The topic you knew was coming- AI & Golf Course Design

[cary lichtenstein]

This topic is interesting for what it represents at the macro level for humanity. I’m not sure if we (humans) are appropriately grappling with the question at hand: what do you want in life?

Advancement always has innovators, early adopters, mainstream, and curmudgeons. Having an innovator or early adopter tell me that the new thing is the better thing is boring. Someone explain to me not how something is better. Tell me why I should care if it’s better. Optimization is so…dull.

I think I have mentioned here before that my wife was an art major.  Before I met her -- which was just before I built Pacific Dunes, coincidence or not -- I had little or no contact with the art world and rarely thought about my business in terms of "art".  Luckily, now, we have a few friends in the local art community, and a couple of them are golfers, so they are very interested in what I do at an artistic level, and it has led to lots of interesting conversations.


Just last night over dinner my wife and I were discussing Picasso and she said that she [paraphrasing] "was okay with his cubist period because he had proven his abilities with conventional forms beforehand".  I objected to this as snobbish, asking what difference it made to the observer of the art whether the artist was truly talented or just lucky?  But Jennifer has had classes in philosophy of art, unlike myself, and for her it was not even an argument that true art could only occur through the conscious application of observation and talent.  An artist "getting lucky" and producing something cool was not REAL art.


She would say the same for AI.  Maybe it will produce a great course, but does it really KNOW what it's doing?  What's the next one going to be like?  More likely it will just look for good spots to place C. B. Macdonald's templates, and there will be long green to tee walks!


But seriously, you could program it to recognize good spots for 1,000 of the best golf holes on earth and insist that it keep the green to tee walks tidy and move earth if necessary to achieve that . . . but that would not be the same thing as what Bill Coore does, or what I do.  You would be unlikely to get anything that was truly original.  Maybe it would be more efficient than the average designer's work, but I had a good guffaw at Anthony's idea that the cost savings would be passed on to the consumer.  That's not how the world works.

Tom:
I disagree with your wife. I'm an artist too and I hate Picasso, I think his body of work is ugly. I think the same thing of Basquait. I think great artists tell a story and I prefer a beautiful one. Golf course design is art and great design should be very eye appealing as well as challenging but not penal, because penal is never beautiful.

[Kalen Braley]

Cary,

I certainly agree with your first bit, ART is soooo subjective, its seems a bit of folly to say what is and isn't "real" art.

However as to your 2nd part, Oakmont and Pine Valley are often described as very penal and they can easily eat top players proverbial lunches, but from the countless pics and video I've seen they both look amazing beautiful.

[Charlie Goerges]

I think we're missing the point a little bit on Tom's post. It's about intentionality. Picasso was a talented and skilled enough artist to have done whatever he set out to do. His cubist period was as intentional as his earliest work. That skill and intentionality is a part of the art. The fact that my kindergartener could have done it is a non sequitur. My kindergartener had no skill or intentionality.


An algorithm also has no intentionality. I feel like the best that can come from an algorithm is artwork, not art. And that's fine, in my life, what I produce is really artwork rather than art, at least most of the time.

[Anthony Butler]

I think we're missing the point a little bit on Tom's post. It's about intentionality. Picasso was a talented and skilled enough artist to have done whatever he set out to do. His cubist period was as intentional as his earliest work. That skill and intentionality is a part of the art. The fact that my kindergartener could have done it is a non sequitur. My kindergartener had no skill or intentionality.

Agreed - when you say Picasso was a poor artist, you lose all credibility - at least in the field of art criticism.

Picasso is like the Tom Brady of Artists.. He also had 3 Hall of Fame careers except his career lasted 70 years not 23..

1.Blue/Rose Period.
2. Cubism > Guernica.
3. Post WW II

[Anthony Butler]

I have given this a couple of days of thought.


