Video Games Training Real-World AI Robots: General Intuition's $2.3 Billion Vision

General Intuition has raised 320 million dollars in a funding round that values the company at 2.3 billion dollars.

The startup uses action data from millions of video game clips to train AI agents that transfer skills from virtual worlds to physical robots.

Game Data Unlocks Better AI Intuition

This method builds on years of gameplay footage collected through the founders' earlier company Medal which lets gamers share clips.

Unlike other approaches that guess actions from video only General Intuition uses exact button presses to teach models about cause and effect in space and time.

The result is a world model that understands walls ladders and moving shadows allowing agents to explore safely without constant real world trials.

Such scalable training could cut the high costs of gathering physical data and speed up development of helpful robots for homes and factories.

Ethics Shape a Responsible AI Future

CEO Pim de Witte brings a European perspective and humanitarian background that sets strict rules against using the technology to harm people.

The company also launched the Nerve platform to let gamers earn money through data tasks and robot control creating new opportunities in an AI driven job market.

Investors like Khosla Ventures see this as a long term bet on intuitive AI that could power everyday tools from search and rescue drones to factory helpers.

For ordinary people this means safer smarter machines that understand the world like humans do rather than just processing text.

Overall the work points to a future where AI feels more natural and accessible while prioritizing human values over rapid unchecked growth.