Advantage Watson

IBM’s artificial intelligence skills fill seats at US Open

The stands are empty, but Watson is still there.

This week saw the US Open kick off at Flushing Meadows in New York City. But this year, the US Tennis Association whittled fan attendance down to zero as a result of the pandemic and the need for crowd restrictions. The world’s top tennis players are now playing in front of empty seats rather than the thousands of spectators who normally fill the National Tennis Center each year.

Live sports taking place without an audience is a surreal picture, but one we’ve all quickly become used to. And again, technology is being called on to fill the void.

For close to three decades, IBM has been providing technology for the US Open, supporting the tournament’s online experience and giving fans digital access to one of tennis’ top venues. And during the most recent years, the company has deployed its artificial intelligence (AI) to put a modern IT spin on one of the world’s classic sports; AI-powered algorithms, for example, were used to compose broadcast highlight reels based on crowd reaction. However, the absence of fans at this year’s event presented a problem for some of its AI projects.

The event organizers have turned to IBM’s AI skill set to support sweeping changes to almost every aspect of the competition. For example, matches are being played with electronic line-calling, and players use food-ordering apps for meal deliveries to their suites at the venue.

This year, IBM is using AI and cloud computing to add audience noise to broadcasts based on past highlights. Sports broadcasters, faced with similar challenges in holding matches in empty stadiums, have also tried this approach, notably in football. The NBA and the NFL will also use artificial fan noise for live games this year.

One of the most interesting deployments I’ve seen of IBM’s AI so far, however, is by non-league football club Leatherhead FC in the UK, where the team shot up the table in the year it deployed Watson to help it track player performance.

Fascinating night in my hood to hear about how @LeatherheadFC shot up the table this yr using #AI & #IBM #Watson. Young, visionary managers like Martin McCarthy & players hungry to improve with data are keys to sports success with AI. The future of coaching & player development💪 pic.twitter.com/If0mvMPSKZ

— Nicholas McQuire (@nickmcquire) April 30, 2019

IBM says its project with the US Open is unique, as the pandemic has introduced problems that have forced it to develop new solutions. Necessity is the mother of invention.

The first solution is AI Sounds, which aims to recreate the ambient noise normally emanating from the stadium. According to IBM, Watson digests video from the 2019 US Open and ranks the “excitement level” of various clips, which it compiles into a reel and classifies to give each a crowd reaction score. IBM will supply this footage to ESPN, which is broadcasting this year’s event, to serve it dynamically based on play. It’s at the discretion of the US Tennis Association and ESPN to use the audience noise system when appropriate.

AI Sounds is similar to a solution deployed at Wimbledon, which is also a long-standing IBM and Watson customer. I went behind the scenes in 2019 to see the solution in action (see IBM Serves Up Its AI Progress at Wimbledon 2019). Wimbledon uses IBM’s visual recognition on real-time video feeds from matches, as well as its audio analysis APIs to assess scoring data, player gestures and crowd noise to determine an overall excitement score for each scene within a match. IBM’s AI also automatically generates highlight packages based on these scores to share with digital editors. This has resulted in a 250% increase in the number of highlights packages produced for the tournament’s online channels.

The second solution that IBM has rolled out at the US Open is Open Questions with Watson Discovery, a text analytics and natural language processing service to facilitate tennis debates between viewers on the event’s website. With human support and AI technology from IBM Research, the solution analyses and summarizes the information, presenting it as pro and con arguments. Watson Discovery also analyses “millions” of news and sports articles for relevant information and provides match insights through the US Open website and smartphone app ahead of each match.

The Importance of Trust in AI and Humans in the Loop

Importantly, IBM confirmed that although match insights are generated by AI, all information is editorially reviewed before publishing. AI transparency and, in particular, AI explainability supporting human-in-the-loop strategies have been a big part of how IBM differentiates Watson for several years. Details about how the US Open is deploying the technology are light, but what we’re seeing elsewhere and particularly at Wimbledon, where Watson has been at work since 2015, is the importance of responsible and human-in-the-loop AI processes, which IBM is enabling.

This approach has been benefiting IBM for the past 12 months. According to our Senior Leadership IT Investment Survey, 2020, IBM is the second most-trusted technology supplier with company data in the market.

Big Steps for Watson

Watson Discovery has been progressing rapidly over the past several years, with more than 16 major innovations in the past year alone, according to IBM. A good example is Project Debater, the company’s AI system that can debate with humans about complex topics (see Watson goes Anywhere at IBM Think 2019). It uses Watson Discovery and several components of IBM’s AI in similar fashion as the US Open.

This week IBM also announced another major milestone with Bloomberg Media. The two partners will launch an interactive debate series titled That’s Debatable, to debut on Bloomberg TV on 9 October 2020. The show will highlight how AI can be used to bring a larger, more diverse range of voices and opinions to debates and help enhance the participants’ arguments. IBM plans to use a new technology in Watson called Key Point Analysis, a progression in natural language processing developed by the same IBM Research team that created Project Debater. It’s designed to analyse arguments submitted by viewers and provide insight into the public opinion about the debate.

It will be fascinating to see the impact this event could have on people’s views on the advancement of AI. What’s also important to watch is how this mix of AI technologies will be adopted by enterprises, especially in the areas of information synthesis and decision support.

The recent technology advances in Watson are positive signs for IBM as the race in AI heats up in 2020.