AWS re:Invent: AI vanishes as
differentiator in sports

Artificial intelligence once offered a competitive edge to the early adopters in elite sport. That period has passed. Teams across codes now analyse their performance with similar intensity, draw from similar tools and feed their planning with cut-and-paste flows of information. Much like science fiction writer Douglas Adams’ legendary answer to the secret of life, the universe and everything – spoiler alert, the answer is 42 – the quest has shifted from getting AI’s answers to learning the right questions.

Even Springbok coach Rassie Erasmus, in an interview for my book, “The Hitchhiker’s Guide to AI”, declared that no elite team could do without AI.

Julie Souza, global head of sports at Amazon Web Services (AWS), has seen that shift from inside the engine room of world sport. She told Gadget during last week’s AWS re:Invent conference in Las Vegas: “Analytics help inform those decisions, yet coaches still carry the responsibility for the choices that shape a game.

“The raw data of sport is x’s and y’s and z’s, degrees of latitude and longitude coming off of the field of play, and it is coming in at tremendous volumes. An average soccer match emits about 3.6-million points of data. A golf tournament, 53-million points of data. Formula 1 has 1.1-million points of data per second off the cars. So, then what do we do with this data?

“It depends on who you are in the ecosystem. If you’re a governing body or a league, use cases like player health and safety, officiating analysis, rule development, equipment changes. If you’re a team, that’s game strategy and scouting and recruiting and training.”

At one time, AI in sports was a proof of concept for the business world in how data could enhance performance. Now, it is a business in itself. Every movement becomes a data point, every data point becomes a pattern, and every pattern feeds coaching rooms, medical teams and broadcast suites.

The value of those patterns appears vividly in making American football safer.

“We’re seeing about 500-million points of data delivered per week through their player health and safety portal, which is a digital twin of the athletes.  That platform is provided to all 32 clubs, and then predictive analytics are placed into this to help understand and alert teams to the players that are at risk of injury. They can then get to those players and change their training, to try to stave off potential injury.

“The first year the portal was put out to the teams to use, there were 700 fewer missed games by players due to injury. There was a play called a hip-drop tackle, when a defensive player wraps his arms around the waist of the offensive player and falls on their legs to bring them down. That was causing a 20x injury rate over a normal pass and run play. Now that the league had that data, they were able to ban the play.”

Even the sacrosanct kick-off rule in American football was changed due to AI revealing a double injury rate and a quadruple concussion rate over normal play.

“The league used AI to simulate 10,000 seasons worth of data to optimise rule changes for two factors: lower injury rate, and higher kick-off returns. That rule change this season has done exactly what they wanted to do. Injury rates came down, normalised with a traditional play. And, as of week nine this season, there were 79% more, which means much more exciting play.”

As Rassie Erasmus has made clear, those who ignore AI pay the price. Souza describes the way availability of data widens the gap between teams that embrace insight and those that hesitate.

“When a data portal was launched in the National Basketball Association, four teams barely used it. Those teams finished at the bottom of the table.”

However, once everyone uses it, surely four teams still have to finish at the bottom of the table? When everyone is using AI, how are teams differentiated?

Her bottom line could have come out of Rassie Erasmus’s playbook: “People always say, well, then data will just take over and make all the decisions in sport. No, it won’t. It’s a tool. These analytics, these insights, are a tool to help inform decision making, but you are still going to have coaches making the decisions.”

* Arthur Goldstuck is CEO of World Wide Worx, editor-in-chief of Gadget.co.za, and author of The Hitchhiker’s Guide to AI – The African Edge.