AI in sports

My introduction to Artificial Intelligence (AI) in sports happened in my teens when I first learned about IBM’s Deep Blue that eventually beat Gary Kasparov in a series of chess games in 1996. I was just starting to learn the subtleties of cricket to prepare for the big leagues, and the idea that a machine could learn the mental aspects of a game – just like I could – truly fascinated me. Now, after almost a decade of international cricket experience, and a much better understanding of the game, my fascination and appreciation of IBM’s mission has deepened.

Since the Deep Blue project, advancements in AI, especially driven by Geoffrey Hinton’s work in Deep Learning and Machine Vision, has brought us closer to building a “human-like” cognitive system. Dr. Hinton’s team used deep learning to identify suitable molecules likely to be effective drug agents from a data set describing the chemical structure of thousands of different molecules. Similarly, in 2012, renowned Super Vision algorithm achieved human-like success in recognizing objects. The breakthrough in achieving drastically better results has taken AI & cognitive computing into its next era – what we, at SparkCognition, call AI 3.0.

Having worked at an AI startup in 1999 and again now, I can attest that the technology and the supporting infrastructure yield results that were not possible several years ago . At SparkCognition, proprietary extensions of deep learning technologies and the IBM Watson cognitive system have successfully and remarkably augmented  security and vibration experts with more than 20 years of experience.

What makes AI 3.0 more promising is its ability to tease-out from experts the knowledge that made them experts. Humans can exhibit expertise, but it is difficult for them to articulate. The machine learning algorithms today can identify the discriminating features and induce patterns or rules automatically to produce an expert system. The progress showcased by IBM Watson in recommending cancer treatments and Google’s DeepMind in learning to play a video game showcase the advancements AI has made since 1996.

So, have the machines that are advancing healthcare and other areas got us to closer to building a machine that can replicate the intelligence of Tom Brady?

 

Many AI evangelists believe that Brady, Ronaldo, Tendulkar and Gretzky have the best pattern recognition at their given sport, but AI is not far from delivering their intelligence via software.

The ability to ingest a variety of data sources* (video, sensors, wearable tech, external factors, stats, soft and hard data), large quantities of training data and self-learning technologies (such as deep learning) have landed us closer to this reality than many might think.

A comprehensive cognitive system, with intuition and reasoning, is the Holy Grail. But we can augment or amplify human intelligence by adopting an “ensembled” application of AI approaches, the right set of training data and domain expertise. 

Today, teams, athletes and ancillary businesses are looking for competitive advantages to create operational efficiencies. Over the next five years (2016-2020), AI technologies will mature and sports organizations will experiment with use cases* where AI can make a difference. In order to be successful organizations must not only embrace these technologies and find relevant use cases, but also develop an organizational culture* to support decisions suggested by machines.

Artificial Intelligence 3.0 is about amplifying the intelligence of elite human experts and minimizing the need to rely on human “gut” instincts or subjective decision making.

Cognitive platforms’ ability to interact in natural language, analyze large volumes of unstructured data, respond to complex questions with evidence-based answers, and discover new actionable patterns and insights empower coaches, scouts, administrators and players’ options to make better decisions – supported by human-consumable evidence* –It is very likely that future coaching philosophies will be determined by cognitive systems trained by sports experts.

Just imagine: the owner of Super Bowl LV thanking its coaches, players and software (all on a first name basis, of course) equally for a memorable season.

*In my next posts, I will discuss real-world use cases of AI in sports and cutting-edge research in this areas.

Usman Shuja is Vice President of Market Development at SparkCognition – a Cognitive analytics startup that is applying Artificial Intelligence to solve cyber-physical security challenges. In this role, he is responsible for developing new markets for Cognitive and AI technologies, including IBM Watson-powered applications.

IoT home security