The global race to regulate Artificial Intelligence

*By Anderson Rohe

There are currently three currents for an eventual outcome of the global race for leadership in Artificial Intelligence. The first follows the prospective view that China will soon surpass the United States of America – USA in some of its applications, given China's potential to become a leader in the sector. The second positions the USA in the same place of supremacy it is in, as it questions how close this Chinese leadership is, since the country encounters external resistance, as well as internal limitations and historical specificities that still keep it behind the USA. The third recognizes that the European Union – EU runs in parallel, trying to make up for lost time in extensive debates, although its pioneering spirit has lost its former strength and is the target of criticism from its competitors. The objective, therefore, is to assess whether China and the EU have effective conditions to surpass the USA as a superpower in Artificial Intelligence. And, secondarily, to determine advantages and disadvantages between these, using as a methodology the contrast between the three main models in dispute. And the expected result is the confirmation of the hypothesis that China is a worthy competitor capable of, at least, (co)leading this competition.

US, China and European Union lead the AI regulatory race

The regulatory models of the USA, China and the European Union (EU) present themselves today as the three main proposals for the governance of Artificial Intelligences (AIs)^[1]. Within this dispute, there is a clear Sino-American leadership, making the EU compete on the outside. It turns out that until recently the scenario was not this, as the USA cared less about technological regulation. And it was the Chinese emergency that probably sparked this concern^[2], because in the face of legislative gaps, those who take the lead in regulation also tend to dictate the sector's rules.

Therefore, the current rush and urgency for technological regulation comes not only from the rapid popularization of generative AIs^[3], due to its supposed existential threat to democracy and humanity^[4], but also about the moral panic created by the danger of AIs taking a leading role in China. For many, an excessive ambition attempted at any cost^[5], due to the Chinese goal of leading Artificial Intelligence by 2030 (Made in China 2025)^[6], thus surpassing the isolated protagonism of the USA.

In fact, the thesis of Chinese protagonism does not come by chance. In How China became an AI superpower, André Gualtieri ^[1], an ethicist and lawyer specializing in AI governance, attributes this turnaround to the fact that “China's mindset has completely changed”, causing this change to occur due to a series of factors, such as: government policies and institutional reforms since 1978 to support a culture of Chinese entrepreneurship; opening to trade and foreign investment in the 1970s; its unique internet ecosystem; public policies aimed at AI; a digital economy model that collects data in greater quantity and quality than in the West.

However, it is still early to say that China will be elevated to the first level of global governance thanks to its leadership in AIs, as there is a transition underway. That is, the progressive rise of an emerging power (China) supplanting another that is still hegemonic (USA). Slow and complex process, whose layers or dimensions are not only technological, but also economic, marketing, competitive, cultural, cognitive (mastery over language), and above all (geo)political ^[2].

Gualtieri^[3] also believes that the dispute between China and the USA goes beyond the issue of AIs, and can be seen in three versions: a) China will in fact take the USA's place as the world's superpower. First for economic reasons and then due to the increase in military power (hard power) and of greater global influence (soft power); b) the world will be divided between the USA and China as the greatest powers (new Cold War?); c) the USA remains a superpower due to innovation, technology and trial and error in the business field.

Based on such scenarios, writer and computer scientist Kai Fu-Lee^[4] reinforces that, despite there being other competitive models, such as that of the United Kingdom and Canada, among the regulatory proposals, only three stand out today frameworks AI regulatory (models). This is evidenced by the most recent initiatives in the sector. Among these: the European ones (Artificial Intelligence Act, Digital Services Act and Digital Markets Act); the American one (Executive Order on Safe, Secure and Trustworthy Artificial Intelligence – EO) and the Chinese ones (Cyberspace Administration of China – CAC, the Internet Information Service Algorithmic Recommendation Management Provisions and The Shenzhen AI Regulation)^{[5] [6]}.

For now, what can be done as a study or prospective research is to bring qualitative and quantitative data that corroborate Kai Fu-Lee's thesis. That is, there is indeed a monopolistic concentration of the market and power of AIs by China and the USA. Like, for example, the fact that “seven Sino-American super-platforms alone hold two-thirds of the total value of the digital market in the world”^[1]. And that “these powers are leading digital advances, alone controlling more than 90% of the market capitalization value of the 70 largest online companies in the world”^[2]. Argument that is supported by American futurist Amy Webb ^[3], in his book The Big Nine: how the tech titans and their thinking machines could warp humanity, with six of the big techs are from the USA (Microsoft, IBM, Apple, Amazon, Google, and Facebook, currently Meta) and three are from China (Tencent, Baidu and Alibaba).

