Artificial Intelligence – the term is used so frequently nowadays that one hardly understands what it actually is. At present, we do not possess computers capable of beating mankind at most intelligence based tasks – in other words, we do not possess Artificial General Intelligence (AGI). We only possess algorithms capable of excelling at very specific tasks. DeepMind’s AlphaZero is incredibly potent at chess, while google’s Inception-Resnet V2 architecture is highly competent at classifying images. AlphaZero cannot classify images, and Inception-Resnet cannot play chess, so one must essentially construct and train new models on new data for every task one wishes to automate. However, our narrow models are incredibly effective. AlphaZero likely has an Elo rating of around 4000 (for reference, Magnus Carlsen has an Elo rating of around 2800), and various deep-learning algorithms are now capable of outperforming humans on image classification tasks.
In order to build these narrow models, one needs large quantities of data, vast resources of computing power, and a pool of talented Machine Learning researchers. There are only two types of entities capable of fulfilling these requirements – large corporations and powerful nation-states. Google has led the way on the corporate front, with Baidu, Amazon, Facebook, and a handful of other tech giants following closely behind. Yet, all of these corporations are part of the national efforts of, for the most part, two nations – the United States and the People’s Republic of China.
China has made great progress in the field of AI over the past few years. China’s Social Credit System, which seeks to punish “societal ills” and reward good behavior, serves as both a method of securing political power for the ruling communist party as well as a gigantic data-collection machine which can be used to test new algorithms. China’s state-run news agency, Xinhua, introduced a digital news anchor, featuring a life-like appearance, sound, and mannerisms. This digital news anchor is likely a combination of GAN networks, generative networks which have been gaining popularity since Ian Goodfellow authored a research paper on the topic in 2014. Chinese tech giants such as Alibaba and Tencent are utilizing AI frequently in business. The main problem faced by the People’s Republic of China is the lack of high quality research – the 2017 NIPS conference (widely considered to be the most prestigious conference for Machine Learning) was dominated by American companies. However, China plans to address this problem via its plan to become a global leader in AI by 2030. The government plans to build a 150 billion dollar domestic AI industry, combining government funding, new AI institutes, and private sector expertise to launch AI “moonshots” aimed at achieving better, faster, and more versatile algorithms.
The United States of America, meanwhile, leads the world in AI prowess. For instance, self-driving cars are already being tested and even implemented in some portions of the United States – following the efforts of massive corporations and breakthrough innovations in image recognition and video processing. Chinese students still travel to America to conduct AI research. But America’s lead is certainly shrinking. Chinese AI startups now receive more funding than American startups, and China’s massive population pool will eventually manifest itself into highly-skilled researchers. This is problematic, considering the idea of Thucydides Trap – that a rising power (in our case, China) often throughout history ends up fighting a war with an established power (the United States). While Thucydides Trap may be somewhat nullified due to the threat of Mutually Assured Destruction, there is no doubt that China and the United States will remain politically and diplomatically in opposition over the course of the near future. Thus, the United States must maintain and grow its advantage in AI – the technology is simply too important to ignore.
America must, to start, implement a national plan for the development of AI and AI-related technologies. The private sector and academia have both produced tremendous results, but the government must now lead the effort in order to produce even greater results. Boeing, Douglas, IBM, or North American Aviation, in the 1960s, would most certainly fail in any attempt to land a man on the moon. However, with the combined effort of all of these companies, under the supervision and overall leadership of NASA, the United States placed man on the moon. Similarly, Amazon, Facebook, Google, or Apple, with no overarching goal, cannot achieve something tremendous – something that these entities would be able to achieve given unity and a common goal. The United States must commit itself to the funding of STEM programs and the sciences – a long term investment to secure America’s future. The development of AI must be a national priority – else, America will lose the AI arms race, and with it, her spot atop the global stage.