Archive — US — techcrunch
General Intuition raises $2.3B on bet that video games can train AI agents forthe real world
General Intuition has raised $2.3 billion to leverage video game data for training AI agents in real-world applications. Their unique approach combines gameplay action data to improve spatial reasoning and robot navigation. Backed by high-profile investors, they aim to scale their AI model while maintaining ethical boundaries, avoiding military uses of their technology.
Key points
- General Intuition raised $320 million at a $2.3 billion valuation, totaling $454 million in funding.
- The company uses video game data, especially action labels, to train AI models capable of understanding real-world dynamics.
- The model can generalize from gaming to real-world scenarios, requiring minimal real-world data for fine-tuning.
- Investors include Khosla Ventures, Google DeepMind, and Jeff Bezos, indicating strong confidence in the technology.
- General Intuition plans to scale compute capacity and enhance model training, while also developing its API for broader accessibility.
- The CEO, Pim de Witte, emphasizes ethical use of AI, refusing military applications and focusing on positive use cases like search and rescue.
Why it matters
- The rise of AI in gaming and robotics is part of a broader 2025 IT trend toward utilizing immersive environments for training agents.
- The emphasis on ethical implications in technology echoes global discussions around AI governance and responsible innovation.
- Investments in AI startups reflect the growing recognition of data-driven insights from non-traditional datasets like gameplay.
General Intuition is positioning itself at the forefront of AI development by integrating unique gaming data, emphasizing ethical boundaries, and hoping to revolutionize robot training, which could shape future technological advancements.