The rapid advances in AI technology are transforming our world. As the performance of computer hardware improves in all respects and a massive amount of data builds up in the Internet age, AI technology has entered into a period of high-speed development in the second decade of the 21st century. Thanks to the unprecedented progress in algorithm and computing power, AI technology, for the first time, has crossed the threshold of industrialization and is regarded by countries as a key driver of a new round of scientific and technological revolution. 

As an epoch-making scientific and technological achievement, AI is fundamentally different from previous scientific and technological revolutions in human history. Past technological changes were invariably about the tools and methods that mankind employed to transform the world. AI, however, is able to understand people’s inherent needs via analysis and learning of  big data and become a creative partner that is directly involved in man’s activities to transform the world. As human society embraces the age of AI, the relations between man and technology will go through adjustments at the deep level. Accordingly, key factors in the international system will change significantly, including the actors, structure, logic and operational rules. Traditional theories of international relations face severe tests.

I. Paths and ways through which AI technology shapes international relations

Theoretically, AI technology develops in three stages:

First, the stage of Artificial Narrow Intelligence (ANI) or Narrow AI. Some researchers characterize AI technology at this stage as Applied AI. ANI is not a single technology path, but refers to a variety of solutions enabled by the application of AI in a full range of vertical areas. At this stage, AI functions have narrow expertise in particular domains. They can learn through practice to improve their performance. Products have to be developed for different domains. Therefore, it has a prominent utility property. 

Second, the stage of Artificial General Intelligence（AGI）, General AI or Full AI. It means that AI systems that will emerge in the foreseeable future will have a level of intelligence comparable to that of human beings and be able to solve problems encountered by mankind in all dimensions and across domains. AI technology at this stage will be able to pass the Turing test and replace human beings in performing most of the routine jobs. 

Third, the stage of Superintelligence. Superintelligence is a hypothetical agent that may appear in the future as predicted by philosophers. It is defined as “an intelligence that far surpasses the smartest human brain in every aspect such as creativity in science, wisdom and social skills.” Yet due to technological limitations, the basic state of AI at this stage is mostly imaginary and hypothetical and can only be approached from a philosophical perspective. 

The ongoing development of AI technology led by deep learning will eventually take human society to the stage of ANI. The current high-speed development of AI technology started around 2010, a period that is marked by multi-layer neural networks and deep learning. Deep learning algorithm is based on neural networks composed of embedded, multi-layered pattern recognition systems and transforms lower dimensional features into more abstract higher dimensional properties, classifications or features to identify the patterns of data distribution. Such algorithm empowers AI to extract, identify and build systems from massive, complicated sources of information. In areas where there are tasks with clear goals and rich data, deep learning algorithm enables machines to learn new skills, develop effective strategies and produce in a short period of time solutions that surpass human learning capacity.

AI technology is able to solve problems in a specific area through “training” and “learning”. The model designer, however, has no idea whatsoever what special patterns are discovered when a problem is resolved by the model. The system is nothing but a “black box” for mankind. In this sense, General AI, which is capable of solving problems across different sectors, is almost unlikely to emerge in the existing theoretical framework. The US National Science and Technology Council (NSTC) projected that it will take several decades for General AI to become a reality. Therefore, human society will, for a long to time, stay in the age of ANI. We still need to analyze the impact on the existing international system brought by the way technologies diffuse their influence in the age of ANI. 

As Figure 1 shows, based on the current momentum of development, the continuous advances of AI technology in different sectors made possible by deep learning in the age of ANI have an impact on the international system at three levels. 

First, at the sectoral level, some AI technologies can be directly applied in specific domains of international relations issues, thereby changing the balance of power among existing actors in these domains and adding new uncertainties to international relations. As AI technology progresses, prominent, direct changes in international relations issues have taken place in the assistance system for strategic decision making and the intelligent military system. 

Second, at the institutional level, the sectoral proliferation of AI technology will reshape the relationship between capital and labor in the world economic system. Structural adjustment will be underway in the global economic system, driving the evolution of the way power is distributed inside a country, the rapid expansion of capital and the spread of influence of technology communities. As a result, the national governance system will experience important structural adjustments in the context of technological changes. 

