Sport and AI

AI (Artificial Intelligence) has become the subject of intense reflection recently, not least due to the rising public profile of Open AI’s ChatGPT, and the spread of AI generated images that readily pass as ordinary photographs (and, notoriously, the image of Pope Francis in a puffer jacket). AI suddenly appears to have acquired an uncanny power to imitate human conversation, as well as matching and often surpassing human powers in decision-making and creativity.

While AI has the potential to improve human life (increasing economic productivity, and even the rate of scientific research and development), some—such as the think tank Center for AI Safety—see AI as posing a threat to humanity’s very existence. For example, its application within military technology may not only increase the destructive capacity of weapons, but place the decisions to use weapons outside human control. More subtly, AI may change the way in which humans think and behave. By stripping away our need for productive labour, and leaving us dependent upon machines for our existence, human desires, motivations and values could be radically changed—we would become post-human (see https://www.safe.ai/ai-risk).

While AI has many applications within sport (see Brady, Tuyls, and Omidshafiel 2022), games in particular have long been part of the history of AI’s development. Crucial milestones in the development of AI have been marked by ludic achievements. IBM’s DeepBlue chess computer defeated Grand Master Garry Kasparov in 1997; AlphaGo defeated Go champion Fan Hui in 2015; and, most recently, Meta’s Cicero proved able to defeat 90% of human players of Diplomacy. In playing Diplomacy (a strategic game in which ideally seven players compete, through forging but also betraying allegiances with other players) Cicero had not merely to plan its own moves, but also to anticipate the potential moves of six other players, and significantly to generate convincing text in order to negotiate with other players. While Cicero seemed able to withhold information strategically, it could not obviously say one thing in its negotiations and yet do another (such as attacking an ally), and thus to perform the Diplomacy equivalent of what would be a feint in sport. This is its crucial, remaining, weakness with respect to its human opponents.

AI is now a routine component in the computing power that contributes to video games and e-sports. But traditional sport may only become part of the history of AI if robotics catches up with AI. Only then will AI become embodied, and thus be able to confront humans not simply over the gaming board, but also on the tennis court, football pitch, or in the boxing ring (see for example Javier Lopez Frias and Luis Pérez Triviño 2016; Lee 2022). Although even here, unless AI can learn to feint and deceive, it will never be trusted to take a penalty kick in soccer.

My reading around AI, and in particular a recent article in The Economist (The Economist 2023), reminded me of Bernard Suits’ The Grasshopper (Suits 2005), and his characterisation of utopia in the book’s final chapter. Suits’ utopia is one in which material scarcity has been eliminated—a prospect that The Economist sees as a potential achievement of a super-AI. Indeed Suits hints that there may be a ‘Computer in Charge’ of his utopia (Suits 2005, 156), and that computers will produce utopia’s equivalent of art (p. 152). Suits’ utopia is thus, arguably, realised and run at least through human co-operation with AI or some similarly efficient computer. Here his vision meshes with that of The Economist. But Suits also seems to embrace the post-human future that critics of AI, such as the Center for AI Safety, fear. Advances in psychoanalysis and pharmacology—and AI could carry out the necessary scientific research—will bring about radical modifications to human psychology. All ‘interpersonal problems’ and ‘psychic disturbances’ will be abolished, allowing peaceful co-existence between people (Suits 2005, 150).

Further, The Economist draws on a distinction made by the economist John Maynard Keynes, between human needs that are absolute (thus felt regardless of cultural difference), and ‘those which are relative in the sense that we feel them only if their satisfaction lifts us above, makes us feel superior to, our fellows’ (The Economist 2023). It can be suggested that a similar distinction shapes Suits’ arguments: while a super-AI might satisfy the first set of needs, for Suits it is game playing alone that allows for the satisfaction of the second set. The Economist argues that ultimately the greater value that is attributed to that which is produced by humans—as opposed to that which is produced by AI—will inhibit AI’s scope for revolutionising economic productivity, and thus for bringing about any form of utopia that is characterised by an absence of material scarcity. Suits makes a related point. In contrast to The Economist, he proposes that his utopia is inevitable. But he sees in it a fragility that is rooted precisely in the value that is placed in human productivity. The human desire to be engaged in genuinely productive instrumental labour will ultimately derail utopia, as humans come to prefer their own inefficiencies over the effortless luxury that advanced technology offers (Suits 2005, 160).

Most profoundly, in this context, Suits notes that sport and games ‘are clues to the future’ (p. 159). The point here is not merely the somewhat problematic prediction that AI could transform the world into a games-playing paradise. Rather it is that, firstly, the philosophy of sport may contain crucial resources for reflecting upon the sort of future we humans might want from AI (and, perhaps not least, Suits’ posthumous Return of the Grasshopper (Suits, Yorke, and López Frías 2022) offers this resource); but secondly that, right now, it is in playing games and sports with AI that we learn, in practice, what the human relationship to AI ought to be. Humans play with AI: both in the sense that AI is used to design and facilitate play in games and sports, and in the sense that we humans compete against AI (for example in video games). The restrictions and regulations that we are already constructing around that playful relationship—to ensure that the games remain meaningful and valuable to us—may say much about how AI regulation in general should be shaped in the future. Put otherwise, it is through thinking about and playing sport that we might mitigate the existential fears of AI’s critics. (Do we really want to allow an AI that can send the goalkeeper the wrong way?)

Disclosure statement

No potential conflict of interest was reported by the author(s).