“Play is a tool the brain uses to generate data on which to train itself, a way of building better models of the world to make better predictions,” Clancy writes. “Games are more than an invention; they are an instinct.” Clancy, a writer and researcher who previously worked at Google’s AI company DeepMind, argues that human intelligence is profoundly insatiable. “Intelligence wants,” she writes. For centuries, games have unveiled the sheer force of this desire.
Clancy probes the history of gaming and the emergence of game theory to at once celebrate, expose and critique “how games have come to dominate modern thought,” as she puts it, and the outsize role they have come to play in the development of artificial intelligence. Gaming, she argues, can help us discover new parts of ourselves, find our moral limits and shift our understanding of ourselves and others. “Our hallmark quality is plasticity. We are, to borrow a phrase from historian Johan Huizinga, Homo ludens, man the player. We slip in and out of identities, value systems, and agencies the way players inhabit the strictures of a game,” Clancy writes. Huddled around a game, we agree to enter a fictional world with fellow players, to abide by the rules and to momentarily abandon the impulses that propel us through our daily lives in favor of an invented set of goals.
Yet Clancy is suspicious of how games have been adopted and interpreted as measures of human intelligence, and critical of their use in military forecasting. Generations of kings, generals and political strategists have used games to simulate the outcomes of war. Today, Clancy writes, militaries treat video games as ripe recruiting grounds. Drone warfare has blurred the line between fictitious virtual battlefields and real conflict zones. Many of the participants in this macabre form of play have committed the cardinal sin of mistaking the rules of the highly constructed world of games for those of the real world. In game theory — the mathematical study of multiplayer systems — all participants are treated as rational actors, bound by the same set of rules and desires. When reality inevitably departs from this neat set of assumptions, games cease to be a useful field from which to learn about human behavior.
An overreliance on game theory contributed to the catastrophic U.S. failure in Vietnam, because our computer simulations could not conceive of, let alone account for, the possibility of guerrilla warfare. One of the pioneers of game theory, the economist Ariel Rubinstein, later warned of its limitations and even argued that it had made the world worse off. “I have not seen, in all my life, a single example where a game theorist could give advice, based on the theory, which was more useful than that of the layman,” he said.
The failures of game theory are, for Clancy, advance warnings of the dangers of AI, which also attempts to simulate reality through “brute computational force” and mobilizes the same blinkered logic of maximization. Game theory laid the foundation upon which AI came to be built, and the theory’s shortcomings threaten to exacerbate the shortcomings and possible dangers of AI.
“Game theory posits humans as a fixed bundle of preferences, when they are, in fact, learning systems,” Clancy writes. “It’s common for people to confuse their personal goals with those quietly manufactured for them by the games in which they participate, wittingly or unwittingly.” We have all been conscripted into various games not of our own design, recruited to play roles that we may not have otherwise chosen for ourselves (or may not even know we are playing). Work environments are increasingly gamified, employees taught to manage themselves and others through automated workflow software; everyday investors allow their savings to be gambled and traded by financiers they will never know or meet. Disappointing, fraudulent and often unjust outcomes are dismissed with what Clancy calls “the callous counterfeit of common sense: ‘Don’t hate the player, hate the game.’” She would like her readers to imagine their way out of these relations: “So what games are we truly playing? And can we design better ones?” she asks. “Does true intelligence entail the ability to choose winning chess moves, or is true intelligence what invents a game like chess?”
At this stage in its development, AI has been likened to a young child that is still learning the contours of the world, hungry for data and always wanting more. “Whereas living organisms are endowed with the desire to survive, artificial agents are endowed by researchers with a desire to win,” Clancy writes. The force of that desire might undo us all, accelerating our already frightening powers of surveillance, data harvesting, optimization and attention-grabbing. “The problem is not that a paper-clip-maximizing AI will arise in the future and turn the universe into paper clips,” Clancy writes. “The maximizers are already here. Any consequences too subtle to measure — environmental costs, civic discord, troubled diplomatic relations — are simply omitted from the score.”
“Playing With Reality” is a foreboding prehistory of AI. Clancy conveys how we became so in thrall to gaming that we forgot where the field of play stops and the real world begins — who gets to be a player and who is merely being played.
Linda Kinstler is a writer who covers European history and politics. She is the author of “Come to This Court and Cry: How the Holocaust Ends” and teaches at the Georgetown School of Foreign Service.
How Games Have Shaped Our World