General Intuition: Gaming Data May Unlock AGI, CEO Says
Original: Why this CEO thinks video games make better training data than the internet
Why This Matters
Gaming-data-trained world models could reshape the path toward physical AI and AGI beyond text-based LLMs.
Bezos-backed startup General Intuition, valued at $2.3 billion, raised $320 million from Coatue, Eric Schmidt, and researchers at MIT and Google DeepMind. CEO Pim de Witte argues video game data trains world models better than internet text for physical AI and AGI development.
General Intuition, a New York-based startup that spun out of gaming platform Medal TV, has closed a $320 million funding round, bringing its valuation to $2.3 billion. Investors include Coatue, Eric Schmidt, and researchers affiliated with MIT and Google DeepMind, with backing also from Jeff Bezos.
CEO Pim de Witte appeared on TechCrunch's Equity podcast to explain the company's core thesis: large language models such as ChatGPT and Claude excel at processing text but lack the ability to understand how objects move through space and time — a capability he considers essential for achieving artificial general intelligence (AGI). De Witte argues that video game data, which is rich in spatial and temporal information, provides superior training material for building 'world models' — AI systems that can simulate and reason about physical environments.
The company is positioning gaming-derived data as a foundation for physical AI and robotics applications. De Witte also addressed ethical considerations, noting concerns about potential defense-sector use of the technology. General Intuition's approach represents a broader industry debate about the limits of text-based training data and the need for richer, physics-grounded datasets to advance AI capabilities beyond language tasks.