First, let's all admit there is an ENORMOUS amount of hype around AI right now . . . story after story in every kind of media about how it is going to change every aspect of life.  Why all the stories?  Because a few Silicon Valley companies are trying to cash in on the hype and ramp their valuations to the moon.  They want to live in those mansions NOW, not in ten years IF the technology proves itself.


Will AI have that kind of effect?  In some businesses, maybe yes; in others, no.  Remember all the hype around self driving cars?  Well, artificial intelligence and enormous investment still haven't produced a self driving car that is reliable enough to deliver on those promises.  (Time to change the subject . . . and they are very glad they didn't label that "AI driving".]


For golf course design - or art - AI is more likely to succeed, because success is mostly subjective -- nobody is going to crash and die -- and only a small percentage of the users know the difference between really good work and schlock.


But, AI is not going to make it cost a lot less to build a golf course.  You might cut out one or two people in the architect's office, and you might cut out the best shapers in favor of guys sitting on a computer . . . in the same way that drone warfare has cut out some pilots, but hasn't seemed to lower our defense budget.  And if you believe that a guy sitting in his basement on a computer, or the computer itself, is going to produce BETTER work than a guy sitting on a machine at the site of a future golf hole, well, then maybe AI will take over golf course design.  But only because of my previous paragraph.

Significant cost savings may only be realized when the entire course delivery chain is both digitized and ML-Enabled. So AI would design or "assist' in the design of the course. The final approved layout and routing would then programmed into GPS connected machinery that would shape the holes (and eventually the green sites).


That technology already exists on farm machinery. As you mentioned Tom, even if these vehicles are unmanned or 'self-driving' the risks are much lower than allowing Musk-built vehicles to terrorize our nation's highways and kill the occupants and innocent bystanders.

[Kalen Braley]

Anthony,

I think the biggest obstacle to self-driving cars is humanity itself.  All of the data we have on them so far shows that they get in accidents at far lower rates than people do.  But all it takes is one bad wreck by a self-driving vehicle to declare they all must be shut off, imagine if same standard were held for human drivers.

P.S.  When/if AI robots or equivalent were running the show in the dystopian future, based on metrics in the aggregate one thing would be certain, humans would never be allowed to get behind the wheel.

[Charlie Goerges]

I just got started reading The Culture novels by Iain M. Banks (80% of the way through the first book, Consider Phlebas). I'm not making any pronouncements about quality etc., but I'm fascinated to be reading a non-dystopian sci-fi book about AI. Most of what we see in media about the far-future of AI is it deciding it doesn't need us and tries to eliminate us. This book considers the possibility that it wouldn't go down like that. In the book, the societies outside of the AI-led Culture think humans are basically pets, inside the Culture society it's a little more complicated than that. If you like sci-fi, it may be something someone would be interested in. (I'm not a big sci-fi reader FYI)

[Anthony Butler]

Anthony,

I think the biggest obstacle to self-driving cars is humanity itself.  All of the data we have on them so far shows that they get in accidents at far lower rates than people do.  But all it takes is one bad wreck by a self-driving vehicle to declare they all must be shut off, imagine if same standard were held for human drivers.

P.S.  When/if AI robots or equivalent were running the show in the dystopian future, based on metrics in the aggregate one thing would be certain, humans would never be allowed to get behind the wheel.

Yes, self-driving car accidents are the new shark attacks in terms of headline-grabbing events.

That doesn't change the fact Elon Musk is still an a__hole.

[Matt Schoolfield]

I live in SF, in a testing neighborhood for Cruise and Waymo. I've ridden in driverless Cruise vehicles more than a few times, and yet I am generally skeptical of self-driving cars' viability. I really think self-driving cars are not a useful proxy for the uses of machine learning with regards to golf course development, because our automobile-focused transportation infrastructure is inherently broken and dangerous, where society seems to care much more about uninterrupted throughput than human lives.

Automatic mowers, for example, could simply be pre-programmed, with fail safes, and wouldn't even need machine learning involved much at all. If it did have ML involved, it would be much more akin to Andrew Ng's self-driving busses with preassigned routes, than general self driving cars.

