Strategy, Evolution, and War: From Apes to Artificial Intelligence by Kenneth Payne. Georgetown University Press, 2018, 272 pp.
In popular imagination, artificial intelligence (AI), strategy, and war typically conjure images of Terminator-like cyborgs or manipulative programs such as 2001: A Space Odyssey’s HAL (Heuristically programmed ALgorithmic computer). What they all have in common is anthropomorphization—we believe AI will make decisions in a humanlike way. Dr. Kenneth Payne, a senior lecturer at the School of Security Studies at King’s College, London, and the author of two previous books on the psychology of strategy and conflict, takes exception to our tendency to anthropomorphize. He contends that AI will arrive at strategic decisions in an unhuman and unfamiliar context.
In Payne’s Strategy, Evolution, and War, he argues that strategy is a biologically human phenomenon. Any future artificial general intelligence (AGI) will lack the evolutionary context to make decisions like us. Further, he sees AGI as transformative to decision-making in war, altering existing power structures, the utility of deterrence, the offense/defense balance, and how we view warriors (p. 1). To arrive at this conclusion, Payne divides the book into three sections: the evolution of strategy in humans, the introduction of culture to evolved strategy, and the impact tactical AI (or narrow AI) and AGI will have on future strategy and conflict.
In part one, Payne defines strategy as a human psychological phenomenon (p. 30), which allows one to see the enduring nature of war through history as the way groups purposefully use violence to achieve their goals over others in a more abstract sense. The first way is through mass—assembling larger groups than an adversary. The second is using deception and surprise to defeat enemies. In both, humans needed and evolved a “theory of mind” to understand the aims of allies and adversaries (pp. 54–55) and an ability to use that information socially to form larger groups (p. 65). The need for group cooperation in conflict helped both consciousness (particularly to understand our social status) and unconscious heuristics to develop in order to navigate meeting threats in a community setting.
The author’s extensive use of sources and ease of integrating evidence from anthropologists to cognitive scientists highlight the first section. Payne is convincing in backing up his argument and acknowledges the speculative nature of some of the evidence, such as the anthropological record or the inferences made with cognitive psychology (p. 70). Notwithstanding, this section is a valuable reminder to the strategist of the strands that weave together the conscious and unconscious ways we make decisions.
In the second section, Payne uses three vignettes to demonstrate the impact of culture on evolved strategy and draws six themes: (1) technology shifts the scale between time and violence, (2) ideas on how to assemble and utilize force are important, (3) technology both shapes and is shaped by society, (4) writing is the most strategically revolutionary development, (5) writing is an extension of our ability to abstractly reason, and (6) the most advanced weapons do not alter the evolved nature of human strategy (p. 91). With Thucydides and the ancient Greeks, a symbiotic nature developed between technology and society regarding the hoplite revolution (p. 106), and the development of writing was the first real “AI” that enabled humans to share our evolved strategic faculties externally (p. 101). With Clausewitz and the Napoleonic era, the idea about the generation and use of force manifested with the French, and Clausewitz reminds us that psychology matters in strategy in refutation of the scientific bent of the Enlightenment (p. 134). Last, the advent of nuclear weapons shifted strategy from the use of force to deterrence. However, the evolved psychological need for the “theory of mind” was still imperative for Kennedy and Khrushchev during the Cuban missile crisis (p. 157).
The second section forms the weakest part of the book, as Payne also admits. Generalizing thousands of years of history into three Western-focused vignettes points to specific Western cultural pathologies but neglects other cultures’ impact on how humans strategize. A non-Western example would have added more support to his subthesis: once humans begin to use tools, a complex interaction occurs between war, society, and technology.
In the last section, Payne outlines the divide between tactical AI and AGI, concluding that while tactical AI (already in use today) comports to our strategic logic, AGI will arrive at conclusions in a unique and foreign manner. However, while tactical AI follows our strategic logic, it will transform how humans conduct strategy because rather than a weapons system, it is a decision-making tool (p. 192). Its speed and freedom from heuristics will make the offense more powerful, giving an advantage to first movers and those who develop advanced systems. Further, remote AI will challenge how we think of warriors and alter the bureaucratic organization of military forces.
Transitioning to AGI, three significant points underpin the conclusion that AGI will act in unfamiliar ways: evolution, interaction, and intrinsic motivation. First, AGI will not have the same survival instinct inherent in humans—a powerful force shaping our evolution. Second, machines will not need to behave in the same social way as humans (p. 195). Furthermore, given those two points, machines will maximize their reward functions differently than humans, looking for perfect solutions vice the evolved human decision-making strategy of satisficing (settling for satisfactory versus optimal solutions) (pp. 201–4). Thus, machines calculating in nonhuman ways may surprise and disappoint their human masters. Lastly, Payne considers the merging of humans and AGI. While speculative, a hybrid AGI-human would still have a biologically evolved consciousness and basis, thus being subject to our evolved psychological decision-making (p. 221).
Strategy, Evolution, and War is a must-read for any strategist, as it is an excellent cross-disciplinary survey of how humans perform strategy. Payne’s research is thorough, his context on everything from the theory of mind to artificial neural networks is informative, and his conclusions are sound while not getting overly speculative. This book is a reminder of the evolved way humans conduct strategy—inherent with fog and friction—and a caution that while we do not need to fear killer cyborgs, AI will make decisions in radically different ways.
MAJ Aaron Cross, USA