General Intuition, which trains AI via video games, valued at $2.3 billion

New York-based startup General Intuition has raised $320 million in a funding round to create a universal "brain" for robots and digital agents based on video game data. Following this round, the company's valuation reached $2.3 billion. This technology offers a revolutionary approach to training AI to navigate not only virtual worlds but also the real world, reports Techcrunch.com reports .

Founder and CEO Pim de Witte notes that their agent can act independently for hundreds of hours in complex games like Fortnite. Interestingly, the same algorithm is being applied to control quadrupedal robots. This means the AI can successfully transfer experience gained in virtual environments to the physical world.

From video games to real robotics

General Intuition's approach differs fundamentally from competitors. While most companies try to train AI by analyzing only videos, de Witte's team also accounts for the buttons users pressed during gameplay. This method helps the model better understand spatial-temporal reasoning.

According to TechCrunch, the company sourced its initial database from the Medal platform, which allows gamers to upload their gameplay. Millions of hours of game clips and their associated control commands served as the perfect training ground for the General Intuition model. As a result, the robot learns to understand real-world objects very quickly.

According to company experts, just eight minutes of real-world data were enough to adapt the robot to a new environment. For example, a robot tuned with street data was able to move freely among chairs and obstacles in an office. This is considered a very high result for modern robotics.

Future opportunities

Unlike text-based models (LLM) like ChatGPT, the model being created by General Intuition specializes in rapid response to visual information and executing physical actions. This technology is expected to be used in the following areas in the future:

• Managing complex logistics and warehouse robots;
• Assistant robots performing daily household chores;
• Virtual assistants acting like humans within games and applications;
• Safer and smarter decision-making mechanisms for autonomous systems.

The company's total investment has now reached $454 million. Such large investments prove that in the tech world, video games are not just entertainment, but the most important data source for creating future Artificial General Intelligence (AGI).