Third, at the system level, the influence of AI technology advances will have spillover effects at the system level. Changes in domestic governance structure trigger changes in international governance structure. Changes in socioeconomic conditions and governance thinking will gradually play out across the world. Consequently, subtle changes will take place in international economic and political governance models. Adjustments will also occur in the nature and rules of international relations to adapt to the changing economic foundation. At the same time, the profound changes in the international system also reflect the general direction of changes in social thoughts globally. 

Based on the above analytical framework, we can have informed discussions about the complicated impact on all the elements of international relations brought by technology advances in the age of AI, and try to draw a picture of future international relations. 

II. How AI technology gets involved in strategic and military fields

Currently, the involvement of AI technology in activities of international relations is mainly in the strategic decision making system and the military system. As strategic decision making and military security bear on a country’s fundamental interests, all countries have a positive attitude toward new technology R&D in this field. At the same time, as the two fields have relatively clear targets and rich data, they have all the conditions necessary for deep learning algorithm to play its role.

1. AI technology drives changes in the model of strategic game 

In a mathematical sense, strategic interactions between international actors are typical behaviors of “incomplete information game”. The participants make and enforce the most favorable game tactics by evaluating their competitors’ capability, intention, interests and determination and analyzing the specific external environment.  In a traditional sense, states participate in a strategic game mostly based on the acumen and judgement of statesmen. As deep learning algorithm advances, AI-assisted decision making systems have demonstrated their potential in this area. 

Firstly, AI systems can provide more accurate risk assessment and early warning, making strategic decision making a process of precise selection and putting it on a scientific basis. 

A strategic game is premised and built on accurate prediction and judgement of prospects and risks. Decision making by human beings, however, is a fairly subjective, speculative attempt based on extremely limited intelligence. When deep learning technologies are involved in a strategic game, the severity of risks in strategic behaviors and the effectiveness of responding tactics can be accurately expressed through technologies and in ways of probability. Decision making can be transformed from mere guessing to choosing a tactic from prospects of different probabilities. This will minimize uncertainties in a strategic game. 

Secondly, deep learning algorithm can provide, at a faster pace, strategic options that are different from conventional patterns of human thinking, and continuously raise the effectiveness of strategy delivery in a game. 

As the features of deep learning algorithm dictate, machine intelligence does not imitate the process of human decision making, but proactively establishes new tactics based on a full understanding of targets. These tactics are quite different from man’s patterns of thinking, but they are often more effective in achieving goals in a game. At the same time, another important advantage of deep learning algorithm is that it can learn from mistakes and more accurately grasp the tactical characteristics of the opponents through repeated interactions. As data continues to accumulate, the characteristics and tactical tendencies of the opponents in a game will become clearer and expand the advantages of interactions. 

Finally, in the strategic game process, the AI system can eliminate the interference of other factors to the greatest extent and improve the reliability of strategic decision-making.

AI is an objective and neutral decision-making system that represents a purely rational attitude. The AI decision-making system will only be affected by the parameters, and there will be no impulsive decision-making. Human beings cannot escape the adverse impact of fatigue and emotions on their bodies. But AI does not make any change just because of the extension of time. Comparatively speaking, AI systems are more reliable decision makers and offer ways that are the closest to rational choice.

In general, decision-making assistance systems based on AI technology will further widen the gap in strategic game capabilities between international actors. Actors, without the assistance of AI technology, will be in absolute disadvantage in many ways, including risk assessment, strategy selection, decision making and execution efficiency, and decision reliability. The traditional strategic game structure will be completely out of balance.

2. Changes induced by AI technology in the military field 

The military field is an important platform that leads to breakthroughs in AI technology. The advancement of AI technology will profoundly change weapon systems, military strategies, military organizations and even the meaning of war. As such, human beings will find themselves in a completely different military and security environment after the dawn of the era of ANI. 

Firstly, at the level of military technology, AI technology will promote the wide application of intelligent weapons.

