Is Artificial Intelligence Turning Athletes Into Robots and Robots Into Athletes?

Michael Lewis’ book Moneyball popularized the use of artificial intelligence in sports. A cocky general manager running a low budget baseball team finds great success joining forces with a nerdy quant, using data rather intuition in making personnel and strategic decisions. Machines analyzing numbers turn out to be better than experienced human scouts using intuition to choose players to draft, and better than old school managers in suggesting whether to steal second. The use of artificial intelligence in sports was novel when Moneyball was published in 2003—now it is ubiquitous in every decision in every sport.

How Extensive Is AI Use in Sports?

If anything in a sport can be measured, it is being measured—and then used to train athletes, inform coaches, scout prospects, entertain fans, and inform betting.

Sophisticated data interpretation technologies are the levers of advantage sports teams search for, from quarterbacks reviewing defensive coverages on sideline tablets to baseball teams turning to AI for management decisions during games.

What Are "Advanced Statistics"?

Sports have always been a gold mine for statistics lovers who pore over box scores to ponder batting averages, points per game, yards per pass, etc. But descriptive statistics (eg, means, percentages) mostly described past performance and are only a rough guide to the future. Advanced statistics take advantage of modern computing power, extensive data sets, mathematical modeling, and fancy algorithms to go beyond the descriptive by uncovering crucial past patterns that predict future probabilities. The scope and seeming precision of prediction is astounding: eg, Apple TV’s Major League Baseball (MLB) programming employs nVenue statistics to display the probability that a given batter will hit a double when he is facing a 2-0 count, against a fresh lefty pitcher, in a given stadium, in the eighth inning, with men on first and third, in 63 degree weather, and countless other variables.

How Successful Are Advanced Statistics in Predicting Sports Outcomes?

Statistical prediction has greatly increased its scope and accuracy since the primitive days of Moneyball. Much more data are now available, computing power has exponentially increased, and algorithms are much improved. Motion tracking turned static AI into dynamic AI. Everything that moves in every sport can be measured, analyzed, and reported instantaneously—the speed of a tennis serve, a pitcher's fastball, a hitter’s bat, a sprinter's kick; the efficacy of different formations in every team sport; whether a batted ball would be a homerun in each major league stadium. It is a good guess that any action in any sport you can think of is being measured, fed into a complex algorithm, and used to guide training, coaching, and gambling decisions. AI cannot always pick winners, but it can predict fairly accurately the probability of who will win, not only at the beginning of the game but before every play.

How Does AI Affect Athletes?

It both helps and hurts them. Measurement guides training and improves performance- how to move your hips to get more speed and spin on a fastball, which weights to lift to get more spring in you jump, how many seconds before releasing a pass, etc. Athletes of the past who shaped their bodies and learned their craft intuitively would have great difficulty competing with athletes of the present who are precisely toned and finely tuned machines. But there is a cost in safety and spontaneity. Thirty-five percent of major league pitchers have undergone "Tommy John" surgery because they have been trained to throw fastballs that create stress that is beyond the tolerance of their elbow ligament. And athletes, even young ones playing sports presumably just for fun, must undergo grueling training regimen that make sport more grind than game.

AI is also beginning to pervade juvenile sports. Countless companies now offer AI models specifically trained to track, synthesize, and coach youth athletes at all levels of mastery. Apps like GameChanger monitor Little League baseball games as if it were the World Series. Top performers become encouraged to exceed safe pitch counts and underdeveloped players get increasingly sparse playtime. These decisions only accelerate the rising prevalence of injuries as well as the overly managed and overly competitive nature of youth sports.

How Does AI Affect Coaching?

In many ways AI replaces the intellectual aspects of coaching. There is little room left for intuition and experience when the statistically probable outcomes of every play call have been calculated in exquisite and exhaustive detail. There is already a strong trend in major sports leagues to hiring younger coaches willing and able to work with artificial intelligence in making decisions. In an extreme experiment, human coaching was made obsolete entirely. The Okland Ballers, an independent professional baseball team, allowed trained AI models to take full control of a game’s live decisions. From the lineup to stealing decisions, all core aspects of coaching were outsourced to the model, which rendered verdicts through a tablet in the dugout. In the end, the Ballers—and their AI manager—bested their opponents 3 to 2. While still early days, coaching is among many other professions that are to some degree at risk of replacement. Of course, it remains to be seen whether AI can give a rousing halftime speech.

How Does AI Affect Officiating?

Video replays made clear what athletes, coaches, and fans have always known—to err is human when it comes to officiating. AI is more accurate in calling balls and strikes, first downs, in or out in tennis, hits in fencing, and performances in gymnastics, diving, and ice skating. Human officiating may eventually be replaced altogether or limited to situations too complex or novel to be easily reduced to numbers.

In an effort to prime baseball fanatics for the introduction of the Automated Ball-Strike Challenge Systems (ABS) system in the coming seasons, televised accounts of MLB games began to display graphics outing human umpires for mistakes during the 2024 season. This caused fans to become increasingly critical of officiating staff. Whether it be Video Assistant Referee systems employed on the soccer pitch, ABS on the baseball diamond, or Hawk-Eye Systems on the cricket grounds, AI can now make almost all the calls.

How Does AI Affect a Fan's Experience?

Ai increases the cognitive interest of sports, but probably at the expense of its emotional impact. Some fans love advanced statistics and become addicted to studying them. Others find it a mechanical exercise that distracts from the human drama of the game. As AI attempts to pull patterns from the seemingly random, the magic of sports can feel lost. “Miracle,” the 2004 film recounting the US Men's Hockey Team's shocking victory over the far better Russian team, would have been far less dramatic if AI had indicated in advance that the odds were far greater than the 1% alleged by coach Herb Brooks in his famed speech.

How Do Robots Compare to Humans As Athletes?

It is still early days in the development of robot athletes. Today’s robots cannot compete with skilled humans in any sport. Autonomous robots compete with each other in soccer, but lack agility, coordination, and a sudden burst of speed. Robots also have a hard time adapting to an opponent’s adaptation to them, particularly on the fly. Robots are still slow runners. The best robot times for a 100-meter sprint are 16 seconds by a 4legged robot and 25 seconds for a bipedal one, the best 1500 meter about 6 and a half minutes. Using precise sensors and machine learning, robots can consistently hit every shot from any position on a basketball court—but so far, they are much worse than humans when in motion. Robots are already very good hitters in baseball but only for pitches under 30 mph. There is no room for complacency. Robot athletes may eventually surpass us in every physical way as much as they already surpass us in almost all cognitive ways. They will have unbeatable advantages in consistency, precision, strength, and endurance. Machine learning on vast data sets will undoubtedly improve performance exponentially in every sport. But I doubt this will attract much fan interest. Humans are still interested in human chess championships even though no human will ever again beat a good computer chess program.

Concluding Thoughts

Perfection and predictability make sports (and life) boring because they reduce the human drama which relies on spontaneity, imperfection, and unpredictability. In perfecting human athletic performance and making it so predictable, artificial intelligence dehumanizes sport and turns athletes into robots. AI's rapid transformation of sport during the past 25 years is just the canary in the coal mine. Soon everything human that can be digitalized will be digitalized. Decisions guided by human intuition judgment, and experience will be devalued. Numbers will rule and silicon chips will become rulers.

Dr Frances is professor and chair emeritus in the department of psychiatry at Duke University.

Mr Frances is studying finance at the University of Southern California.