec 12th, 2023 at 10:30am EST on “China’s AI Exports: Implications for Asia and Africa.” This public webinar will convene a multidisciplinary group of experts to surface insights from a groundbreaking new dataset and report on China’s AI export projects with official development financing. Register here.
Key features of the new dataset
The China’s AI Exports Database (CAIED) covers projects from 2000 to 2017, sourced from AidData’s Global Chinese Development Finance (GCDF) Dataset, Version 2.0.
- CAIED provides an unprecedented analysis of 155 AI projects funded by Chinese official sector institutions over an 18-year period. These projects span 51 countries across Asia, Africa, Eastern Europe, the Caribbean, and Latin America and showcase China’s increasing commitment to global AI dominance.
- CAIED categorizes projects into two groups: (1) AI Applications and (2) AI Infrastructure. This classification allows for a nuanced understanding of how China not only deploys AI technologies but also builds the necessary infrastructure to support their implementation.
- CAIED features detailed information on projects funded by Huawei and the Chinese military, offering a more comprehensive perspective on the various sources of funding behind the country’s AI export activities.
AidData released an updated version of the GCDF dataset in November 2023, which will allow for future analysis of China’s AI exports across an extended time period.
Top insights from the analysis
- The study identifies a significant surge in official sector AI exports from China, with a five-fold increase from 2005 to 2011 and a doubling again between 2012 and 2017. The initial rapid growth can be attributed to China’s official sector financed AI infrastructure projects such as data centers and fiber optical cables that dominated the market. By contrast, later years saw an increasing reliance on AI applications like facial and speech recognition and algorithms for medical diagnoses, and Chinese official sector projects experienced relatively slower growth.
- The dataset and report highlight three key factors driving China’s progression in AI exports to the developing world: (1) an integrated approach to technology exports, (2) competitive affordability, and (3) flexibility in operations.
- China’s AI exports are concentrated in low- and middle-income countries, particularly those participating in the Belt and Road Initiative across Asia, Africa, and Latin America. Most of these countries are not liberal democracies, but rather electoral democracies, electoral autocracies, or closed autocracies. An analysis of sectoral allocation reveals a focus on “Safe City” or “Smart City” projects, e-government initiatives, communications projects, and medical imaging technologies.
- Most recipient countries lacked robust data protection laws as of the end of 2017, emphasizing the need for increased attention to data safety measures in AI projects. However, compared with non-AI projects funded by China, a much larger share of Chinese AI exports were directed to countries that have robust data protection laws. This could mean that countries that have more experience with using AI are both (1) more likely to adopt data protection laws and (2) more likely to import AI technology from China, as well as other countries.
- The report and dataset reveal different perspectives on China’s AI exports among administrators, medical professionals, and the public in recipient countries. While administrators and medical professionals view AI exports to their countries positively, public sentiment generally tends to be more negative, driven by concerns over their governments’ lack of capability for effectively using AI technologies while ensuring data safety.
Supplemented by in-depth interviews in two of the recipient countries, Kenya and Pakistan, the report offers the following recommendations:
- Policymakers and civil society in recipient countries should work to ensure transparency regarding AI imports, especially for projects involving governance systems and communication technologies. This includes outlining AI procurement policies, and providing public information on project outcomes and impacts. Donor countries and agencies (including the United States) can also support recipient countries’ governments and civil societies in transparency efforts.
- Recipient government regulatory branches should aim to build data safety policies and implementation processes, develop customized data governance frameworks. Donors should conduct thorough assessments of existing policies on data and AI in recipient countries and work with public officials to establish robust monitoring and evaluation systems.
- Donor countries should help build local AI literacy and prepare ancillary support for the AI application ecosystem. This includes providing youth education around AI; fostering native AI talent; supporting local tech entrepreneurship; and creating a conducive regulatory environment for AI innovation.
The CAIED’s launch comes at a pivotal moment, as China increasingly asserts itself as a global technology powerhouse. The dataset and report address pressing questions related to the financing, technology, and data safety of Chinese AI exports, providing valuable insights for policymakers, researchers, and international organizations. As AI exports from China continue, understanding their on-the-ground impacts will be crucial for informed decision-making and responsible governance.
For media inquiries, please contact Alex Wooley, firstname.lastname@example.org; +1 757-585-9875.
AidData is a research lab at William & Mary’s Global Research Institute that provides data, analysis, and innovation to make development finance more transparent, accountable, and effective.
About the RAND Corporation
The RAND Corporation is a research organization that develops solutions to public policy challenges to help make communities throughout the world safer and more secure, healthier, and more prosperous.