AWS for M&E Blog

How generative AI is changing the game

At the National Association of Broadcasters (NAB) Show 2024, held annually in Las Vegas, Nevada, generative artificial intelligence (AI) was a prevailing theme—from technology demonstrations to speaking sessions, interactive installations, and more. On the NAB Main Stage, Thursday Night Football’s Kaylee Hartung led a lively generative AI-focused session comprising panelists Amazon Web Services (AWS) Vice President of AI Products Dr. Matt Wood, AWS Global Leader of Strategic Business Development for Sports Julie Souza, NFL Deputy CIO Aaron Amendolia, and PGA TOUR Senior Vice President of Digital Operations Scott Gutterman.

Throughout the discussion, the group covered a broad range of topics. Wood and Hartung first took the stage for a one-on-one, where Wood described generative AI as the single most significant shift in how we interact with data and information since the internet. He noted energy, enthusiasm, invention, and investment around generative AI unlike anything he’s ever seen. Wood also explained how organizations willing to invest in it now will see exponential ROI in the coming years, and that generative AI holds the potential to increase our productivity a hundred times.

The other panelists then joined Wood and Hartung on stage to continue the conversation. Following are some key highlights and the session recording:

The cloud: a stepping stone to generative AI experimentation

While generative AI was the session’s primary focus, Hartung opened the panel with a discussion around cloud enablement. Gutterman shared how the PGA TOUR is developing a new tool with generative AI that will infuse all kinds of live stats into videos to help players better evaluate their performance.

Amendolia credited a layered approach with AWS to National Football League (NFL) broadcast and storytelling capabilities like NFL Next Gen stats. He also covered operational benefits, noting how AWS makes it easier for the NFL to run the game, describe the action on the field, and support player health and safety. With the NFL’s infrastructure and data platforms powered by AWS, the league can store all its data in the cloud and build on top of it—making it easier to take advantage of emerging technologies like generative AI and introduce new ways for fans to engage with live game content.

“Everybody wants everything to happen in real time today,” Amendolia said. “So, you’re bringing the data, that overlay, and insight straight to the fan, whether they’re watching on a traditional broadcast television, a D2C platform, or app experience. They’re at the game and want to get an AR overlay of what’s happening on the field; this is the type of experience we’re trying to build and bring to our fans.”

Souza expanded on innovations in sports enabled by the cloud, including the NHL’s live cloud broadcast production in March, as well as the role of data in creating these experiences. “NHL went live to air with a game broadcast produced entirely in the cloud,” she explained. “That’s significant, especially when you think about the time savings and carbon footprint. A PGA TOUR broadcast used to require 14 trucks on site to support its events and now, with AWS, the number of trucks required is nine, reducing the overall footprint. An NHL game production previously might emit 2.05 metric tons of carbon dioxide; to counteract that, you would have to plant 34 trees and let them grow for ten years. So, at scale, there’s a real positive environmental impact that live cloud production can have.”

Spotlight on AI’s scalability advantage

Scalability arose as an early topic in the discussion, with Gutterman explaining its importance in evolving the PGA TOUR’s Every Shot Live platform, which incorporates a host of stats into live stream videos to drive what people can see. The platform started four years ago with THE PLAYERS Championship, where the PGA TOUR became the first-ever golf entity to show every shot from every player. For the tournament, it produced 48 streams, 24 of which ran simultaneously, and all of which coincided with the PGA TOUR’s ESPN+ broadcast. While the videos featured natural audio, Gutterman plans to introduce multi-lingual commentary into these streams and expects generative AI to be hugely beneficial.

Amendolia added that one of the NFL’s goals is to build systems with multiple layers of value that can coordinate various types of AI, computer vision, and machine learning, and then add an orchestration layer on top with natural language processing. This goal will prove crucial as the league juggles massive amounts of real-time data flowing in. He emphasized that making this data usable for the production team will require more work in the cloud and training of AI and machine learning models.

“A really critical element is scaling up the viewership, and scaling up the engagement by which fans can invest in a particular game or team,” Wood noted. “What these leagues are building totally changes how you experience a game, and generative AI will help broaden that audience even further with translation and the ability to be able to take what we’re doing and expand it to new audiences; I wouldn’t be surprised if we’re looking at audiences in the hundreds of millions or even billions within our lifetime.”

Exploring the multidimensional impacts of generative AI

As the conversation transitioned to AI’s potential impact on the future of live sports, Wood predicted that generative AI would open the doors to unprecedented audience growth, largely through the experiences it enables. Amendolia then highlighted the near-limitless advantages that generative AI introduces, especially for metadata tagging and leveraging real-time contexts to reimagine how fans interact with sports. He explained how AI can be trained to recognize certain football actions and contexts that would be challenging to tag manually.

