Fact, Fiction, and the Future of Technology in U.S. Grand Strategy

Futurism and prediction are perilous exercises - especially for technology, security, and farming where humility is forcibly ensured by nature or a sentient adversary on a periodic basis.  Peering into the unknown to assess the specific sources of strategic surprise and to catalogue uncertainty requires a willingness to demonstrate strong conviction and to prune the set of hypothetical scenarios down to a more useful illustrative subset which form a communicative whole.  Clarity in this arduous task revolves around clarity of assumptions coupled with artful blends of deductive and inductive reasoning to make leaps which might be followed by readers who join such a constructed journey.

Michael Kanaan's T-Minus AI is an example of extrapolative commentary about the near-term future of artificial intelligence (AI). Kanaan's bounds his discussion with a functionally focused definition of his interest here: "the single, underlying goal of AI is to empower computer systems to perform the higher kinds of intellectual functions we've traditionally thought are only possible by humans" (p. 118). As the current director of operations for the Air Force/MIT Artificial Intelligence Accelerator, Kanaan’s writing is demonstrably aware of the cyclical nature of interest in this field, which has already waned and waxed three times in the post-WWII era. The conceptual development of AI is summarized in the first third of his book, but seriously interested readers would be well advised to pursue deeper treatment available from The Master Algorithm by Pedro Domingos. It is in the later sections where Kanaan makes the framework of his thinking clear.

Kanaan first describes several phenomena that exemplify the increasing importance of AI in current affairs. They include the social issues stemming from algorithmic bias, the economic issues driven by the global exponential growth of data, and several technological trends, notably, competitive games as proving grounds for AI development, the resurrection of machine learning and neural network approaches to AI after the most recent cycle of "AI winter," and the proliferation of software bots (many with malign intent) in contrast to the hardware robots of the physical world. Kanaan comes down on the side of the near-future likelihood of "narrow" or "weak" AI, limited to specific domains, as opposed to "strong" AI, artificial general intelligence, a point of view popularized by the University of Oxford philosopher Nick Bostrom, among others.

Kanaan still emphasizes that narrow AI will still be transformative, even geopolitically. He highlights Vladimir Putin's statement in 2017, "Whoever becomes the leader in (AI) will become the ruler of the world," and China's near-simultaneous announcement of its Next Generation Artificial Intelligence Development plan, calling for China to become the world's leader in AI by 2030, only ten years away.

In this section, Kanaan's analysis turns to international relations. Unlike the twentieth century's experiment with totalizing ideologies, most nations currently exist on a spectrum from democratic to authoritarian. China's authoritarian government is using mass surveillance, the Great Firewall, and an Internet-based social credit system to create conditions that favor mass conformity. Under this technologically mediated system, nonconformist minorities like the Uighurs can be placed under mass detention with little public outcry, both domestically and internationally. Yet the economic success of China makes aspects of this model an attractive one to countries with weakened democratic institutions or none at all. For example, Kanaan mentions its surveillance model being directly exported to Ecuador, Zimbabwe, Uzbekistan, Pakistan, Kenya, Venezuela, Bolivia, and Angola among others (pp. 189-190).

At the same time, Russia, the quasi-authoritarian successor state to the Soviet Union, now shares a common economic ideology with the United States, yet remains adversarial in its geopolitical aims and means of pursuit. Russia today has relatively far fewer resources at its command than it had during its days as a superpower. Kanaan brings up Russia's development of its electronic warfare programs to level this unequal playing field, and in Lethal Autonomous Weapons Systems (LAWS) as an example of near-future AI improvements whose possibility we should recognize, but also its surveillance, propaganda, and disinformation campaigns. He especially notes the efforts of the Internet Research Agency operating out of St. Petersburg and its success with social media disinformation, not dismissing its effects on the 2016 U.S. presidential election.

Nations interested in military power see in AI an opportunity to leapfrog militarily. Kanaan compares the current geopolitical moment with respect to AI as parallel to the Soviet launch of the first artificial satellite, Sputnik 1, in 1957. These are what Kanaan calls "moments that awake nations." There have been very few of these moments in human history unassociated with direct military conflict. The Sputnik moment was a rare event that took place in a particular historical context following the defeat of the Axis powers, the victors with directly competing ideologies. Today, the majority of nations have democratic forms of government, most make rhetorical obeisance to some democratic ideals, and nearly all participate in the same economic system. The competition is less stark. Is this truly a moment that will wake the nations?

P.W. Singer and August Cole answer this question affirmatively in their 2020 speculative technothriller, Burn-In: A Novel of the Real Robotic Revolution. Singer is currently a Strategist at the New America Foundation, and Cole is a nonresident senior fellow at the Brent Scowcroft Center on International Security at the Atlantic Council. Their novel is extrapolative near-future science fiction, heavily end-noted with citations to recent technological and social developments related to AI, using Cole's method of "FICINT" (fiction and intelligence) to explore the evolution of armed conflict.

Unlike their 2015 novel, Ghost Fleet, which contemplated a technologically advanced great power war between a near-future China allied with Russia against the United States, Burn-In is focused on the problems of policing and counterterrorism in a setting where the development of AI—in a form closer to artificial general intelligence than to a narrower, domain-specific AI—has exacerbated social and economic inequality to the point of domestic insurgency. Its topics of interest thus rarely directly overlap with Kanaan's geopolitical emphases, but rather complement them with a domestic perspective.

As a technothriller, many elements of Burn-In are purely Hollywood high concept. An FBI agent is partnered with an artificially intelligent robotic Tactical Autonomous Mobility System, TAMS (clearly meant to be related to the aforementioned LAWS), to investigate a nihilistic domestic terror group disrupting American infrastructure in physical and cyber attacks based on the Biblical ten plagues of Egypt. The strength of Burn-In is in the attention Singer and Cole pay to continuity of behavior in a changing technological environment. The authors are aware that technological transformation does not lead to immediate changes in patterns of behavior or structures of thought. Professionals and extremists alike will first use AI as tools to further their ends before making more fundamental changes in their practice or mindset. Longer term doctrinal innovations will likely follow for established players, but both Kanaan and Singer make little reference to the kind of strategic surprise driven by more purist doctrinal innovation and technological repurposing as noted by Meir Finkel in On Flexibility.Both Burn-In and T-Minus AI make the case that the potential for AI to erode trust in the structures and institutions of society is significant and pressing. For Kanaan, this is framed as largely a geopolitical issue, through channels such as bot-driven disinformation campaigns directed by an adversarial power. For Singer and Cole, this erosion also may develop through AI-driven processes of social and economic dislocation, and through domestic actors using AI to disrupt existing norms to their advantage. Strategically, the former case is more actionable for policy makers than the latter. Kanaan therefore believes that it is imperative that AI be developed “in ways consistent with fundamental human dignities… and only for purposes consistent with democratic ideals, liberties and laws.” In his conclusion, Kanaan allows the GPT-2 deep learning language model to finish his thought for him: “Our job is now to convince the public in particular that using AI to achieve these aims is a necessary and desirable part of our society, but we cannot afford to do so unless we know how it will best be used and when.”