Bengaluru: Vidit Gujrathi, the 31-year-old Indian grandmaster, expresses his thoughts with remarkable speed as he discusses his passion for hackathons, app development, and artificial intelligence tools. Recently, he received recognition from NVIDIA for his innovations. Gujrathi has developed two notable applications: one for playing chess without sight and another that presents puzzles from actual tournament scenarios. His most recent project, named Kibitz, has attracted interest from tech industry leaders and prospective investors.
Event Context
Vidit has always had a keen interest in technology. Reflecting on his earlier days assisting Anish Giri, he recalls experimenting with various software but felt constrained by his limited technical skills, particularly his lack of coding knowledge. “In 2023-24, with the advancements in LLMs, I became more hands-on and began creating innovative applications. It felt like magic,” he shared with HT.
Recently, during the World Teams Rapid and Blitz Championship in Hong Kong, Vidit stumbled upon a hackathon submission call from NVIDIA on X. “It lingered in my thoughts. Post-World Teams, I allocated 75% of my time to chess and 25% to building an AI tool, motivated by a desire to tackle a genuine issue rather than merely submitting an entry for the hackathon. I’ve always felt that chess engines fail to capture the grandmaster’s thought process, often leaving spectators baffled by seemingly simple move predictions. I decided to train a model from scratch to reflect how a grandmaster thinks, and I was pleasantly surprised by the concept’s originality—especially given that serious broadcasters haven’t explored it yet. The model I developed, Kibitz, predicts the moves grandmasters are likely to make and can be utilized by broadcasters alongside AI commentary.
Currently, Kibitz achieves a first-move accuracy of 56% and an 86% accuracy for the first three moves. Vidit is optimistic about enhancing its performance: “I tested it on an archived game set, which was quite limited. During the recent Grand Chess Tour Rapid and Blitz, I achieved a 60% accuracy for the first move in rapid games, but I aim to assess it on a broader data set to see how reliably it performs.”
His technical skills have been honed through self-education, encompassing blog articles, academic papers, and YouTube videos by industry leaders. “I actively follow top AI and ML experts on X and found Andrej Karpathy’s YouTube content invaluable for grasping fundamental concepts like neural networks. I engage with the latest research and seek to comprehend the foundational principles. Recommendations from my X followers, such as Paul Graham, an influential figure in the startup ecosystem, also guide my learning process. I strive to familiarize myself with available tools, experiment with them, and learn through practice.”
Despite his active participation in tournaments, Vidit remained intrigued by the intersection of technology and chess. In preparation for the 2024 Candidates tournament, he enlisted software engineers to assist in developing new tools. “I explored forums and consulted domain experts to leverage technology effectively in my preparation. For instance, while many chess tactics platforms focus on immediate gains, elite chess demands a different training approach, often distinguishing between equal and slightly superior positions rather than just winning outright.”
Looking ahead to the Global Chess League in September, Vidit has reduced his competitive appearances but is still weighing a few tournament invitations. With a current world ranking of No. 33 and an Elo rating of 2697, he continues to balance his chess career and technological pursuits.
Player Focus
He envisions a future where he will naturally take on a mentoring role for emerging chess players. Excited about the potential of technology and artificial intelligence, he aspires to travel to San Francisco, the epicenter of generative AI.
“Chess is inherently a zero-sum game where my victory means someone else must lose. However, in the realms of technology and AI, I see vast opportunities for collaboration where everyone can thrive together.”
Match Outlook
Chess players were pioneers in facing the emergence of artificial intelligence. Even before Generative AI became prevalent, they had begun training with machines capable of faster calculations and deeper insights than any human. A pivotal moment occurred in 1997 when IBM’s Deep Blue defeated Garry Kasparov in their rematch in New York, signifying a crucial point not just for chess but also for the field of artificial intelligence as a whole.
Vidit’s pursuit of this passion project stems from his decision to step back from competitive chess following the 2024 Candidates tournament. In the year leading up to this event, the grandmaster from Nashik delivered some of his finest performances. By early 2024, he achieved a peak rating of 2747, ranking him No. 14 in the world. “Around 2021-22, I committed myself fully to chess for a couple of years. I closed my YouTube channel and deactivated my social media accounts to focus intensely on the game. I aimed to realize my full potential without regrets. After qualifying for the Candidates tournament, I found it disappointing when the results didn’t meet my expectations. I realized I no longer had the drive to compete with the same level of dedication and effort. Burnout set in, and I began to ponder my next steps outside of vying for major chess titles. I struggled to maintain that intense lifestyle. Ultimately, it wasn’t a lack of chess knowledge or technical ability, but my inability to manage pressure and nerves that led to lost opportunities.”
“I have concerns about my chess performance. I believe that my current level is not reflective of my true capabilities, which leaves me feeling dissatisfied. If I’m going to pursue semi-professional chess, I aim for a higher standard. This notion is a significant concern for me. My experiments with technology and AI have emerged from the challenges I face in chess. I can devise solutions that surpass others because they may not even recognize these issues exist. Generally, I believe that anyone who is an expert in their domain should leverage today’s technological advancements to address problems effectively. India should develop its own pioneering models in this field.”
