LLMs Will be Like Ozempic for Golf
The Robot Caddie
About a year ago, I got a golf lesson and it didn’t go well. I was out of practice and set to play a round with my brother-in-law in a meetup with family friends. When you’re bad at golf, part of what kills you is all the unanswered questions. Was that the right swing? Why did I miss? Should I be swinging this way instead?
My instructor was sour tempered, and stingy with specifics. Backswing slower, grip the club looser. This might have been good advice, but it didn’t offer me much in the way of why. Perhaps I was a worse student than she was an instructor, but the experience didn’t feel entirely positive. I left a less confident golfer than I arrived.
The round went fine even if I didn’t play particularly well. A year later, I was set to play in the same annual family get together, and was still similarly out of practice.
So I did an hour TrackMan session with an instructor named Rob Boldt. You hit the ball at a screen, it flies towards a virtual target, and the system spits numbers back out at you. Rob’s advice was far more specific than last year’s instructor. He explained that this is because the TrackMan is “objective.”
I played the annual round at more or less the level I’d played the year prior. The main difference was that, this time, I actually recognized my screw ups as part of a pattern.
Rob was fantastic but it’s not as though I can just do a TrackMan excursion all the time. Yet I remained curious about why my swing was what it was. I had decent power, but lacked the ability to square the ball up with anything approaching consistency. Was I destined to always have this problem?
And so I turned to LLMs, feeding the TrackMan stats into GPT. Based on 12 numbers, from one swing, GPT had me clocked. It knew my strengths and weaknesses. It fully understood the specifics of my poor technique. I’m sure Rob could have walked me through as much, but his time is limited. The machine had all the bandwidth in the world to deal with my “over the top” swing, how to fix it, and any other questions I might nag it with.
A day later, my swing was different and self recorded video sent to Google’s Gemini confirmed the change. Swing errors that were decades in the making were corrected in the span of minutes. I’m not saying that I’ve suddenly made a leap from “Struggles to break 100” to “scratch golfer.” I’m just saying that a process that could have been expensive and arduous was instead efficient and relatively cheap. I apply the LLM’s fix, and it tells me whether I’ve actually applied it. The feedback is instant and objective.
Gemini isn’t perfect, but it understands a golf swing. Without getting into the self indulgent details of my own, the machine can easily discern which process is better between two swings that yielded a similar result.
Not every sport is like this. I doubt I could improve much as a basketball player by having someone record my approach in the chaos of a pickup game. But golf is basic physics, against a relatively static backdrop. It’s the sport where it’s pretty easy to conceive of a robot besting human competitors.
My main takeaway from this experience is an expectation that average golf skill should improve in the world, similar to how people have suddenly gotten skinnier in certain communities due to GLP-1s. There are other individual sports, like tennis, where I could envision an advantage gleaned from robot assistance, but golf is the easiest one for me to project.
As with many LLM discussions, the obvious caveat is that an improvement process can’t be LLM alone. I benefitted from a lesson with a person (Rob), and still need to play more to make further strides. But I’m convinced after this dabbling that this is the future, for novices and pros alike.
I am predicting the Golf Revolution, or perhaps decline, if your perspective is that optimization tends to ruin hobbies. A sport for obsessives has been gifted the ideal tool for refinement. I believe it will lead to an overall improvement in skill at all levels, even beyond the standard march of progress we’re already seeing in the other major sports.
Perhaps I’m overly optimistic and you can rebut me here, or elsewhere. Will talk about it more on Chat BCC, interested to hear counter arguments, even if they’re LLM-aided.
Hi Ethan, scratch golfer here (not that it matters). I do think LLMs could enhance golf instruction, but I don't think it will make much of a difference for Average Joes.
For Average Joe, the problem *isn't* that his instructor lacks knowledge of the golf swing. His problem is how own inability to exactly replicate an explosive, athletic movement over and over again. His other problem is that he doesn't have the time and/or dedication and/or money to successfully execute a major swing change, which takes thousands of repetitions.
I don't think Average Joes feeding videos of their backswing to a LLM will produce better results than seeing a swing instructor. Even if the LLM is way smarter than any human instructor, the three dimensional in-person observation contains exponentially more data than the two dimensional video recording.
The group of golfers that is best suited to benefit from LLMs is the group of golfers that already has access to the technology that would be easily digestible to an LLM: professional golfers with lots of expensive tracking equipment. I can imagine a scenario in which professional golf instructors learn how to leverage LLMs along with existing technology (Trackman, pressure plates, motion capture) to provide better information to high-level players with the physical ability to consistently execute the instruction.
I think you just accidentally leaked the plot to Happy Gilmore 3 (scheduled for 2045)