Deep learning algorithm can enable intelligent weapons to acquire the basic ability to control weapons in a virtual environment, and use AI systems to coordinate the action of intelligent weapons to maximize the advantage on battlefields. Intelligent weapon systems can not only achieve a substantial separation between humans and weapons, thus bringing the actual casualty close to zero, but also maximize the efficiency of the use of weapons and cooperation between different weapons. When AI weapon systems meet traditional models of military power on the battlefield, laws and ethics that are applicable to war will be under tremendous shocks.

Secondly, at the level of military strategy, AI technology will bring military strategy and tactics to an unprecedented level.

War itself is a relatively simple and closed field compared with complex human society. Deep learning algorithm has been able to recover almost all battlefield information, and simulate the battlefield in relatively accurate terms on a sand table. On a real-life battlefield, AI has the ability to collect battlefield information more comprehensively and efficiently. Commanders therefore have a more full and accurate picture of the battlefield and get more precise tactical advice.

At the level of military strategy delivery, AI has two advantages that humans can’t match. First, AI systems can quickly process battlefield information and have rapid response capabilities that humans do not have. Second, AI systems have multi-tasking capabilities to pursue multiple actions simultaneously, and propose complex strategies that human thinking patterns cannot understand. With new technologies, the imbalance between conventional military forces will be magnified like never before. Traditional armed forces will find it increasingly difficult to compensate for their disadvantages on the battlefield through tactics and strategies.

Lastly, at the level of military implementation, AI technology advances may lower the threshold of war. 

In the contemporary international system, war is widely regarded as an extreme means in international political behavior. The huge economic costs and domestic political pressure caused by casualties have actually set a high threshold for war. However, with the involvement of AI technology, there is a trend of significant decline in the cost and risk of war.

On the one hand, the involvement of AI technology will be able to effectively reduce the cost of military operations. The use of intelligent weapons can significantly save time and cut labor costs in the training process. The training of unmanned combat weapons relies on relatively mature deep learning algorithms. After the initial training, it can be quickly replicated to all unmanned combat weapons of the same type so as to fast track the training process of combat weapons. From the perspective of long-term effects, it is a more economical and effective combat training method. Due to the replicability of algorithm and data, the wear and tear of some weapons will have less impact on overall operational effectiveness.

On the other hand, the cruelest aspect of the traditional war model is casualty, which is also the most serious political risk of warfare in contemporary society. The widespread use of intelligent weapons actually reduces man’s direct participation in combat. The separation of people from weapons changes a war into a task of using unmanned weapon systems. With the widespread use of intelligent weapons, casualties can be greatly reduced. So are political risks. This situation essentially encourages big countries to exercise less self-restraint and take more aggressive actions to achieve their goals. It also creates new factors of instability on the international security landscape, providing additional impetus for the technology arms race between big powers.

III. The Impact of AI technology on the international system

AI technology has the most direct impact on the economic system. Most notably, technological progress brings shocks to the stable economic structure of contemporary society. The change of economic structure will in turn lead to the change of the power distribution model within international actors, and ultimately transform the system from inside.

1. Change of socioeconomic structure in the age of ANI

It is the core logic of the AI industry to replace human beings in production by machines with similar human intelligence. The increasingly mature AI technology will disrupt the original industrial structure and production model, and change the status of different elements in the economic system. With the full application of AI technology in social production activities, two unacceptable negative phenomena in current human society will be widely seen during the period of transition and become the normal state of economic and social structure in the era of ANI:

First, persistent unemployment. 

The core logic of AI technology is to replace human labor with machine automation to improve social productivity. Therefore, the persistent and widespread unemployment will become a constant of the ANI era. Unlike the cyclical unemployment traditionally caused by the decline in production scale, the joblessness caused by AI is essentially structural. Capital eliminates the need for labor in new ways and through new means. The structurally unemployed will not come back to work because of changes in the economic cycle. On the contrary, the application of AI technology in various vertical fields will onlyenlarge the scale of unemployment and  lengthen the duration of unemployment.

Second, yawning gaps in social wealth.

Yawning gaps in social wealth represent another important social phenomenon accompanied by persistent unemployment in the era of ANI, and willtear asunder the traditional social structure.

