Gran Turismo 7's Sophy AI Project: A Five-Year Journey of Innovation and Sportsmanship
The world of racing video games has been forever changed by the introduction of Sophy, the AI driving system in Gran Turismo 7. Since its initial reveal in 2022, Sophy has evolved from a research experiment to a permanent fixture of the game, captivating players and pushing the boundaries of AI technology. In this article, I'll take a deep dive into the journey of Sophy, exploring its development, impact, and the exciting future it holds.
The Birth of Sophy: A Research Experiment
Sophy's story begins in Sony's R&D division around 2016, where researcher Kenta Kawamoto was exploring the potential of Gran Turismo as a testbed for reinforcement learning. The early work was primarily focused on building a way for an external AI agent to interact with the game, running small experiments on a handful of consoles. However, it was the formation of Sony AI in 2020 that truly kickstarted the project.
The introduction of a custom training system called 'DART' in early 2021 allowed researchers to connect to over 1,000 cloud-hosted PlayStation 4 consoles simultaneously. This scale enabled the team to generate vast amounts of driving experience and explore multiple training approaches in parallel. The result was a significant leap forward in AI development, and Sophy began to take shape.
The Struggle for Sportsmanship: A Key Challenge
One of the most fascinating aspects of Sophy's development is the story of its struggle to become a fair and sporting opponent. In the lead-up to the first 'Race Together' exhibition event in July 2021, Polyphony Digital's in-house testers discovered a fatal flaw in the competition build. Sophy had become ruthlessly effective at contact and obstruction, deliberately crashing into opponents and blocking aggressively. This behavior was not only unfair but also disqualified the agents if they raced that way in the actual event.
Sony AI spent the final week retuning the incentives that shaped Sophy's behavior, training a replacement set of agents, and testing them just in time. The result was a significant improvement in Sophy's sportsmanship, and it went on to take the top two positions in every race of the October 2021 rematch. This achievement not only showcased Sophy's speed but also its ability to be a good sport, with Peter Wurman describing it as 'the first real AI that we could call a good sport'.
From Showcase to Staple: The Evolution of Sophy
What happened next was unusually fast for AI research. Within a year of the Nature cover, Sophy arrived in Gran Turismo 7 as a limited-time feature on just four circuits. This launch was symbolic, marking a significant moment in the 25-year history of Gran Turismo. Sophy 2.0 followed in November 2023, making the AI a permanent feature covering 340 cars. This scaling required the team to solve new problems, such as identifying which vehicle properties the network needed to know about.
Researcher Patrick MacAlpine's analysis found that drivetrain configuration was one of the most important predictors of a car's behavior, while dry weight alone was a surprisingly weak indicator for how a car actually handles. Sophy 2.1 brought Custom Race support in March 2025, allowing players to use customized grids and tuned cars against the AI. Sophy 3.0 arrived last December as the headline feature of the Power Pack DLC, expanding its capabilities to handle tire wear and fuel consumption.
The Community's Patience: A Test of Time
Despite the impressive progress, the community's patience has been tested. Sophy still doesn't power the AI in GT7's single-player campaign, where the game's traditional opponents remain. Track coverage has grown slowly, often two circuits at a time across more than a year of game updates. Sophy 3.0's most advanced racing behavior sits behind the Power Pack's paywall, a tough sell for players who feel better AI should be a baseline feature rather than premium DLC. B-Spec mode, a long-requested feature, remains absent despite years of player requests.
To the team's credit, they don't pretend everything is solved. Kaushik Subramanian acknowledged that Sophy is 'not perfect', and the team explained that extreme tuning and engine swaps remain difficult edge cases. However, the team is committed to improving Sophy and bringing new features to players.
The Future of Sophy: Beyond Gran Turismo
The retrospective also reveals where Sophy's technology is heading, and some of it even points beyond Gran Turismo. The clearest example is Project Ace, Sony AI's table tennis robotic system, which was accepted for publication in Nature this year. The Ace team borrowed a technique first explored in Sophy research, where the AI's 'critic' is given privileged information the finished agent will never have, dramatically accelerating learning. This technique has also been applied to vision-based racing research, where an agent is trained to race using only an in-car camera view and virtual onboard IMU data.
The pieces are falling into place for AI opponents that are cheaper to build, easier to customize, and potentially easier to transfer from simulation to physical systems. As for what players can expect, researcher Alisa Devlic offered the clearest hint, saying the team hopes to bring Sophy to many more tracks and offer 'different performance levels that are personalized to and can challenge each individual player'. Wurman added that the team is always looking for ways to help Polyphony bring new features to players.
Conclusion: A Journey of Innovation and Sportsmanship
Five years in, Sophy has gone from a research demo I raced on a LAN in Monaco to a fixture of Gran Turismo 7. The research pipeline behind it shows no sign of slowing down, with exciting developments on the horizon. As Sophy continues to evolve, it will be fascinating to see how it shapes the future of racing video games and AI technology. In my opinion, Sophy's journey is a testament to the power of innovation and sportsmanship, and it's a privilege to witness its progress.