Comparative (dis)advantages between the regulatory models in dispute

a) American model:

It is a decentralized governance model. For now there is no National Strategy on Artificial Intelligence (which aligns with the fragmented form of government between states and federal agencies), but rather an executive order^[4]; although the current Biden-Harris Administration brings eight voluntary commitments to Big Techs^[5] (therefore, no guarantee of effectiveness). We do not think well in terms of public policy (but in a transversal way, as, as a rule, the emphasis is on the State not being strong). Sector-based^[6], according to each application given to AIs (in the area of health, education, banking, for example). Criticized for being overly pro-market^[7].

b) Chinese model:

By way of comparison, the benchmarking Chinese (with centralized governance) should not be discarded, as it at least serves as a counterpoint to that of the USA and the EU (in the sense of knowing whether or not there is a guarantee of effectiveness). Since this tends to be more effective than the other two due to its more imposing form of governance (for others, draconian, oppressive, authoritarian). It is thought, then, in terms of public policy, and there is a way to enforce its compliance (enforcement) by the strong hand of the State.

c) European model

The European regulatory model runs in parallel, although the European Union – EU took the lead in proposing regulation even before China and the USA. Discussions began in 2018 ^[1], however, European pioneering has lost its former strength and has been criticized by its competitors; mainly regarding the rigor of a pre-classification of the risk of AIs that would supposedly stifle entrepreneurship, free enterprise and innovation^[2].

It is a decentralized governance model. There is no National Strategy as such, as it is an economic-political bloc that only provides recommendations (directives) to its Member States. Therefore, there is no guarantee of effectiveness and/or applicability in practice. This means that people think more in terms of public policy (foreign policy as a genre of public policy) than the American model, e.g. e.g., however, it does not have the enforcement to force National States to internalize the proposed model, and rather try to negotiate with them.

O framework European would then be a middle ground between the other two models, although it runs the risk of being overshadowed by the Chinese-American protagonism. The European fear of being left behind lies, in fact, in the danger of becoming a mere supporting player in this clash, also becoming a victim of “data colonialism” by China and the USA, to the point of losing its digital sovereignty^[3]. Which justifies his initiative in starting regulatory debates about AIs, even as a precaution.

Final considerations

In the end, it is clear that the European Union, as an alternative to China and the USA, is today trying to recover its initial prestige^[4]. As for the expected results of the research, the confirmation of the hypothesis that China already has the potential to lead, or at least co-lead the regulatory race with the USA, is partially refuted. Due to factors even extrinsic to AIs. Therefore, there is no confirmation of Kai Fu-Lee's thesis that China will soon overtake the US in global AI leadership. There is more of a desire for the author than an actual opportunity.

*Anderson Röhe is a Fellowship Researcher at the Artificial Intelligence GT at Think Tank ABES. The opinions expressed in this article do not necessarily reflect the positions of the Association

^[1] SOARES, Matheus. China, USA and European Union lead the regulatory race. Available at: https://desinformante.com.br/riedade-regulatoria/.

^[2] SOARES, Matheus. Op. Cit.

^[3] JUNQUILLO, Tainá Aguiar. A different and historic 'sextou' for the regulation of artificial intelligence. Available at: https://www.jota.info/opiniao-e-analise/artigos/um-sextou-different-e-historico-para-a-regulacao-da-inteligencia-artifi cial-11122023.

^[4] REUTERS. Rapid proliferation of artificial intelligence is a threat to democracy, experts say. Available at: https://noticias.r7.com/tecnologia-e-ciencia/rapida-proliferacao-de-ia-e-ameaca-a-democracia-dizem-especialistas-09112023.

^[5] ZHANG, Angela Huyue. China's Short-Sighted AI Regulation. Available at: https://www.project-syndicate.org/commentary/risks-of-beijing-internet-court-ruling-allowing-copyright-of-ai-generated-content-by-angela-huyue-zhang-2023 -12?barrier=accesspaylog.