Firstly, gaps in individual wealth. In the important process whereby capital squeezes labor, productivity gains brought by AI will not translate into universal wage income growth. The newly created social wealth will be tilted to the capital side in a disproportionate way, with low-income groups and people with less education seriously disadvantaged in the new round of social resource allocation.

Secondly, the distribution of wealth across industries will notably diverge. The use of AI technology will bring enormous capital to industries related to the Internet and big data. In the next five to ten years, there will emerge a huge gap between AI-related industries and traditional industries.

Finally, the wealth gap between countries will be wider. There is an uneven development of AI technology among different countries. In the early days of ANI, some countries will be in a monopolistic position over AI technology for some time, and make enormous benefits. This will only aggravates the already yawning gap in global wealth.

The development of AI technology plays a role as a “creative disrupter” in the global socio-economic structure. Under the influence of AI technology, the status of capital and technology in economic activities has been comprehensively improved, while the value of labor has been seriously weakened. Globally, capital and talent will be pulled to developed countries with technological advantages, while developing countries have a limited window of opportunity to embark on the path of modernization. The power of capital will rapidly expand in the political system.

2. The adjustment of national power distribution model in the age of ANI

The distribution of national power is fundamentally determined by the characteristics of social and economic production modes. When the social and economic structural changes driven by AI technology unfold, new social power emerge with technological change, and lead to adjustments in the national governance structure and power distribution model.

First, the power of capital will continue to expand thanks to technological monopoly. The development of the AI industry relies on huge and long-term capital investment. Therefore, major data resources and advanced AI technologies in today’s world are basically monopolized by large multinationals. Command of data and algorithm in itself means power. They can use the black box of algorithm to provide the public with what they want to see, subtly and quietly changing the way public products are offered. The combination of capital and technological monopoly will gradually lead to the sharing of important powers such as finance and information, which have been traditionally controlled by nation-states.

Second, technological power expands in a short period of time by virtue of the asymmetric advantage of knowledge. During the transition to the ANI age, a small group of elite scientists who are responsible for the core tasks of developing AI technology will gain influence that goes beyond conventions. With their own voice and influence, they form a special class of technological power, a scientist community. As the community of top scientists is small during this period and is irreplaceable, such a group is relatively stable. The industrial revolution and the large-scale adjustment of the social structure triggered by AI technology will appear at the same time, making the scientist community itself an intermediate link in the change of technology and social structure. And they have an important say in shaping technological development going forward. 

In the early years of the ANI age, traditional political power, capital power and technological power will form a pyramid structure. Traditional political power will continue to control the core elements of state power such as security, taxation, and social management, and continue to monopolize the legitimate right to use force and endorse the use of other forms of power. As representatives of capital power, large multinationals will accept power transfer in certain areas to provide public goods for the whole society, including data and economic information. The scientist community constitutes a key link between the development of AI technology and social production, providing basic standards and moral influence for the future society, and shouldering the responsibility of promoting further development of AI technology. The three forces support yet check each other, adapt to each other in a state of constant interaction, and try to explore the most reasonable model of state governance in the new stage of social development.

3. The transformation of the international system from the inside

The transformation of the international system in the age of ANI is actually a natural extension of the distribution of domestic power in a wider space. As shown in Figure 3, the capital and technological powers on the receiving end of the transfer of traditional political power will gain the legal status as international actors in the new stage of the international system. Such a model of interaction between actors of this type will be introduced into the international system, making the whole system more complicated, with diverse actors, rules and relationship. 

First, diverse actors. Sovereign states, multinationals and the scientist community across borders will become important participants in the international system of the new era.

In the age of ANI, the status of sovereign states as the most important actors in the international system will remain unchanged. Super multinationals will gradually develop into international actors with commercial interests as the core and certain independence and discourse power. They provide smaller countries with public goods in specific areas and participate in political transactions on other topics. The emergence of the scientist community as an independent force in the international system will play an important role in building global consensus in the age of ANI.

Second, diverse rules. There are significant differences in the rules of conduct followed by different actors in the international system in the new era.

