Geopolitics of Artificial Intelligence: Research Proposal

he modern world runs on algorithms. The news you get from Facebook, the search results from Google, to the products suggested by Amazon are all driven by increasingly powerful Artificial Intelligence algorithms. These AI systems nudge and shape our daily decisions but also undergird the control of critical infrastructure and weapons systems whose complexity far outstrips what any human or even a team of humans could reasonably react to and process. When so many of our decisions have been outsourced to the infinite black chasm of unseen data-centers, the question becomes, who is really in control? These algorithms once deployed run with little outside human supervision and in the coming years the deployment of unsupervised learning models may lead to black box systems in which the AI researchers themselves are unable to discern how their progeny came to their decisions. Given such a digitized world and the degree of control which has been granted to these algorithms, I seek to explore how AI will come to shape the global balance of power. Are nations and their governments still the driving force in international affairs when tech titans can rival them in wealth and population? What kind of power structures does AI enable and how does this emerging digital order reshape our classical notions of International power distribution.?

Classical realist conceptions of power rest upon the notion of a finite nation-state imbued with land, people, and money which challenges other nations for hegemony within the international system. States were built on the hard power of money and guns to secure and defend their interests. After World War II, the marker of a “Great Power” amongst the nations evolved from those who could muster the largest armies to those who could muster the technological prowess to produce a nuclear weapon. The new rivalry of the 21st century is no longer dominated by a race to build the most warheads but rather one focused on accruing the most data to feed the voracious appetite of deep learning algorithms that drive the world’s economy. National governments, however, are becoming mere spectators to the development of AI; overshadowed by the global tech companies like Facebook, Microsoft, and Google or Tencent, Alibaba, Baidu of China. The power polarity of the 21st century world order therefore becomes not just multipolar but also multi-dimensional as states compete not just with other states but also with corporations (sometimes within their own geography) in order to maintain their influence. The implications of the democratization of such power is profound as it upends over 300 years of centrality for the nation-state being the driver of international affairs.?

The Industrial Revolution relied upon states either being endowed with coal (and later oil) or seeking out arrangements to secure it by diplomacy or force. States such as Saudi Arabia or Iran could extract rents for wealth and military assurances from the Great Powers who courted them for their oil. As the world moves from an industrial to a knowledge based economic model, the value of natural resources such as oil no longer hold the same prestige or leverage. Given the plethora of public cloud computing resources and a world awash in data, the barrier to entry into the AI arms race is ostensibly quite low. Small nations such as Israel and Singapore can rapidly move up the global value chain even though they are bereft of natural resources due to their investments in education and technological development. These small nations are only limited by their own human talent and their ability to master the mathematics behind the intellectual magic of AI.?

This new tech-driven multi-dimensional world order has, on its face, a positive-sum quality, owing to the inherently non-rival nature of digital goods. While the oil under the sands of Arabia can only belong to one man at a time and thus spurs conflicts for its control; the world’s knowledge is freely available on Wikipedia to anyone with an Internet connection. This leads to questions of digital-haves and have-nots. Even with free digital resources for education such as Khan Academy and computation services such as WolframAlpha, only 60% the world’s population currently has internet access. While anyone can learn to code for free on a cheap laptop, it takes the resources of a global tech giant to orchestrate the world’s top talent and build the massive data centers necessary to run state of the art algorithms whose power requirements dwarf that of cities. AI-powered firms are able to slingshot their growth based on the first mover advantages afforded by a digital first business model. Firms with a solid digital offering can rapidly attract users and scale accordingly. This influx of data trains the backend algorithms which allow the firm to make ever better products and attract ever more users leading to a runaway positive feedback loop until it becomes nonsensical to compete and creates a de facto monopoly. The research question to investigate then becomes, does this accelerated AI-dominated internet still support a free, open, and democratic market supporting a global user base or does it naturally portend a winner-take-all monopolistic system??

I plan to base the primary research question regarding AI and the global balance of power within the neorealist tradition of international affairs. Whereas the focus shifts from having the largest yield warhead to having the most functional algorithm; the concept of capitalizing on the technology of the day to drive power and influence remains the same. What is novel, however, is that it is no longer just states who can realistically drive the international arena. All actors, be they states, major tech firms, or even individual billionaires will use the resources available to them: money, talent, and algorithms to ensure their interests. An immediate difference in the character of power here is that geography and territory, so integral to concepts of statehood and national power are no longer as relevant for private actors. The global tech giants, be they firms or individual moguls, are not geographically bound in their influence and can move anywhere which is most advantageous to them (and even off the planet in the case of Elon Musk). The research will examine the emerging multipolar system and the potential points of conflict and rivalry amongst states like the United States and China and their national champion firms. It will explore the various economic and political models used to encourage and drive AI development and its relationship with building and supporting overall national power. The research will also examine the inherent rivalry between private sector actors and their ostensible national benefactors as they eclipse their home countries in speed, wealth, efficiency, and public trust to solve problems. States are no longer only in competition for influence globally amongst other states but are in competition with the private sector writ large in providing services at scale and efficacy for their own populations.?

In order to develop the specific evidence to support my conclusions, I will first evaluate case studies whereby private actors solved problems that were henceforth the purview of states such as SpaceX’s launch of humans to the international space station or Elon Musk’s promise to Australia to solve its energy shortage within 100 days using his battery packs. I will then evaluate the extent to which AI supports or enhances the actor’s capacity to solve the specific problems in order to determine its relative importance. I will also need to evaluate counterfactual examples where AI had little to no role in solving the state level problem. The other key measure is the level of reliance any individual actor has on the use of AI for its operating model and the inherent advantages and limitations this may impose. The dependencies which are created as a result of deploying AI and maintaining the underlying infrastructure may lead to emerging national security prerogatives as industrialization once drove military conquests for oil.? However, emerging AI systems may eventually act as a sorcerer's stone granting its owner such immense advantage over competitors that it renders them impotent. This leads to a fundamental question: What kind of AI would imbue its owner with such power over its rivals that men would kill to possess it?