By Michael Cheng
Why excellence is impossible without human imperfection
Imagine that you are watching a basketball game where every player is an AI athlete. With names like SlamBot and DribbleDroid, these AI athletes make every shot and never tire, with perfectly synchronized movements and perfect strategic plays. As the game progresses with robotic precision, you begin to wonder whether something essential is missing. Can a game dominated by robots ever capture the heart and soul of sports? From the thrill of unexpected athlete and referee errors, to the drama of human endurance, to the grit and determination of humans who train for decades in the pursuit of greatness, there appears to be something missing. In this article, I explore some of the potential impacts of AI on sports and explain why human athletes are unlikely to be replaced by AI. I argue that part of the appeal of sports involves shared histories, experiences, role-playing, and relatable failures that robots, by their natures, cannot provide. This article is not merely an argument about the role that robots might play in sports, but also a discussion of the roles that sports play in the lives of humans and how those roles might change as technology advances. The rise of AI is not only changing what we do, but also forcing us to reflect on why we do.
How are human and robot athletes different?
Consider the “crude thesis” that the main goal that athletes have is to win while following the rules, and that winning by a larger margin is more desirable than a smaller margin. This seems sensible enough at first glance. A child who excels at basketball from a young age will often be praised for their athletic talent and funneled into progressively higher-level training camps before potentially playing in professional-level leagues. Now suppose that a robot basketball player existed that was substantially more skilled than the human athlete. Perhaps the robot could land perfect 3-point shots on every attempt and never foul another player on day one, without the expensive training camps or decades of athletic development that a human requires. The robot basketball player would be objectively better at “winning” than the human basketball player, but the fact that it makes no mistakes, or different kinds of mistakes than a human might make, places a limit on the extent that humans can sympathize with, or derive entertainment from the robot athlete. Part of the appeal of sports is that spectators can empathize with and follow the development of human athletes, and that humans are by nature imperfect when they pursue their goals. Imagine what would happen if flawless robot athletes were integrated into other sports. If a robot runner could run a mile in 3 seconds, it would likely be boring to watch compared to a human runner that set a new mile record of 3 minutes. If a robot fighter knocked all its opponents out with a single blow, it would not necessarily be more entertaining than a human who elegantly bobs and weaves and wins through grit and determination.
Even though the athletes who play sports often have the goal of winning, it is not the case that the individual sub-tasks that must be accomplished to win must be performed as efficiently as possible, irrespective of the means. There is something lost when robot athletes play sports instead of human beings; something essentially human that artificial intelligence cannot fully replicate. After all, we would rather watch human athletes run than cheetahs or cars, even if human runners are slower. We often derive substantial entertainment value, or even learn more, from watching someone who messes up (e.g. bad calls and bad referees are core components of many sports). The fact that humans make mistakes is a feature, not a bug.
One of the most salient differences between human and robot athletes is the length and nature of athletic training involved. A human athlete spends years or decades training to perfect their skills and their training involves sustained interpersonal interaction with other human beings. The high-level human athlete has likely worked with multiple coaches (who each emphasize different aspects of the sport), travelled to compete in many locations as they develop their skills, and developed bonds with teammates and rivals alike. In the course of training, the human athlete forms relationships with other human beings and develops inside jokes and lingo. It is often these distinctly human relationships that are uniquely appealing about sports, for both athletes and spectators alike. Human spectators enjoy hearing about great rivalries, stories of overcoming the odds through hard work, and the intrigue over which players and/or coaches might be swapped around in a particular year. Human spectators have often participated in sports (even if at a low level) and can sympathize with the unique challenges and skills displayed by human athletes.
Over time, human athletes develop a shared history with other athletes, fans, and competitions (e.g. the tennis player who has lost Wimbledon three times but keeps coming back for more). A robot runner might be equipped with aerodynamic “skin” that allows it to outrun human beings, but it has not participated in the training camps or sporting traditions that human athletes have. It does not have the same shared history as human athletes. A robot athlete might trace its history back to the development of computing and the wild dreams of a few entrepreneurs and engineers, but it does not “grow up” like a human being does or face human illnesses and setbacks. There would be a team of humans who support the robot—such as the robot’s inventor or the mechanics who “fine-tune” the robot’s program for better performance—but fundamentally, the robot athlete does not share human life experiences or physical limitations. This lack of shared experience inherently limits the extent to which human spectators can sympathize with the achievements of robot athletes.
When a human athlete joins a sports team (e.g. the All Blacks in rugby) or competes in a particular competition (e.g. the Olympics), they are not merely taking on an arbitrary affiliation, they are joining a long legacy. They might aspire to exceed the achievements of past athletes who came before them, live up to the values of legendary teammates, or avenge a bitter series of losses. In contrast, if a robot athlete could join the same sports teams and competitions, it would not be able to look back to a long history of robot triumphs. The robot athlete might be able to start a new athletic legacy, beginning in the year 2024, but it would not be able to situate itself within centuries and millennia of human athletic traditions and histories. For instance, in the National Football League, the New England Patriots have a history as one of the most successful teams. Football players who join the Patriots understand that they are joining a long legacy of athletes like them, while teams who play against the Patriots often receive additional support from fans who are inclined to favour the underdogs. A football team composed of robot athletes would not share as long of a history. Even if a robot football team was allowed to play against human teams, it would not be one of the “original” teams; it would always be marked as an outsider in a human sports league. This might reflect prejudice, but the reality is that history provides human athletes a bulwark against robot replacement.
