Houser and Raymond 21

C. AI Risks Because China and the U.S. are viewing AI as a race, they are negligently moving forward without acknowledging or addressing the risks. Little effort has gone to the creation of ethical standards and international agreements on use by either. Europe, on the other hand, is spending most of its ramp-up time in this regard but only recently determined that its industry is falling behind in actual development.307 Studies disagree on whether AI will eliminate or create jobs,308 but there is a need to understand, at the very least, what skills should be taught in the school system to meet the eventual AI revolution.309 One of the greatest concerns associated with the fourth industrial revolution is inequality.310 Although beyond the scope of this paper to discuss how these risks will manifest, these issues should be addressed prior to forging ahead with AI development in order to win a “race.” These include risks with automated decisions,311 human rights,312 programming ethics into autonomous vehicles,313 autonomous weapons,314 social and political manipulation,315 invasion of privacy, 316 destabilization of the financial system,317 product safety, liability and consumer protection,318 proxy discrimination,319 and cybercrime and cyberwarfare.320 One particular risk is that AI systems are especially valuable in assisting authoritarian regimes. A failure to seek agreement now on international standards and limitations for the use of AI could lead developments down a path that could endanger the safety of the world.321 Although some point to China’s social credit score as an example of a repressive regime, others see it as filling a void in Chinese society in alignment with Chinese culture. 322 A positive step with respect to at least a local level of acknowledgement was taken in May 2019, when the Beijing AI Principles were released, which include a call for “the construction of a human community with a shared future, and the realization of beneficial AI for humankind and nature.”323 These principals were created by a coalition of stakeholders from government, industry, and academia in China. “They include focus on benefitting all of humanity and the environment; serving human values such as privacy, dignity, freedom, autonomy, and rights; continuous focus on AI safety and security; inclusivity; openness; supporting international cooperation and avoiding a “malicious AI race”; and long-term planning for more advanced AI systems, among others.”324 While the AI Initiative indicates that “[t]he United States must foster public trust and confidence in AI technologies and protect civil liberties, privacy, and American values in their application in order to fully realize the potential of AI technologies for the American people,” it does not provide any explanation on how this would be accomplished.325 The final risk that needs to be addressed is that of cyber warfare. The main government sponsored initiative for AI in the U.S. is for the military. The Defense Advanced Research Projects Agency (DARPA) has pledged $2 billion for their “AI Next” campaign.326 When people think of AI uses by the military, they automatically jump to killer-robots. It is important to remember that technology can be used for the good, bad, or neutral (efficiency) purposes. AI can be used to assist with training systems, surveillance analytics, logistical support, the use of unmanned vehicles, and aircrafts to carry supplies, all of which can all be used without threatening lives.327 AI can also be employed to decrypt military communications, hack into weapon systems, distributed denial of services, and cyber-attack utilities and other infrastructures. The real concern is not that other countries may be working on these types of technologies, because most assuredly they are; rather the concern is that there is no understanding of what the future of war will look like or how to respond in kind when it is not just state actors engaging in cyberwarfare but also individuals and groups hostile to an administration.328 This is an area in which international institutions and agreements would be of enormous benefit. 329 Much like the 1968 Nuclear Non-proliferation Treaty (NPT) with 191 signatories,330 countries could agree that the risks are not worth the potential harms in weaponizing AI for assured mutual destruction. Some argue that even if countries agree to stop developing autonomous weapons, countries could still secretly violate such agreements. However, bringing this issue out in the open and agreeing to a set of principles at the very least provides a system of check and balances system to monitor the development of such technologies. As this NPT treaty will be reexamined in 2020, now is the time to get around a table to promote cooperation and peaceful uses of AI. RECOMMENDATIONS Although each region’s strengths and weaknesses have been touched on, the current situation is highly complicated. The interdependence of these regions remains unacknowledged and the U.S. and China are moving at full tilt without regard to how their trajectory will impact the world.331 Given the impact that AI will have in areas beyond where the technology is created, attention to the sweeping societal and economic effects must be addressed. While arguments can be made to name China or the U.S. the winner, each will have difficulty advancing by shutting out the other. However, it is the U.S. who risks losing its position due to its failure in policy with respect to 5G. Regardless, it is imperative that all countries come to an agreement on how to proceed responsibly.332 First, the AI race narrative must be discarded. With the U.S. seeking to shut out the spread of Chinese technology, not only is industry in these two regions impacted, but the EU is also caught in the middle. The U.S. and China must acknowledge their interdependence and work to responsibly advance AI, addressing health, safety, and fairness issues. Rather than isolating researchers from one another, funding must be dedicated to the responsible development of AI. While the EU is well on its way to meeting this goal, global agreement must be obtained on its trajectory and standards.