Wood added that the level of detail available in tags is more extensive than ever, making metadata tagging a good testbed for experimentation. Reflecting on a recent experiment, he noted, “We tried out a system the other day with an athlete running, and we asked AI to tag it. We expected it would tell us it saw a woman in sneakers running on concrete, but the feedback we got was incredibly detailed, down to the type of concrete as well as the sneaker brand, make, and color. These AI systems are capable of providing a level of detail today that’s never been possible.”

Gutterman mentioned that when people often talk about AI and ML, data and statistics come to mind, but what the PGA TOUR and other leagues are really talking about is how generative AI will give them greater access to content. The path generative AI is taking will result in a series of tools and efficiencies that will improve how teams work and give fans greater depth of view. Gutterman explained that generative AI might allow commentators the opportunity to follow six specific players and generate five talking points after each player finishes a hole in golf that they can weave into the broadcast.

Amendolia also pointed out, however, that leveraging generative AI to its full potential requires a focus on developing and following a set guiding principles as you begin to adopt it. “Speed to market,” he said, “is important for everyone, and it’s crucial to be prepared to put something in the context of the principles you’ve set when you want to act on it.” He then underscored the importance of controlling your own model. “With AWS,” he explained, “the NFL has been able to move that model around without giving away its secret sauce.”

Addressing the AI competitive advantage

When Hartung asked panelists if they could envision a time when teams will use generative AI to gain a competitive advantage, each provided valuable insights. Amendolia reiterated that the NFL’s mission is to ensure that the integrity of the game is preserved. He explained, “The league’s place is not to hold back competition. It’s to enable competition, ensure it’s fair, and on game day, have the right policies that allow for the best possible game and the best possible fan experience, so we are making sure that there is fair access to data across all the teams, that all of them have resources and knowledge.”

Souza added that just like innovations before it, generative AI will soon just become part of the way we do things in sports. Gutterman added that generative AI can help improve the fairness of the game. The PGA TOUR’s new scoring system can calculate within a few feet of where the ball will land from the moment it’s hit. The solution includes 120 cameras across the entire course, allowing the PGA TOUR to move from a manual process of capturing scores and balls around the course to a visually driven process enabled by AI. The solution combines camera footage with radar from tees on the green so that the PGA TOUR can capture ball movement. The organization is building a new room for officials where they will be able to see all these inputs, receive feedback, and then react accordingly on the course. It’s providing a definitive source to see where the ball crossed a boundary line, even if it’s not been televised or shown on a live stream.

Getting on board with generative AI

As the discussion wrapped up, Hartung asked all participants what advice they’d offer to organizations experimenting with AI, and this was the consensus:

Ask the right questions

Gutterman stressed the importance of asking questions to get thinking about the experience you want to create for people, like “What are you creating, and why? What’s your goal?” Similarly, Amendolia said it’s vital to consider whether you’re envisioning the right application and if it’s safe, encouraging the audience to ask questions like, “Will this represent my brand well or operate at scale across my organization?”

Don’t be afraid to experiment, but be patient

Another recurring theme was a willingness to understand how AI is evolving, and experiment while exercising patience. Gutterman expressed, “We’re trying to create meaningful value for our fans, players, and the business overall, and there’s a lot of hype around this [AI] right now, but the real value is in taking your time and making sure that you’re following through.” Amendolia encouraged, “Have fun experimenting. Try this technology and apply it to things in new ways. But, know you can go back to your structure and your guidelines to check your work”

Consider the layers required to execute and make the right business case

Souza touched on the importance of putting thought into data layers when embracing AI—including how you structure data and where it’s accessible. “Starting with the right foundation in place will allow you to add generative AI on top, experiment, and then make a business case,” she shared. “At the end of the day, the application must serve the fans, broadcasters, or players, so it’s crucial to think about the use cases that will be the most meaningful and deliver the most value.”

Don’t discount the role of culture in AI success stories

While technology providers like AWS provide the models, grounding, and guard rails needed to securely leverage generative AI, companies need buy-in from the entire leadership team for success, according to Wood. They also must be able to show how generative AI applications can deliver more meaningful experiences for fans and even the larger leagues. He urged, “Clearly convey the big top-down goals that will motivate your teams and give them the energy to move quickly. That energy will flow through to your partners like AWS and others you work with and get them excited; it will carry you through all the good times and all the hard times.”

Educate and showcase the infinite potential of AI

Illustrating what’s possible with AI is essential to getting stakeholders across an organization on board with generative AI. Those “aha” moments that get people excited about the technology can make it easier to secure financing for further development, Amendolia explained. “The moment that you train the model to recognize your special context as a business, and you say, ‘show me a spike,’ and it shows it, the model starts to get it and that accuracy spikes up,” he said, “You’re like, wow! It understands football, and that’s cool. It’s an awesome experience and not just for the technologists and the data scientists, but for everyone at your organization. They’re going to use AI in the future to do their job more efficiently but also to produce new and novel things.”

Watch a recording of the full panel, and find more information about AWS solutions for sports on the AWS website.

Lisa Epstein

Lisa Epstein

Lisa Epstein is a Senior Industry Marketing Manager at Amazon Web Services.