How the PGA TOUR Is using generative AI to change the golf fan’s experience – SiliconANGLE

Generative artificial intelligence has the power to change almost every aspect of how we work, educate and entertain, and nearly every business is seeking to infuse AI into its business process to create new ways of engaging customers and employees.

The sports and entertainment industry remains one of the most competitive vertical industries. Fans have many choices today and want an experience above and beyond the typical TV broadcast.

Recently, Amazon Web Services Inc. invited a few analysts to the TOUR Championship at East Lake Golf Club in Atlanta to meet with AWS and PGA TOUR executives and experience some of the generative AI-based experiences coming to golf fans soon. The PGA TOUR and AWS entered a 10-year partnership back in March 2021 with a focus on making golf more entertaining and accessible for fans to watch and follow. 

The foundational component of doing this is data. The PGA TOUR has decades of video that was previously on tape and terabytes of PGA TOUR data. It also has decades of scoring data, PDF-based fan guides and other data. The PGA TOUR also captures about 2 million data points at any given time around the course from the 120 to 150 cameras and 36 radars across the course.

All this data is in a media lake with a whopping 25 petabytes and stored in the AWS cloud. The PGA TOUR also recently modernized its scoring system to run in the cloud versus as an on-premises solution. I had the opportunity to interview Scott Gutterman, senior vice president of digital operations at the PGA TOUR.

“The scoring system hasn’t seen a major upgrade in over a decade,” he said. “It was a well-run system, but it was time to upgrade it to use modern technologies to increase the data points we captured. This helps our players use the data and gives our fans new experiences. So, we rebuilt it to run completely on AWS.”

The PGA TOUR has been looking to use Amazon Bedrock for generative AI to tap into this massive data set. Gutterman explained that it has 32 identified use cases for generative AI, ranging from efficiencies for internal staff to data-driven insights and fan-facing applications.

One of the more compelling internal use cases is to use chatbots to tap into the information in the 30-plus years of PDF-based media guides. I asked Gutterman how the data was accessed before generative AI, and he said it was a manual process that involved looking up information and then having to do a manual cut and paste of the information, which is very time-consuming.

One more intriguing fan-facing use case is PGA TOUR’s play-by-play commentary, of which we analysts received a demo. One of the challenges for the PGA TOUR is letting fans follow their favorite players, but the TV broadcast typically only provides commentary for the big-name players and those contending for the tournament. The typical tournament has between 30,000 and 32,000 shots, making it impossible for humans to comment on every shot.

The example provided was a shot (pictured) from PGA TOUR professional Joel Dahmen where he hit a shot and AI commentary was generated. It stated, “On the par-5 16th hole, Joel Dahmen sank a 6-foot birdie putt to move to 10-under par for the tournament and tie for 7^th place with a 2-under round score. Dahmen’s birdie extended his current one-putt stream to 1 hole and brought his one-putt percentage for the round to 50.0.”

The goal is to give commentary that goes beyond just “Shot one was 320 yards and shot two was 150 yards,” by providing greater context. PGA TOUR Commentary is still in beta with the hopes of having it available to the public for The PLAYERS Championship in March of 2025.

Gutterman discussed the value of the application. “This lets us cover a non-linear sport for three and half days until the closing holes on Sunday,” he said. While it’s always important to understand what the leaders are doing, many fans want to know what’s going on over the rest of the course. We’ve had the ability to provide basic text for a while, but the goal is narrative level commentary.”

I asked Elaine Chiasson, global golf team leader for AWS, for her thoughts and what makes Amazon Bedrock unique. “We give a choice of models,” she said. “Amazon Bedrock isn’t just one model, and we are continuously adding models. We have models from partners as well as our own as we want to provide organizations like the PGA TOUR choice based upon their use cases.”

As for privacy, she added: “We do not use our customer’s proprietary data to train our models. Their data is kept in their accounts and is private to each customer.”

Given the PGA TOUR’s experience with generative AI, I asked Gutterman for a couple of pieces of advice to other IT leaders in how to get started with it. “Start with experimentation,” he said. “2023 was the year of experimentation and proof of concepts to try things so we could learn. This year has been to take those learnings and create production-ready tools, efficiencies, and experiences. The goal should be to work generative AI into your software development models so it’s easy to adapt when a model goes from a 1.5 version to a 1.6 to ensure the models will still be useful for your tasks.”

Chiasson added her thoughts on the importance of co-innovating with a partner. “The PGA TOUR is extremely collaborative with us,” she explained. “At AWS, approximately 90% of our roadmap and features directly result from customer feedback, and the TOUR constantly gives us feedback through mechanisms like beta programs.”

Gutterman seconded that when he said, “We are golf experts where AWS are experts in generative AI and machine learning. So, we lean into the relationship for help in executing the ideas we have for our players, fans, and business.”

It was great to see the advancements the PGA TOUR has made with generative AI since I met with the team at AWS re:Invent in November 2023. Though the example provided was golf-related, one could see how this could extend to other industries. A retailer could provide commentary around accessories when someone is buying clothes, or a cruise line could offer up AI-generated images of vacations. The possibilities are unlimited, and one needs the ideas to “prime the pump,” which is what AWS and the PGA TOUR did.

Zeus Kerravala is a principal analyst at ZK Research, a division of Kerravala Consulting. He wrote this article for SiliconANGLE.

Image: AWS