This article was originally published on SportsPro Media.

I vividly remember the 2011 sports drama ‘Moneyball’.

The movie is based on the true story of Oakland A’s general manager Billy Beane, who was finding different ways to achieve great results for his team with the least possible expense. His quest leads him to Peter Brand, who develops a method of empirical analysis to find winning combinations of players for his baseball team. It inevitably invites the skepticism of players, coaches, and scouts alike, who had to unlearn what they knew about the game and accept the role of data in their decisions.

But here’s the thing: Moneyball happened two decades ago. The underdog story of the A’s winning 20 games in a row is now an incident of the past.

Today, artificial intelligence (AI), machine learning, and analytics are no longer buzzwords and are very much common parlance within the sports ecosystem. These technologies are continuously advancing and beginning to carry a new set of baggage, including privacy concerns, fair use, enterprise scalability, and not to forget the inflated expectations whenever something new comes up.

AI in sports is a US$3.5 billion market that grows by one-third year after year. Today, sports-based use cases are springing up left, right, and center, with advancements in computer vision, deep learning, cloud computing, and data analytics.

Volumes of data are generated from individual training sessions, player wearables, live match feeds, and social media chatter from fans. Organizations supporting enterprise-level digital transformation are licking their lips while the sports industry figures out how much of it is of actual and enterprise value.

Motivations for leveraging AI

The increase in the volume of both on-field and off-field data generated by various sports organizations has led to a growing requirement to manage this data and analyze it. Due to the increasing demand for monitoring and tracking player data, the number of wearable devices in the sports industry is subsequently increasing.

Therefore, various stakeholders within the sports ecosystem are liaising with tech providers to make sense of this change. Their broad motivations to apply AI include:

Media intelligence applications

Automating existing content operations
Monetising legacy sports content and driving storytelling in sports
Creating personalised content for diverse audiences

Sports analytics

Supporting in-game decisions with data-driven insights
Supporting scouts and coaches with player performance and match insights
Making competing safer for players

Fan engagement

Bringing fans closer to the game through novel and personalised experiences

What ultimately rings true is that the end goal is to scale the solution to an enterprise-grade environment and make it available for different levels of users in the sports ecosystem.

Applications of AI in sports

1. Media intelligence applications

This application class deals with augmenting existing content operations with AI. These operations could include content meta-tagging, creating custom highlights for a game, and day-to-day editorial operations.

For example, the German Football League (DFL), which oversees the elite professional tiers of club soccer in Germany, has created a video data library with more than 175,000 hours of content spanning six decades. Making this video archive searchable is a must for licensing to different brands, advertisers, and media partners, and tagging this volume is not possible manually.

Quantiphi worked with AWS to build a scalable solution to automatically identify and capture essential information around players’ emotions, desired camera angles, and specific events such as red cards or goals.

Quick highlights can be created using AI-enabled editing workflows, which have been learned from editorial actions. For custom highlights, these can be targeted to different user segments depending on the fanbase, interest, and other business needs to be based on that same metadata.

2. Sports analytics

These solutions deal with elevating the game itself. Multiple-use cases emerge with technologies like a player and ball tracking, action recognition, and analyzing data from player wearables. Some of them include AI-assisted coaching; player scouting and recruitment; injury detection and management; team selection and match strategy; and live match analytics.

3. Fan engagement applications

Fans today want to participate, analyze, critique, and connect with their favorite teams and players in real-time. They are becoming true brand partners in this era of the sport. Understanding and leveraging this relationship is vitally important for clubs, which are therefore looking into micro-segmentation and offering personalized promotions with AI.
A large chunk of club revenues comes from season ticket holders. Promoting in-stadium experiences using augmented reality and enhancing in-home viewing with match facts is another area gaining traction with sports tech providers.

The right ethics to build AI solutions

With the emergence of multiple use cases, it is necessary to have a parallel conversation about the ethical and legal challenges of implementing AI solutions.

In sports, AI will collect and process athlete data when analyzing and monitoring performance. Therefore, compliance with data protection regulations is required. However, in this case, the application of general data protection regulations is not very straightforward. It is essential to understand who qualifies as the controller or processor of the data. This will influence respective roles and responsibilities under data regulations.

Primarily for accuracy, the data should be sufficient, high quality, and prevent bias. Any personal data news concerning an athlete must be processed lawfully, fairly, and transparently. An athlete has the right to receive meaningful information on the logic involved in any automated decision-making. Consent must apply a specific, informed, free and unambiguous indication of the athlete’s wishes.

Over time, we could see multiple instances of ‘power imbalances’ between elite athletes and AI systems. We could also witness inflated sports contracts where athletes agree to share their biometric data.

Both the industry and technology providers are testing new ground with AI and sports. Therefore, human involvement and understanding the context of these applications becomes crucial. What goes without saying is that human involvement has to be active and not just a token gesture, which is a sports context still requires the participation of coaches, analysts, or medical staff when an AI system makes a decision.

At the end of the day, the intention is to blend technology with the game to elevate how it is being played, watched, and celebrated, and not build to something for the sake of it.