The term artificial intelligence comes shrouded in mystery, its potential giving rise to both enthusiasm and alarm. It’s so ubiquitous in everyday life — whether it’s ride-sharing apps searching for the best route for your commute home, or taking a mobile picture of a check to be deposited on the spot — most of us assume the devices around us can somehow learn information and adjust in real-time.
How that same A.I. has manifested in golf, a game of countless inefficiencies waiting to be optimized, is a fascinating study. Last month, we wrote about Toptracer implementing its shot-tracing, statistic-producing technology at driving ranges across the world. Other companies such as Arccos are using similar technology to evaluate data from a player’s game. Just one month ago, the company launched an A.I.-powered rangefinder that factors in wind, temperature and altitude to reach conclusions beyond what a human can process. And to think that some of us still locate a yardage marker and count steps.
A.I. in golf, with computers adapting to behavior, seems every bit as predictable as booking a single tee time online and having an ad for that course follow you like a single waiting on a foursome.
But another A.I. implementation has been far more surprising. The technology, which has been slowly developing for decades at some major manufacturers, is being heavily relied upon to produce golf clubs. It’s not as complex as creating self-driving cars, but A.I.-inspired golf clubs are, at least in part, designing themselves and offering answers that run contrary to expectations.
“The process used to be something where a human made design choices and a computer went to analyze it and came back with data where we had to decide what happens next,” said Alan Hocknell, Callaway Golf’s senior vice president of research and development. “Now the computer is running simulations, we are asking more difficult questions, and then it’s making the judgments for itself. We have to be open to the idea that it’s going to come up with something that us humans don’t even understand.
“That is thrilling and frightening all at the same time.”
So, what has A.I. learned on its own that challenges how clubs have been made?
Driver design across the industry has been fixated on larger faces as companies have relentlessly competed to add every mile per hour of ball speed they could. With Callaway’s recently debuted Mavrik driver and fairway woods, Hocknell and his team went against that trend based on super-computing output.
“We were able to create a face that minimized the variation of ball speed, no matter where contact is made,” Hocknell explained. “Because we had this performance in the face, we realized we didn’t need a large body that was less efficient. We could shift to a smaller head that doesn’t pay those penalties aerodynamically. That meant we moved the center of gravity closer to the face, which increased spin consistency.
“It’s all to say that often as designers, we have to make trade-offs. And in this case, everything kind of lines up without having to make sacrifices. This use of A.I. moved our design thinking to a new place where we had to tear down some institutionally held beliefs, but once we saw it being tested, we knew it was real.
“This is taking driver design in an unexpected and new direction. I’m excited about what we don’t know, to be honest.”
Of course, like any other hard good, design and production have to meet in the middle. The A.I.-designed face required an upgrade in material — titanium alloy that could be heat-treated to a higher strength — and creating a lightweight carbon crown allowed proper balance in the head.
“It’s sort of a symphony with artificial intelligence as the anchor,” Hocknell said.
The end result in theory, and confirmed in testing, is that swinging a club with a smaller head and more consistent face will narrow the shot dispersion. Farmers Insurance Open winner Marc Leishman recently noted that it took him all of four practice balls on the range to be convinced that the Mavrik technology could be installed in his bag. The different design has also drawn in world No. 3 Brooks Koepka, who suddenly switched to the Mavrik earlier this month.
As Callaway has used A.I. to go against current trends in driver design, Titleist has used it to enhance many aspects of the brand.
At Titleist parent company Acushnet, Chuck Golden’s title is director of golf club research, but he is jokingly referred to as the “supercomputer” given his expertise crunching numbers. One point he makes is that A.I. can influence more than just the shape of a club.
“One thing we really pride ourselves on is the sound,” Golden said. “The pitch, the volume, the length of time of sound after impact that is pleasing to the ear. I remember watching a TV show about BMW and they have something like a 40-person team dedicated to the sound one of their car doors makes. It’s sending subliminal keys to consumers about the overall feeling of the quality of a car.
“It’s the same thing when someone hits a golf ball. With A.I. we are able to ask the computer, ‘OK, with this design, how would it sound?’”
Golden and Hocknell both make the point that this usage of A.I. is not, on the whole, a new phenomenon. It’s engineering optimization, a process that has been going on for decades. The speed, however, makes A.I. more powerful than in the past.
“At the core of it, the problems we are facing haven’t changed,” Golden said. “But we have the power to give a supercomputer a thousand problems at once. Instead of evaluating one idea in 30 days, now we can do it in 30 minutes.
It’s not that A.I. has taken away the art of engineering. There is still a certain instinct required to build a hard good used by humans. All the technology in the world can’t replace a gut feeling golfers have when they make contact.
“If you came to me with a golf club design idea this afternoon, we know by tomorrow morning whether or not it’s going to be durable, how it’s going to sound, ball flight characteristics … and in the span of a day, we can give it a thumbs up to make a prototype of it.”
That makes Golden’s job one filled with experimentation and possibility. When you can try multiple concepts at once, the pressure of failure is altered. The cutting room floor is full, and that’s a good thing.
Could a supercomputer design a driver on its own? Which is to say, could it learn so much about optimization that a human’s role in the process becomes greatly limited, like, say, in a Terminator movie?
The answer is not certain, but Golden believes engineers play a central role in interacting with A.I.
“I could give an optimization routine to the computer and say we are going to use steel on our next hybrid,” Golden said. “Well if I give it a generic material property for steel, I can tell you the computer’s answer will be wrong, because it’s not going to know what happens when you heat-treat the steel, for example. There’s a part of engineering that a computer can never encompass.”
It’s not that A.I. has taken away the art of engineering. There is still a certain instinct required to build a hard good used by humans. All the technology in the world can’t replace a gut feeling golfers have when they make contact.
But it can make a more efficient game for designers and players alike. What that means in the future will decide how colossal of impact it has on the industry.