Sports offer a venue where coaches and athletes are empowered to exercise agency, develop character, and form meaningful, reciprocal relationships with other humans. These values are sometimes difficult to experience in other areas of modern life, especially for many working class people. For instance, a taxi driver who volunteers as a boxing coach might gain a sense of meaning from working with athletes and guiding their development. At the gym, the taxi driver can be a leader who earns respect from human athletes, even if the taxi driver does not necessarily feel respected in other areas of life. The taxi driver is not a mere object acted upon by impersonal, systematized economic forces; the athletic coach is a subject who exercises agency in the lives of others. Humans have an emotional need to form relationships that involve both receiving and contributing support to other humans’ lives, but this need is sometimes left unsatisfied by computerized, modern life. This need is unlikely to be entirely met by robot athletes, even though we might find great value in human-robot relationships. One of the greatest joys of sports involves experiencing the feeling that “my life means something to someone else,” a core desire of human nature. To be human involves exercising agency in our individual lives and our communities alike, experiencing the blessings of human connection and care, and weaving ourselves into the unfolding narrative of human existence.
Can robot athletes replicate “humanness” in sports?
It could be argued that robot athletes can replicate many of the traits that make spectating sport appealing. For instance, robot athletes might develop their own histories, traditions, and lore. Nevertheless, human spectators would not emotionally sympathize with robot athletes the way that they can with human athletes. Although humans have a tendency to humanize non-living things and animals, the physical traits embedded in human nature limit the extent to which humans might sympathize with AI athletes, or view them as equal competitors. Many humans have played sports (even if at a low-level) and can sympathize with the physical limitations that human athletes have. Human spectators derive value from following the human experiences, challenges, and excellences of their favourite athletes.
Consider the example of Lionel Messi, the famous professional football player who grew up in a working-class family, suffered from a hormone deficiency that restricted his growth, and worked hard to rise through junior athletic leagues before becoming one of the world’s most revered athletes. Messi’s story of the rags to riches underdog who achieves greatness through grit and determination inspires many of his fans and would be impossible for a robot athlete. A robot athlete would not have been born to blue-collar human parents, battle human medical ailments and injuries, or inspire the proud fans of a nation and world. Although we might engineer AI athletes that have “analogues” of these features of human athletes, human spectators would not necessarily identify with AI “analogues” to the extent that they identify with other humans who share their life experiences. For instance, how could a robot be engineered to grow up with a single mother in an impoverished neighbourhood and overcome the odds? Robots might replicate some human experiences, but by definition they could not replicate all human experiences, including some of the most relatable experiences (e.g. growing up in an impoverished neighbourhood and overcoming the odds), and that limits the ability for robot athletes to provide the same value as humans.
Nevertheless, there are several clear objections to my arguments. Although there might be positive features of sport bound up with our humanity that robot athletes cannot replicate, robot athletes can bring other valuable features to sport, so some might argue that watching robot athletes could be as valuable or more valuable than watching humans. For instance, in violent sports such as mixed martial arts, having robot athletes compete could alleviate the concern that spectators are watching humans engage in health-threatening activities. Robot athletes might perform impressive, crowd-pleasing moves that humans cannot. Humans could watch a more appealing sport without facing the moral ambiguity of sanctioned human violence. Nevertheless, if part of the value of sports involves developing and displaying the full extent of human capabilities—even when those capabilities include tackling, wrestling, and punching—then replacing human athletes with robots would inevitably lead to some loss of value. There would no longer be the opportunity for consenting humans with particular inclinations or skill sets, no matter how violent, to develop pride and excellence through sports. Moreover, some societies might derive value from developing even violent athletic skills in their human populations—perhaps those skills have military value or are correlated with desirable personality traits. However, humans who train with robot athletes might plausibly improve their athletic skills, just like human chess players have learned from competing with computers. It is possible that the desire to outperform computerized capabilities will encourage humans to develop new techniques and strategies when they play sports. Although it would be unwise for robot athletes to replace human athletes, the advent of robot athletes promises gains and losses for humans.
Examining human-robot “teams”
Up to this point, this paper has largely engaged with the possibility that human athletes would be completely replaced by robots. However, what if existing sports teams are permitted one robot member, new sports where humans collaborate with robots spring up, or entirely robot sports are created? Some human athletes and spectators might plausibly welcome robots, and the introduction of robots might not undermine the histories and traditions of human athletes. The rules of sports often change, and it is entirely conceivable that the rules of sports could change to accommodate robot athletes. It is possible that training with or against robots could help human athletes develop superior skills; imagine a superhuman wrestler who wins every competition and develops a superior level of skill after competing with robot sparring partners. Moreover, if robots competed directly with humans in major competitions, then the differing capabilities of robot athletes could still lead to unfairness or a lack of interesting play. Nevertheless, it is important to remember that the purpose of sports is not merely competitive; sports also serve as playing fields to examine and develop human values.
Although AI appears increasingly capable of performing human intellectual work, AI will never fully replicate the values contributed by human athletes. Even if robot athletes achieve superior performances to humans across many domains, there are some human skills, challenges, and values that robot athletes cannot completely replace. Human athletes develop shared histories, face human challenges, and make mistakes that human spectators sympathize with. Even a perfect robot athlete would not be able to fully replicate all the values that human athletes display. We celebrate excellence, but excellence is impossible without human imperfection.