Politics 101.
One of the first things they will teach you in any international relations course at university is that the US is a unipolar hegemony. In simple terms: the US is pretty much in charge of setting the rules for the international community. Put more candidly: the US is the boss of the world. Another lesson which you really don’t need an international relations degree to know is that those in power like staying in power. Now, enter AI: a powerful technology with the ability to completely revolutionize the defense industry, political functions and the way information is created and distributed.
You can start to see why in the last five years countries have invested billions in AI (the US tops the chart with around $329 billion followed by China with $133 billion). You can see why most of the major world powers have declared intentions to dominate the world of AI and integrate it into their societies. And you can definitely see why China and the US are going head-to-head with a race to AI.
The race of a lifetime.
With world domination on the line, I’d go on a limb and say it’s probably the most exciting race we are going to be witnessing in our lifetime. Unlike the races that created the current world order (industrial revolution, internet etc.), the ‘contestants’ (read countries) are getting access to this technology at the same time.
That’s not to say that the starting line for all contestants is equal. “Computing power, talent, data, and physical infrastructure are widely recognized as the key building blocks for AI development and adoption.” How accessible these building blocks are for a country will put them at a serious advantage or disadvantage in the race to come. It’s the lack of access to these resources that will cause less developed countries to fall behind in the race for AI domination.
Computing power has been defined as the most critical factor which influences a country’s potential in developing high-level AI systems. China found this out the hard way. In October 2022 the US coordinated a strategy with the Netherlands and Japan known as ‘small yard, high fence’, this imposed export controls to block the transfer of vital US semiconductor technology to China.
Dirty Tricks?
This ‘small yard, high fence’ strategy has been under some criticism. While it has been justified by the US government as a national security measure, there are those who believe it is more suitably described as a “protectionist measures taken by nations seeking to preserve or increase their own competitive advantage__”. __Those who consider the strategy an act of techno-nationalism further criticize the strategy on the grounds it holds back the development of AI and stifles international collaboration.
In one of the latest issues of The Economist it was suggest that if the US really wanted to restrict Chinese progress in AI, then it should focus on finding ways to attract and keep top researchers, for example through easier visa rules. The article claims that “the breakthroughs ai needs will come from giving ideas and talent the space to flourish at home, not trying to shut down rivals abroad.”
Yet, discussions at the 2024 Global Conference on the Geopolitics of AI suggested that the reasoning behind the strategy was less concerned with economic competition and more concerned with global security. At the conference, Karen Kornbluh (the former US Ambassador to the OECD) argued that it wasn’t the intention of the US to prohibit open trade competition, its goal was to ensure fair competition. Kornbluh goes on to explain that the strategy did not arise from nothing and that when the US brought China into the World Trade Organization (‘WTO’) it hoped they would liberalize and be obedient to the rules of the multilateral system. But instead, she argues, they used cyber tools to steal intellectual property and breach the rules of the WTO. Therefore, the export controls are a way of ensuring that an unfair and potentially dangerous player doesn’t get too far ahead in the AI race.
At the same conference, Kevin Rudd (the former prime minster of Australia), argued that it was unfair to criticize the strategy as an exercise in crazy American protectionism. Describing the strategy as a ‘slow down operation’, he claims that when you look at the policy instruments available there was really no alternative. Anne Neuberger (Deputy Assistant to the President and Deputy National Security Advisor for Cyber and Emerging Technology at the national security council) followed these statements up by raising concerns of what China might use AI for, specifically with regards to advancing the chemical sciences to build bioweapons.
Is self-sufficiency feasible?
There are those who claim that the strategy has been a failure as it has unintentionally stimulated the growth of a research system in China and pushed it to develop self-sufficiency. After redirecting capital, talent and government bandwidth the Semiconductor Manufacturing International Cooperation recently managed to build a 7nm chip. Yet, this achievement was reliant on western machinery. A vulnerability which suggests that claiming China has attained ‘self-sufficiency’ may be somewhat of an exaggeration.
Oliver King-Smith points out that it is difficult for us to know exactly how efficient the export controls have been and how just far AI technology has been developed in China. According to King-Smith there are rumors of chips smaller than 5nm being produced, while we can’t be sure of what technology has been developed, it’s better not to underestimate the advancements taking place.
Then again, what nation can be considered 100% self-sufficient in the development of AI? The US is almost entirely reliant on imports of critical minerals for the creation of semiconductors. China responded to the export controls with controls of their own on critical minerals such as germanium, gallium, and more recently graphite. In these areas, China has a historic supply-chain advantage.
Karen Kornbluh suggests that it is this international dependence on critical minerals, chips and compute power that may drive international cooperation and regulation. Kornbluh argues that by creating a group which involves not just the usual close allies of the US, but also those who have deposits of critical minerals, you can incentivize the authoritarians to come to the table. While involving critical resources of the supply chain is sure to get the authoritarians to listen, getting democratic countries and authoritarian states to come to an agreement on a common vision is a much more complicated matter.
Visions of AI
To know what really matters to countries, the values or priorities they really hold, you have to look at what they are willing to sacrifice. Both the US and China have made some choices which will certainly restrict their capacity to innovate in the AI sector. But the reasons for these choices are vastly different and reflect the very different future applications for AI that these countries envision.
The restrictions that may limit innovation implemented by China are all in response to the fears created by the unpredictability and black-box nature of AI. This unpredictability creates serious worries for closed societies where AI may jeopardize the states control on free speech. The Chinese government is willing to push for and support AI innovation, as long as it doesn’t affect its ability to control its population.
On the other hand, the US and other democratic societies have chosen to limit AI innovation in the name of ensuring adequate protection of citizen’s fundamental rights. Anne Neuberger explains that “When we talk about strategic competition, we fundamentally begin that competition with our values. Who we are as a group of countries. We are competing on the fact that a citizen feels confident that their government’s, for example, facial recognition models are not made to surveil them in a way that’s inappropriate.” Neuberger says that this comes with an acceptance that we may be allowing countries like China, for example, to have better developed facial recognition models.
It should be noted that some Chinese citizens argue that it’s not just the government that wants to prioritize these values:
It’s possible that the different visions for AI simply reflect the differing cultures between China and the US.
These vastly different priorities reflect some of the challenges that will be faced in finding a middle ground between these two great powers competing for AI domination. For now, it seems the US holds an advantage in its compute power and access to talent, but the race isn’t over yet. The winner is yet to be crowned.
Written by Celene Sandiford, smartR AI
Image 1 credit: https://www.globaltimes.cn/page/202304/1288893.shtml
Image 2 credit: http://usanasfoundation.com/us-hegemony-ends-by-seppuku-part-2