Haenel 18
June, Fabio Haenel currently works at the Global and European Studies Institute, University of Leipzig. Fabio does research in Information Technology and Politics, Human Rights and Foreign Policy. “The Prospects and Dangers of Artificial Intelligence on International Security: The Case of a Sino-American Arms Race.”, https://www.researchgate.net/publication/325542003_The_Prospects_and_Dangers_of_Artificial_Intelligence_on_International_Security_The_Case_of_a_Sino-American_Arms_Race – ECM
2.5. The Ultimate Weapon: Is AI Determined to Fuel a Sino-American Arms Race?
But it is not necessary to venture that far down the line of what destructive force might and might not be deducted from AI in the future as contemporary autonomous weapons already are an integral part of the armories of modern armies. “The technology to deploy autonomous weapons is largely in place, and the engineering tasks seem to be easier than those involved in creating self -driving cars,” stated the UK Ministry of Defense in 2016.58 Non-proliferation activists argue that the development of Lethal Autonomous Weapon Systems (LAWS) is particularly worrying given the parallel development of privately accessible autonomous drones, effectively introducing the use of precise lethal weapons anywhere and anytime, not just on dedicated battlefields, and thus further tear at an already crumbling sense of security within industrialized societies. Three leading computer scientists and philosophers thus ask, “are autonomous weapons likely to reduce political aversion to conflict, or perhaps result in accidental battles or wars? Would such weapons become the tool o choice or oppressors or terrorists?”59 And, most importantly, who will be held responsible or autonomous attacks legally, if not ethically, given that it is only software that calculates who to kill?
Analogue to Google’s AlphaGo automatically improving itself in order to beat the best human players in Go only by following its goal to maximize the probability o winning no matter the cost, a similarly programmed autonomous weapon “would see no difference between victories that required it to kill one or 1,000” men outlining great challenges to humanitarian law.60 Reducing the amount o humans involved in carrying out invasive strikes against the defense organs supporting the current international security system, e.g. by coordinating swarms of autonomous drones overwhelming an adversary’s defense system, will uproot its equilibrium o prioritized, mutual defense: “This would lead to a very unstable international configuration, encouraging escalation, arms races and the replacement o deterrence by preemption.”61 The advent of widespread introduction o LAWS thus echoes the looming threat o the early Cold War be ore the ratification of the Treaty on the Non-Proliferation of Nuclear Weapons, especially so today when two central forces manifested as the catalysts of the weaponization of Artificial Intelligence: the United States of America and the People’s Republic of China.
The specific characteristics of AI applications cater tremendous benefits to the relevant power mastering its use in weapon systems first, or holds the decisive advantage over its trailing adversaries. Any new generation of weapons invented by humans have become either more powerful, precise, or both, over time by maximizing the inflicted damage while minimizing the ef ort needed or its execution. AI-assisted remote and autonomous weapons will cater to both of these values, effectively reducing the operational limits of its deployment to a new minimum. Naturally, this sparks the interests of defense contractors, governments, and international organizations alike. But unlike the Manhattan Project of the 1940s, the current development of AI is spear-headed by private technology companies using their wealth, expertise and user databases to steer the challenging task of funding, researching, developing, and testing their own innovations internally successfully, and do so “principally in the US and China.”62 These almost entirely decoupled markets already experience a parallel augmentation of AI in consumer products, led by Google, Facebook and Microsoft on the one side, and Baidu, Alibaba, and Tencent on the other side o the Pacifc.63 But Google’s AlphaGo did not just de eat the world’s champion in Go in May 2017, it defeated Ke Jie, a Chinese citizen, in a game streamed live online rom Hong Kong. Once it became clear the US-American algorithm would de eat the incumbent Chinese champion the streaming website was immediately blocked within China by authorities.64 Shortly after that, the government has announced an enormous funding program to foster and catalyze Chinese talents and start-ups to take over the market, and becoming “the global leader in the field by 2030.”65 Already, the trend or Chinese computer scientists to look to the United States or education and employment is reversing, boosting Chinese companies with high-skilled expertise and private connections to Silicon Valley.66
But while Beijing is rallying financial and political support to help build up its domestic AI industry to become globally competitive with increasing velocity, such a centralized program is missing in the Trump administration’s budget plans thus ar. Commentators are yet undecided if this will help the Chinese companies to level the playing field with their American counterparts or outright outpace them, or if China will continue to run a parallel domestic market large enough to be sufficient without the meddling of foreign actors. Their reinforced policy to dominate the sector by 2030, however, indicates that it will not only be intended domestically but also to anchor Chinese high-technology, and the intellectual property at its foundation, at the core of the interdependent global economy of the unfolding century. Such a “race dynamic” favoring hastened deployment of AI technology is expected to diminish the interest to invest in its robust development, putting AI Safety researchers on alert.
Bostrom, or example, has thus made the case or collaborating efforts to innovate rather than competing against each other, allowing for a safer pace of innovation, thus avoiding potential conflict, and above all setting a stepping stone to an international resolution to take the prospects and dangers of Artificial Intelligence seriously.67 Particularly worrying is the act that most, if not all, calls or robust development are heeded only by Western academics and their institutions, defining a linguistic, academic, and political rift between the West and China. And while both consumer and government adoption of autonomous technology is slowly taking up pace in Western industries, the Asian market, and China in particular, appears to welcome drastic innovations in AI-assisted products much more easily, offering higher potentials to grow or comparatively less-regulated domestic companies in Asia. These developments already outline a “very unstable international configuration, encouraging escalation and arms races” from an economic point of view alone. As their respective national security bodies ramp up military programs to further implement AI in their tactical arsenal the most pressing question remains how—or if at all— Artificial Intelligence, ultimately, can be governed.