Why Gaming Data Outperforms the Internet for AI Training
When it comes to achieving artificial general intelligence (AGI), large language models simply don’t have what it takes. Models like ChatGPT and Claude excel at text but struggle to understand how things move through space and time — an essential skill for producing intelligence that generalizes. That gap, according to General Intuition CEO Pim de Witte, might be filled by gaming data. This is the core bet behind General Intuition, a Bezos-backed, New York-based startup that has attracted investment from Coatue, Eric Schmidt, and researchers at MIT and Google DeepMind.
The Limits of Language Models and the Promise of Gaming Data
Why Physical Understanding Matters
De Witte explains that current AI models lack a fundamental understanding of physics and temporal dynamics. While they can parse text, they cannot reason about how objects interact in the real world. Gaming data, by contrast, provides rich, structured environments where cause and effect, motion, and spatial relationships are embedded. This makes it an ideal training ground for world models — AI systems that can simulate and understand physical reality.
General Intuition’s Bold Bet on World Models
The startup, which spun out of gaming platform Medal TV, is building world models trained on vast amounts of gameplay data. De Witte argues that this approach could bridge the gap between narrow AI and AGI. On this episode of TechCrunch’s podcast, he joined host Rebecca Bellan to dig into why world models trained on gaming data might be the next big leap in physical AI. The episode was published at 10:47 AM PDT on July 8, 2026, and runs for 26 minutes.
Backed by Top Investors
General Intuition’s investor list includes prominent figures and institutions: Bezos (via his venture funds), Coatue, former Google CEO Eric Schmidt, and researchers from MIT and Google DeepMind. This mix of tech industry leaders and academic experts underscores the potential of the gaming data approach.
Ethical Boundaries and Defense Applications
De Witte also addressed the ethical implications of AI in defense applications. As world models become more capable, questions of dual-use arise. The CEO acknowledged the need for careful governance, noting that the company is committed to responsible development. The podcast also touched on the technical details of how gaming data is collected, processed, and used to train AI systems.
For more on this topic, follow the Equity podcast, produced by Theresa Loconsolo, available on all major platforms. Also, don’t miss the TechCrunch Founder Summit in Boston on November 4 — early bird savings of up to $190 end June 26, 2026. Subscribe to TechCrunch’s newsletters for the latest in AI, startups, and space.