AI and the future of sports

In the aftermath of an Indian Premier League auction, have you wondered why teams battle wildly and eventually end up shelling out massive money for an unknown entity? If you did, you wouldn’t be the only one because the move by the franchise doesn’t align with conventional logic.

Conventional logic, apparently, is too limited. Enter Artificial Intelligence.

Humbling as it might be, AI is far better at picking out talent – obscure or otherwise – to ensure it increases a franchise’s chances of winning the title. Moneyball, a book and a film on Oakland Athletics’ inexplicable success in baseball by the use of sabermetrics in 2002, is a nascent version of what is actually being used these days.

“AI can comprehend more information in lesser time than we ever could,” says V Kamakoti, Professor at the Department of Computer Sciences, IIT Madras, and Chairman of AI task force for the Government of India. “The only difference is the consumption of energy. Our brain uses up about 25 watts generally. AI takes a lakh and more. If consumption isn’t a criterion, it can do more computation, and faster, that we ever could.”

AI’s only job is to ensure it completes the task at hand better than it could minutes ago, meaning it is evolving at a rate unusual for sentients to comprehend.

“Say a team says it wants to win the IPL on a limited budget, the AI engine is fed all the data and it’s asked to pick the best squad for the conditions to come. AI does that in no time, and then it narrows it down to opposition, matches, conditions and so on. At this rate, a scout, selectors and a coach are redundant because AI does it faster and evidently better.”

More on this later because it still deals with the sport and the athletes in question without detailing the impact AI has on viewers and their consumption of content.

At a time when a global pandemic has forced everyone indoors, television and OTT (over-the-top) platforms have hit the forefront and they offer an array of data snippets to keep the viewer hooked.

Broadcaster and streaming platforms have always looked for ways to retain the attention, and after having exhausted their options, they gravitated towards AI to offer a solution. It did, by providing new camera angles based on user compulsions, finely curated highlights packages by assessing social cues, subtitles based on the location of the said user and much more.

Recently, head of sports in Star and Disney India – Sanjog Gupta revealed that AI is playing a massive role in their dissemination of Indian Premier League content. “You will see over the next 12 months the quality of replays required to make the decision improve significantly. What will happen with better technology is that decisions will start becoming more and more accurate and more precise,” Gupta told PTI in an interview.

“So I think it’s a matter of continuing to invest in technology and continuing to constantly upgrade the quality of cameras.”

While the possibility of increasing the number of cameras per game has occurred to Star as the games are being played without crowds (long-range cameras take up space and that could ruin spectators view at the ground), Gupta noted that all the 34 cameras used during the course of an IPL game can double-up as slow-motion cameras, meaning they won’t need to have more cameras.

But there’s more. IBM Watson, which now handles the automated curation of tennis highlights, also remastered classic Wimbledon matches from over 40 years ago. While remastering old footage is hardly a new concept, when you watch every blonde strand of Boris Becker’s luscious mane bounce while Stefan Edberg forehead crinkles, you can thank AI for the elevated experience.

Imagine the same for cricket. The sweat on the brows of Kapil Dev and his men during their 1983 World Cup victory. Perhaps, AI could even recreate Kapil’s unrecorded 175 not out against Zimbabwe in the same World Cup.

“It’s not entirely impossible,” says an AI expert from Bengaluru, who has worked on recreating tools in the past. “Scorecard data is available and there are some texts on the innings. If we can get people at the ground to talk about the knock, including Kapil Dev himself, it won’t be hard to come up with a rendering. It won’t be accurate but it won’t be impossible.”

At this point, you begin to feel like anything is possible.

In the 2010s, a team from Sheffield Methods Institute at the University of Sheffield was working on a research paper titled: Formula for success: Multilevel modelling of Formula One Driver and Constructor performance, 1950–2014. Drab academic title aside, their readings didn’t enthuse those in the business of Formula One because the most technologically advanced sport in the world was being told by technology that the sport has too much technology.

Dr Andrew Bell, the team leader of the F1 project, noted that on average over the period 1979 to 2014, 86 per cent of the performance stems from the car/team and 14 per cent from the driver. In 2018, it was down to around 10 per cent. Those numbers have dwindled further since. What that means is that the oft-criticised aspect of Formula One being more about the car than about human skill isn’t unfounded.

Lewis Hamilton, especially, wouldn’t be too happy with this because as far as this research paper is concerned, the seven-time World Champion is the twelfth on the list of greatest drivers of all time. Juan Manual Fangio is numero uno followed by Alain Prost and Michael Schumacher.

In fact, Fernando Alonso is the highest-ranked among those in the current field, and he’s sixth on the overall list.

The process began with Machine Learning (ML). The scientists let the machine decipher data from all the algorithms and statistical models fed. Once the data was deciphered, AI narrowed the field and acted on the specifics in need.

Think of ML as a child offered pieces of information. Eventually, the information becomes the basis for deeper thought and quicker learning. AI is then the child’s ability to act on the available information in the most efficient way possible.

Essentially, AI mimics humans’ ability to sense, think and, sometimes, act. In this case, if you had every last strand of data on Formula One, and if the question posed was ‘who is the greatest driver of all time?’, and if you were purely rational, Fangio would be the answer.

But had AI been only about ending barroom brawls, sport and its purveyors wouldn’t have consumed it as greedily.

From the time IBM’s Deep Blue defeated world chess champion Gary Kasparov in 1997, it was obvious that sport was going to milk the data-crunching system.

“AI is a big deal,” says Ramji Srinivasan, the former strength and conditioning coach of the Indian cricket team. “At the rate at which AI is going, fitness tests will be redundant in five years. All the data a coach/ trainer needs is perpetually available to them because all athletes use wearables now.

“These wearables also create bespoke workout routines and diets for the players. This individual data is then fed into a system which is built to compute for the team, meaning how will a team benefit from his/ her training, diet and many such parameters. You can even pick teams with this data,” he says.

“In a few years, a chip will be inserted into an athlete’s body and his every move will be monitored,” he adds.


Original post:

Leave a Reply

Your email address will not be published. Required fields are marked *