Again, I think machine learning applications could assist in development of golf courses throughout all the stages. I just think it's important for folks to remember these things take time, and models are trained to do a specific task. Generalized intelligence is still wildly theoretical.

If I had a few spare million dollars, one application I've had in my head for some time, is to train a model walk through topo maps, and find a piece of land similar enough to Augusta National to create a copy for the masses to be able to play. That's the type of application I think would be useful. Another application would just be having a reasonable, if not genius, artificial agronomist to chat with.

Artistic layouts and subtle strategic thought I don't think is something we should be worried about just yet. The types of games that algo's can play are all perfect information games (Go, Chess, etc., though I've heard it played Diplomacy well, but I'm skeptical), and all the artistic creations currently offered do still lack a lot of subtleties (see: it can't draw hands) that are exactly what GCA fans are looking for. I just think it's easy to imagine a fantastical AI future, but I see a lot of speculation with regards to ML that sounds a lot like the flying cars we were promised in my youth, with no clear path from the tech that exists now to get there.

[Charlie Goerges]

Another application would just be having a reasonably, if not genius, artificial agronomist to chat with.

Actually, this sounds wonderful. Given the vagaries of this type of discussion on here, it can be difficult to get much for specific information from the supers on the board about maintenance or agronomy or whatnot. Being able to feed in plat information and the algorithm being able to look up soil, climate, topography and other stuff and give me a few estimates for the cost of maintenance or appropriate turf varietals etc. would be terrific. (I'm talking in considering new construction, but it might work for existing courses as well.) I understand totally the good reasons why a super might not want that type of information available. If a course's maintenance budget was above what the AI tool estimated, I could imagine a moronic member or committee using it against the super.


I don't know the answer to that sort of problem, but we probably need to start thinking about it, because it's probably not particularly far off into the future.

[Tim Martin]

Again, I think machine learning applications could assist in development of golf courses throughout all the stages. I just think it's important for folks to remember these things take time, and models are trained to do a specific task. Generalized intelligence is still wildly theoretical.

I agree with Matt’s take as instead of looking at it through the lens of nixing people and skill sets view it as “an assist in the development of golf courses.”

[Zack Molnar]

My initial thoughts in this area went less to generative AI being used to develop original designs, and more to being used for restorations. For example, if you trained a model on the designs of a particular architect, could it be used to make suggestions for how a course should be restored based on that architect's tendencies?


I particularly thought about this in regards to recreating the Lido, where I think Tom & team had to make some decisions on the ground that were different than the computer model output because a  particular element/feature was out of scale/out of character with how CBM would have designed it. A model that was trained on CBM designs could find these inconsistencies in a review of the course and provide suggestions for an architect to implement.

[David Kelly]

It seems whenever I go back to a classic course, my favorite holes are tweaked because they were "too severe" "too blind","too steep", or "too quirky", and then holes/terrain like this rarely get left alone on newer courses for all of the above reasons, combined with modern turf speed.

Can you cite a couple of examples of this kind of work on older courses?

We've seen it over the years at Hoylake with the demise of Dowie and Royal and the work of Steel, Hawtree and M&E.

[Tim_Weiman]

This is clearly the best discussion we’ve had in some time. Very excited.

OF COURSE there’s a different standard for new courses. The overhead for innovation and newness is so steep that things have to be good day one. Artificial Intelligence has the ability to scour and combine in a way that even AI’s first shot at something is pretty high level these days. Imperfection is designed out of the product for fear of being imperfect. But AI can’t compete with true creativity…

Old Barnwell had a Little Lido competition where kids got to design a hole to go on the Kids Course. Brian and Blake chose a winner. Without divulging too much, the winner has a ramp and a penguin involved. Suck on that AI.

Ben,


I’m looking forward to the kids course. Glad Nick came up with that idea.