Sovereign states will continue to follow the existing basic rules of the international community and engage in competition over traditional themes such as power and interests. Multinationals take economic interests as their core goal. Because they cannot obtain the legal right to use force, peaceful competition will be the main mode of their activities. The scientist community is distinctly global. The existence of different rules means that many elements in the new system need to go through a long period of adaptation to establish boundaries for stable power.

Third, diverse relationships. Sovereign states and multinationals will carry out complex games around the boundaries of power transfer. Restricting and dissolving the monopoly of the scientist community on technology will become the common political operation of sovereign states and multinationals.

Sovereign states need to take advantage of the technological and commercial success of multinationals to drive their economic development and technological progress, but they should also be alert to the erosion of the core power of sovereign states caused by their development. However, sovereign states and multinationals have full consensus on the political means to limit and break the monopoly of the scientist community on the technology discourse power. Due to the lack of experience in dealing with the monopoly of high-level technological power in the history of mankind, how to use the new mechanism to bring into full play of the positive role of the scientist community in AI and limit the possible abuse of technology power will become the core political operation in the early days of the ANI age.

IV. China’s Strategic Choice

The rapid development of AI technology will change the traditional path of industrialization and eventually lead to a new round of industrial revolution. The advancement of AI technology will not only promote the substantial increase of social productivity, but also have a far-reaching impact on the economic production pattern, power distribution model and structure of the international system of human society. Under the impact of the technology, the differences in power between international actors will be further expanded, and human beings will enter an era of growing inequalities.

In order to effectively cope with the technological revolution and institutional changes of the time, the world’s major powers are making active efforts, making AI development plans as suited to their national conditions, coordinating their resources in the field of AI, striving to take the lead in this technology and industry, and seeking a more favorable strategic position in the new international system.

The industrial revolution led by AI is an invaluable opportunity for China's development. This will be the first time that China embraces the industrial revolution as a country on the cutting edge of science and technology. As one of the world’s most important cyber powers, China has a natural advantage in advancing AI technology. For one thing, domestic capital favors AI for its future business potential. When the technology is in the ascendant, a lot has been invested in R&D, with a large talent pool formed. After more than 20 years of development, China’s top Internet companies, represented by Alibaba, Tencent and Baidu, have grown into international giants. They have sufficient research funds and are able to draw talents in the field of artificial intelligence from across the world. Technological strengths have been built in different sectors. For another thing, as the world's largest Internet market, China's Internet population has exceeded 750 million in 2017. Thanks to the easy availability of mobile Internet, a massive amount of data has accumulated in relevant vertical fields, providing sufficient feedstock for the development of AI technology. China's Internet market is large enough to facilitate the commercialization of research results. More importantly, China’s systematic industrial planning and policy guidance can provide strong support for AI development. 

General Secretary Xi Jinping clearly stated in the report to the 19th CPC National Congress that “to promote the deep integration of the Internet, big data, artificial intelligence and the real economy”, taking artificial intelligence as an important part of the supply-side structural reform and the growth of the real economy. From 2016 to 2017, the State Council promulgated the Three-Year Action Plan on “Internet Plus” Artificial Intelligence and the Plan on Developing New-Generation Artificial Intelligence. These documents outline detailed plans on the strategic positioning, development goals, major tasks and guarantee measures for China in AI technology and industry. Through the implementation of the above-mentioned plans, promoting the development of AI technology has become an important national strategy in China, which will effectively promote China's progress in artificial intelligence technology and industrialization. It will also have a positive impact on the development of global AI technology. With the implementation of national plans and through the effective integration of resources, China is fully capable of leveraging its comparative advantages in future competition over artificial intelligence, strengthen weak links in development, and take an important position in the new international system as it endeavors to building a great, modern socialist country.

That said, we must also realize that the AI revolution will bring to the international system profound changes that are unprecedented in human history. The process of change itself means unpredictable risks. How to make precautions when the potential risks are in the making, so that human beings won’t be overwhelmed by the impact of the technological revolution, is also an important responsibility of academics in social sciences. 


-----------------------------
Feng Shuai is Assistant Fellow, Shanghai Institutes for International Studies.