By Cai Cuihong is Professor of the Center for American Studies at the Fudan University 19-09-2024

Topic I:

2023 marks a key juncture for artificial intelligence (AI) to be deeply embedded in international relations. From a technical point of view, the rapid development of generative AI not only provides new carriers and capabilities for countries’ foreign exchanges, but also brings new risks and challenges to national security.

In terms of topics, AI has created a series of new agendas for international relations, such as security, ethics and development, which need to be accelerated under the framework of global governance. Formally speaking, the international competition and cooperation in AI are evolving fast and different power relations prompt countries to follow different paths of competition or cooperation, giving rise to various forms of international interaction in AI.

Therefore, comprehensive analyses of AI influence on national power acquisition, holding and expectation, and discussions on the power logic behind AI international competition and cooperation can provide new ideas for resolving the dilemma of international competition, promoting AI international cooperation and improving global governance.

AI Influence on National Power Strategy

As a strategic and holistic emerging technology, AI is developing rapidly, which not only provides new capabilities and means for the country at the technical level, but also brings about profound changes in the international political, economic and security domains. It has become one of the focuses in international relations studies how AI affects the power strategy between countries.

First, from the perspective of power acquisition, the costs and benefits of “punishment”, “reward” and “regulation” have been seriously affected, which has comprehensively changed the country’s strategy of obtaining power.

 The influence of AI on the acquisition of state power is complex and profound. On the one hand, AI technology has the function of digital and intelligent empowerment, which can greatly reduce the cost of state power acquisition. With the help of intelligent technology, the state can perceive and judge the strategic situation in real time and execute power more efficiently.

Since 2018, the United States Intelligence Advanced Research Projects Agency (IARPA) and its affiliated offices have initiated dozens of AI projects, including the Mercury project of automatic political event prediction, which has significantly improved its information superiority and decision-making ability in intelligence. On the other hand, AI technology has created a new field of power confrontation and provided a new carrier for power interaction between countries.

 As a general technology, AI power interaction can generate vital spillover effects and increase power gains in other fields. For example, in the Ukraine crisis, the United States and its scientific and technological enterprises imposed punitive power on Russia through sanctions, blockades, evacuation, etc., which, as pointed out by Thea Kendler, US Assistant Secretary of Commerce “not only made it difficult for Russia to access cutting-edge digital technology, but also made it impossible for its defense industry, military and intelligence departments to obtain most products made in the West, thereby undermining its military strength.” AI is therefore considered as a key, subversive and strategic technology.

AI brings changes in the cost-benefit dynamics of power acquisition, which makes it possible for more countries to use three kinds of power flexibly, namely punishment, reward and regulation. This is a new phenomenon different from traditional geopolitics. First of all, as far as punitive power, which is almost bound to major countries in traditional geopolitics, is concerned, power acquisition often requires strength and control, thereby giving rise to strategic concepts such as “sea control power” and “land control power”.

In the field of AI, countries and even non-state actors without strong military strength can exert influence through cyber-attacks, information warfare or control of data streams and algorithms. The acquisition of this ability does not depend entirely on traditional military strength or economic aggregate, but more on scientific and technological innovation, talent and ingenious application of strategy and tactics. Secondly, with rapid development of AI, the ways for countries to use the means of reward to make others obey and get rewards consequently are increasingly diversified, and the cost is descending fast.

This is because the spread and replication of digital technology will not increase cost, but rather enhance value by engaging more users. Therefore, the traditional economic law of diminishing marginal revenue is not fully applicable to the digital field where AI is located. Judging from the actual power interaction between countries, reward in the form of technical support, data sharing, economic assistance, etc. is almost of zero cost to the countries that obtain power.

This will make countries with relatively backward technology fall into dependence on technical assistance from advanced countries, resulting in new inequalities in international relations. Finally, the regulatory power means in AI is not only more closely combined with punitive and rewarding means, but also has a far-reaching impact.

On the one hand, regulatory power can provide a more “legitimate” implementation framework for punitive and rewarding means through international standards, technical agreements and codes of conduct. For example, since 2019, the United States and Europe have revised the Wassenaar Agreement to include dual-use technologies and emerging technologies in the control list, which provides more possibilities for joint efforts to exercise power through punitive means such as blockade, sanctions and embargo.

On the other hand, different from the traditional field, the regulatory means in the field of AI have a far-reaching influence on power. At present, AI is basically in the stage of technological nascent and development. Once a certain country forms a strong regulation on another, it may exert structural power so that the receiver end of power directly fall behind in the wave of technological development, and is thus difficult to gain the dominant position in the future technological revolution.

 The power imposition can not only shape international technical standards and norms, but also determine the direction and speed of technological development, thus affecting the economic competitiveness and strategic position among countries. Therefore, the application of regulatory means in AI plays an important role in shaping the future international order.

Second, from the perspective of power holding, the gap between potential power and real power in the field of AI has gradually been bridged, thus changing the evaluation method of power resources held by the state. There is a difference between potential power and real power. Realistic power regards power as the point of departure of causality, which illustrates the ability to change others’ intentions, while potential power refers to the possibility of transforming resources into actual power.

 Given the high added value of power resources such as data, algorithms and models of AI and the forward-looking strategic AI evaluation in various countries, potential power is often regarded as an evaluation factor equivalent to real power.

 For example, though there is still a gap in the development level of AI between China and the United States, the US has always been on the alert for China’s market and innovation capabilities, and believed that China can tap into its strength to “damage” the hegemonic dividend that the US is currently in possession of based on its technological advantages. It therefore can be seen that in the field of AI, the evaluation of power relations often includes a wider range of potential resources, which has greater uncertainty and room to maneuver.

Third, from the perspective of power evolution, AI competence level and conflicts of interest between countries are constantly changing, which in turn affects the setting of national strategic goals and power expectations. The formation of power relations has two indispensable elements, one being the gap of competence level, and the other conflict of interest.

In the emerging field of AI, technology breakthrough and uncertainties of technical implication make the relationship between countries more complicated. On the one hand, countries have broad common interests in the macro aspects of AI security, ethical risks and AI development and empowerment. On the other hand, as AI is widely embedded in all levels of national security and social development, there generally exist conflicts of interest in specific governance issues. These two aspects make the power interaction between countries more complex.

In addition, technologically developed countries take the risks in the initial stage and also gain first-hand advantage. At the same time, they are worried that other countries might make use of technological mutation to catch up and surpass them.

They link their technological advantages with national security and interests closely, and thus become more sensitive to changes in power relations. Countries that have not yet established technological advantages are worried about a “winner-takes-all” by the developed countries and being trapped in a passive position. But they also seek to use the uncertainty brought about by AI development for broader national interests and more favorable international status. These two aspects lead to more diversified power expectations of countries in the field of AI.