While Startups like ScoutEdge aim to address data gaps for athletes in tier two and three cities, experts stress the importance of human judgment in sports analysis.
Artificial intelligence (AI) is reshaping sports analytics by providing insights that enhance performance, reduce injuries, and refine strategies. Platforms and applications collect and analyse data from wearable devices and video footage, helping coaches and athletes understand performance metrics and trends.
AI systems process large amounts of data from wearables, video analysis, and historical records, providing a comprehensive view of an athlete's performance and revealing previously undetectable patterns. When combined with computer vision (CV), AI increases the accuracy and depth of performance insights.
In conversation with AIM, David Gladson, co-founder and chief AI scientist at KhiladiPro, said that performance data is used throughout practice sessions and live matches for various purposes, such as identifying players' strengths and weaknesses and addressing those issues during practice. It is also used for tactical analysis to gain an advantage over opponents. Additionally, performance data helps monitor player workload management and track progress during recovery phases.
However, he mentions that "Gen AI on Indian athletes' data is a little hard because we are behind in data management and adaptation to data tools in sports. It does appear well at the top level, but as we move down the ladder, there are visible gaps in data management and decision making."
India is a significant data producer, exemplified by the National Sports Science Database, which athletes and coaches use to shape national performance protocols. AI is also increasingly used to scout talent across India, with the help of coaches and club management. However, upcoming startups like ScoutEdge primarily use data from platforms like Google, AWS and Microsoft.
Satyendra Kumar, founder of ScoutEdge, told AIM that this raises concerns and that they are committed to leveraging data from across the country. Currently, India lacks a dedicated platform, and the company's data mainly pertains to professional athletes already participating in the IPL or representing the Indian national or state teams.
Unfortunately, athletes from tier two and tier three cities, as well as those from rural areas, often lack access to data regarding their performances or the areas where they may need improvement.
ScoutEdge aims to create a talent scouting platform for these unrecognised athletes. It focuses on tier two and tier three regions as a starting point. Kumar asserts that using AI offers numerous benefits, particularly in identifying patterns and trends within data. Manually sifting through data can be chaotic and inefficient. However, with AI, companies can effectively leverage these insights, making it incredibly helpful for athletes.
However, Rushil Munjal, a consultant at Wicky.ai, a sports analytics company that provides AI-powered solutions, has a different opinion on AI replacing scouts and talent spotters. He strongly disagrees that AI can remove human contextual judgment.
He pointed out several reasons for this. First, AI in sports hasn't reached a level where it can surpass the expertise of coaches and scouts who possess years of experience. Additionally, he mentioned that the selection process involves more than just statistics; one cannot accurately assess an athlete solely based on their performance metrics.
Factors such as emotional values, mindset, and morals are also crucial and can only be evaluated through conversations and a genuine understanding of the individual.
"Millions of talented athletes in Tier 2, Tier 3, and rural India go unnoticed. There's no structured, data-driven system to scout, rate, or nurture grassroots talent, leading to a massive gap between potential and opportunity," ScoutEdge said in an email.
When asked about AI tools measuring impact in athletes, Gladson noted that the term "impact" is frequently used across various sports and can be measured from different perspectives by analysts.
"But in any sport, winning is always the objective, so the best measurement of impact still remains in their contribution towards victory instead of individual milestones. Recovery speed and biometric gains are very different for each athlete," Gladson said.
The Indian government has also promoted the use of AI in sports performance analysis through various initiatives that could enhance athlete training, scouting and sports development.
For instance, the KIRTI (Khelo India Rising Talent Identification) program utilises artificial intelligence (AI) and data analytics at Talent Assessment Centres (TACs) across the country to identify and nurture young sporting talent. By implementing standardised protocols and advanced IT tools, including AI, KIRTI ensures a transparent and merit-based selection process for young athletes.
Additionally, the Khelo India program emphasises the development of sports infrastructure that can be utilised for AI-driven training and analysis.
Considering the potential of the sports analysis market in India, Kumar believes that the government already has considerable funding available. However, these startups need to incubate their ideas in some capacity to qualify for this funding. He added that the resources may seem limited initially, but they offer a good starting point.
IIT Madras demonstrates another instance of government support by collaborating with startup companies. They provide working space, government grants, and mentorship. This year alone, they have incubated more than 200 startups. The institute boasts excellent infrastructure and ensures comprehensive mentorship, leveraging the support of its alumni network.
As advanced sports AI systems evolve, they move beyond simple analytics to prescriptive and autonomous decision-making. Munjal believes that while AI tools have proven more capable and faster than humans in making autonomous decisions, they can still make silly mistakes. It's important to note that if prompted incorrectly, these systems may provide two different answers even when the context is the same. That said, the use of AI in sports is still in its early stages, and it is expected to learn and improve much faster than humans, he added.
However, no AI tool is without its challenges. One is data ownership, which is essentially controlled by the associations, broadcasters, or clubs. "Unfortunately, Indian sports data lacks open source, which means it's very less available to the public," Gladson underscored.
He noted that despite challenges in fundraising, the enduring popularity of sports in the age of AI is encouraging. People continue to engage with sports, signalling India's potential to become a true Sports Nation.
Munjal said, "The biggest barrier, in my opinion, is the regulatory authorities. For example, BCCI could sue you if you try to display live stats/graphics from an IPL match. Another side is that tools analysing live sport are mostly used for fantasy gaming/betting, most of which is illegal in our country."
Technical barriers include creating a product that could be used during live matches or races, as well as the high costs associated with these items. Even after addressing these two issues, he reiterated that there was no guarantee your model would succeed in the market.
Mujal highlighted that two key factors are essential: consumers must recognise and be ready to embrace a product, and testing must be confined to a limited scope. There will always be more consumers than testers, resulting in a range of inquiries and requests that may not align with the product's offerings.
Gladson further explained that a key challenge in building a sports company is finding the correct monetary value for our offerings.
"We have identified AI-driven Sports Assessments as a value, conducting the world's First Sports Olympiad by KhiladiPro. The tech we use is driven by computer vision & deep learning on top of mobile-captured video. With so much going around AI, we truly want to be at the centre of cutting-edge AI technology in sports," he concluded.