[cary lichtenstein]

This topic is interesting for what it represents at the macro level for humanity. I’m not sure if we (humans) are appropriately grappling with the question at hand: what do you want in life?

Advancement always has innovators, early adopters, mainstream, and curmudgeons. Having an innovator or early adopter tell me that the new thing is the better thing is boring. Someone explain to me not how something is better. Tell me why I should care if it’s better. Optimization is so…dull.

I think I have mentioned here before that my wife was an art major.  Before I met her -- which was just before I built Pacific Dunes, coincidence or not -- I had little or no contact with the art world and rarely thought about my business in terms of "art".  Luckily, now, we have a few friends in the local art community, and a couple of them are golfers, so they are very interested in what I do at an artistic level, and it has led to lots of interesting conversations.


Just last night over dinner my wife and I were discussing Picasso and she said that she [paraphrasing] "was okay with his cubist period because he had proven his abilities with conventional forms beforehand".  I objected to this as snobbish, asking what difference it made to the observer of the art whether the artist was truly talented or just lucky?  But Jennifer has had classes in philosophy of art, unlike myself, and for her it was not even an argument that true art could only occur through the conscious application of observation and talent.  An artist "getting lucky" and producing something cool was not REAL art.


She would say the same for AI.  Maybe it will produce a great course, but does it really KNOW what it's doing?  What's the next one going to be like?  More likely it will just look for good spots to place C. B. Macdonald's templates, and there will be long green to tee walks!


But seriously, you could program it to recognize good spots for 1,000 of the best golf holes on earth and insist that it keep the green to tee walks tidy and move earth if necessary to achieve that . . . but that would not be the same thing as what Bill Coore does, or what I do.  You would be unlikely to get anything that was truly original.  Maybe it would be more efficient than the average designer's work, but I had a good guffaw at Anthony's idea that the cost savings would be passed on to the consumer.  That's not how the world works.

Ask your wife about Basquait's body of work, I'm curious

[cary lichtenstein]

It would be interesting to ask AI to blend in "quirk", given the number of great quirky holes in Ireland and Scotland.

[Anthony Butler]

I'm glad to see this conversation has continued at a low boil during my one-month of self imposed Internet absence in New Hampshire. My few trips away from the Granite State all involved journeys south for various lacrosse events.. these inevitably take place somewhere in Maryland or nearby- including an overnight camp just for goalies.

I told my wife I'm happy to drive my son there, but I'm not hanging around for 8 hours in the sun to watch what is essentially practice. That allowed me to squeeze in a round last week at Bulle Rock - a Pete Dye course in Havre de Grace which I thoroughly enjoyed. Like most Dye Courses, it has a set of challenging green sites, particularly as the greens are quite small compared to the scale of the course. And, despite the fact the greens are not large, if you're not on the right section two-putting can be difficult.

I realized that in some ways what Tom Doak has done during his career is build greens on a larger scale where possible that capture the same qualities as the greens you'll find on Pete Dye's notable courses - at least the ones I've played - TPC Sawgrass, Ocean Course, TPC West, Harbour Town, Long Cove etc.

That mentor or guild relationship would not exist if AI took over the design of golf courses to a large degree... You wouldn't have designers learning their craft from their elders, then providing the public with their own take on what they've been exposed to.


I'm note sure how that applies in the art world, but if music continues down the path of sampling and artificial creation as much as other fields impacted by AI - we may never get further examples like Woody Guthrie > Bob Dylan, Chuck Berry > Rolling Stones or, more recently, the Black Keys channeling Howlin' Wolf. If a human being hasn’t learnt from the ideas and creations of all that have come before, they have no basis from
which create their own vision - and probably no skills or experience on how to make that vision reality.

That I think would be a huge loss.

[John Kavanaugh]

How will we know when an AI architect has been dead long enough to love?

[Anthony Butler]

How will we know when an AI architect has been dead long enough to love?

John - funny comment... do you love your current laptop, or do you think your previous laptop was better?