^[6] CBS News. China announces goal of leadership in artificial intelligence by 2030. Available at: https://www.cbsnews.com/news/china-announces-goal-of-leadership-in-artificial-intelligence-by-2030/

^[7] GUALTIERI, André. How China Became an AI Superpower: The factors behind Chinese leadership in AI. Available at: https://andregualtieri.substack.com/p/como-a-china-se-tornou-uma-superpotencia?

^[8] CORTIZ, Diogo. Should research into artificial intelligence be temporarily halted? NO. Available at: https://www1.folha.uol.com.br/opiniao/2023/04/as-pesquisas-com-inteligencia-artificial-devem-ser-paralisadas-temporaria-nao.shtml.

^[9] GUALTIERI, André. Artificial intelligence is the continuation of politics by other means: the dispute between China and the United States. Available at: https://andregualtieri.substack.com/p/a-inteligencia-artificial-ea-continuacao?

^[10] LEE, Kai-Fu. Artificial Intelligence (English edition): how robots are changing the world, the way we love, relate, work and live. Ebook. Translated by Marcelo Barbão from AI Superpowers. 1st ed. Rio de Janeiro: Globo Livros, 2019.

^[11] EUROPEAN COMMISSION. Regulation of the European Parliament and of the Council Establishing Harmonized Rules on Artificial Intelligence (Artificial Intelligence Regulation) e Altera Determinados Atos Legislativos da União. Disponível em: https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=celex%3A52021PC0206&print=true.

______. The Digital Services Act package. Shaping Europe's digital future. Available at: https://digital-strategy.ec.europa.eu/en/policies/digital-services-act-package.

^[12] LAWGORITHM. Brazilian Responsible AI Forum. Available at: https://www.lawgorithm.org.br/forum-brasileiro-de-ia-responsavel.

^[13] WENTZEL, Marina. How the global race for data processing can 'colonize' Brazil and other countries? BBC News Brasil, 13 Oct. 2019. Available at: https://www.bbc.com/portuguese/internacional-49981458.

^[14] WENTZEL, Marina. Op. Cit.

^[15] WEBB, Amy. The Big Nine: how the tech titans and their thinking machines could warp humanity. Public Affairs; first edition, 2019.

^[16] THE WHITE HOUSE. Fact Sheet: President Biden Issues Executive Order on Safe, Secure, and Trustworthy Artificial Intelligence. Available at: https://www.whitehouse.gov/briefing-room/statements-releases/2023/10/30/fact-sheet-president-biden-issues-executive-order-on-safe-secure-and-trustworthy -artificial-intelligence/.

^[17] THE WHITE HOUSE. Fact Sheet: Biden-⁠Harris Administration Secures Voluntary Commitments from Leading Artificial Intelligence Companies to Manage the Risks Posed by AI. Available at: https://www.whitehouse.gov/briefing-room/statements-releases/2023/07/21/fact-sheet-biden-harris-administration-secures-voluntary-commitments-from-leading-artificial-intelligence -companies-to-manage-the-risks-posed-by-ai/

^[18] BUTCHER, Isabel. Op. Cit.

^[19] KAUFMAN, Dora. US contributions to AI regulation. Available at: https://epocanegocios.globo.com/colunas/iagora/coluna/2023/07/as-contribuicoes-dos-eua-na-regulamentacao-da-ia.ghtml.

^[20] CORTIZ, Diogo. What is the AI Act? Available at: https://www.linkedin.com/posts/diogocortiz_tecnologia-inovaaexaeto-inteligenciartificial-activity-7139316954720378880-jM-w/?

^[21] THE WHITE HOUSE. Fact sheet. Op. Cit.

^[22] MELLO, Patrícia Campos. Global South needs to act quickly to avoid losing sovereignty over its data, says UN envoy. Available at: https://www1.folha.uol.com.br/tec/2023/11/sul-global-precisa-agir-rapido-para-nao-perder-a-soberania-sobre-seus-dados-diz -sent-from-un.shtml.

^[23] BRADFORD, Anu. The Brussels Effect: how the European Union rules the world. Oxford: Oxford University Press, 2020.

Notice: The opinion presented in this article is the responsibility of its author and not of ABES - Brazilian Association of Software Companies