General Intuition Raises $320M to Train AI Agents Using Video Game Data

General Intuition, a New York-based startup, has secured $320 million in funding, elevating its valuation to $2.3 billion. This investment underscores the company’s innovative approach to training artificial intelligence (AI) agents using video game data to develop spatial-temporal reasoning capabilities.

Founded by Pim de Witte, General Intuition emerged from Medal, a platform where gamers upload and share video clips. The vast repository of gameplay footage, complete with action labels detailing player inputs, serves as a rich dataset for training AI models. Unlike competitors who rely solely on video data, General Intuition leverages these action labels to enhance the learning process, enabling AI agents to understand and navigate complex environments more effectively.

During a recent demonstration at their New York office, the company’s AI agent showcased its proficiency by playing Fortnite continuously for 100 hours. Remarkably, the same AI model was also deployed to control a quadrupedal robot, highlighting the versatility and adaptability of their technology. The robot, equipped with a single camera, navigated the office space autonomously, requiring only eight minutes of real-world data for fine-tuning. This rapid adaptation emphasizes the efficiency of General Intuition’s training methodology.

The latest funding round attracted prominent investors, including Jeff Bezos and Eric Schmidt, alongside existing backers such as Khosla Ventures and General Catalyst. This financial support is expected to accelerate the development and deployment of General Intuition’s AI agents across various real-world applications.

By harnessing the intricate data from video game interactions, General Intuition is pioneering a novel path in AI training. This approach not only enhances the agents’ understanding of spatial and temporal dynamics but also facilitates their transition from virtual environments to tangible, real-world scenarios. The success of this strategy could revolutionize sectors like robotics, autonomous vehicles, and beyond, where precise navigation and interaction are paramount.

As the AI landscape evolves, General Intuition’s innovative use of video game data positions it at the forefront of developing adaptable and efficient AI agents. The substantial investment reflects confidence in their methodology and its potential to bridge the gap between virtual training environments and real-world applications.