IBM’s Watson Is Playing at This Year’s US Open

The supercomputer’s newest task is running one of the biggest events in tennis: this month’s US Open championships.

|
Sep 8 2016, 7:29pm

Image: Serena Williams/Wikimedia Commons; IBM Watson/Wikimedia Commons

There are few jobs that IBM's Watson won't do. The supercomputer has created movie trailers, written musical scores, and even goofed around with Pokémon Go. Now, its newest task is running one of the biggest events in tennis: this month's US Open championships.

Every year, hundreds of thousands of people gather at New York's Arthur Ashe Stadium to watch tennis' biggest names compete for the prestigious title. But coordinating an event of that size takes a lot of work, which is why IBM and the United States Tennis Association, the organization that runs the tournament, asked Watson to do a lot of the heavy lifting.

The US Open has launched what it's calling a "cognitive concierge," powered by Watson's AI capabilities. Visitors can interact with Watson (to an extent), through the tournament's app that integrates with Watson's Natural Language API. Instead of poking around the app for the nearest concession stand or bathroom, attendees can ask Watson questions like, "Where can I buy a hotdog?" and generally receive a helpful reply.

"We've been working with the USTA for over 25 years," John Kent, the program manager of Worldwide Sponsorship Marketing at IBM, told me. "We wondered, what can we do, from a fan experience perspective, to make the tennis experience compelling? This year, [the app is] enabled in such a way so that beacon technology helps to ID where on grounds you are, and give you the most precise answers."

The engineering and development teams started working on the project eight months ago, Kent said. Since Watson's platform is now enabled through API, the team was able to use its natural language classifier, which helps to make Watson conversational, to teach the supercomputer about tennis, the US Open, and the stadium venue. Visitors will be able to use the new app for two weeks, the duration of the tournament.

Read more: Watson Made a Movie Trailer, Paving the Way for Police-Worn Body Cameras

This all seemed like a lot of work for a two week championship, but Kent assured me that Watson's presence at the US Open will help to engage audiences in newer and deeper ways. Plus, he said, there are plenty of sponsorship and revenue opportunities that come with a highly marketable app. I asked the USTA how much the project cost, but a spokesperson for the nonprofit declined to comment.

Another benefit, which actually makes tangible sense, is the app's ability to enhance match-watching for people who aren't lucky enough to have seats. With the capacity to hold more than 23,000 people, the Arthur Ashe Stadium is the largest singular court in the world. But sometimes matches are held at smaller arenas, and the US Open isn't always broadcast internationally. So, if someone's sitting at home or in another country, the app can provide a live-feed of match analysis, such as a player's historical performance, which points were won and how.

Kent told me IBM wanted to apply Watson's prowess to solve problems. "We don't do technology for technology's sake," he added. So in addition to its app integration, Watson will also inspect every single photo taken by the USTA, and use its visual recognition capabilities to tag people it was trained to identify. Watson will do the same for video content, only it will also add subtitles and produce transcripts to allegedly make the tournament more ADA compliant.

The USTA didn't have any engagement statistics to share with me, so it's unclear if visitors think the app is a love match. In broad strokes, however, the organization knows that its most popular online and mobile destinations are its score feed, followed by the player draw, and then other content like videos and photos.

Watson also has its limitations when it comes to interacting with app users. Developers are still trying to improve its conversational abilities, so it doesn't have to ask people to rephrase their questions so often. For example, if someone says, "I'm hungry," Watson will be able to deduce the intent of their remark as a question about food venues.

And when it comes to having an opinion about the actual matches, Watson is sadly silent. I'm sure it would have bet on the winners, anyway. Now that would be a racket.

Want more Motherboard in your life? Then sign up for our daily newsletter.