If you want to achieve artificial general intelligence (AGI), a large language model is simply not what you need. Models like ChatGPT and Claude are great for text, but less adept at understanding how things actually move in space and time. This is an essential skill for generating intelligence that generalizes. It turns out that this gap can be filled by game data. That’s the bet behind General Intuition, a $2.3 billion New York-based startup backed by Bezos. The company just closed a $320 million round, adding Courtuet, Eric Schmidt, and researchers from MIT and Google DeepMind to its list of investors.
In this episode of TechCrunch’s Equity podcast, General Intuition CEO Pim de Witte joins Rebecca Bellan to dig into why world models trained on game data are the next big leap in physical AI, how the company spun out of gaming platform Medal TV, and where the ethical red lines are when the models might end up in defense applications.
Listen to the full episode to hear more about what’s next.
How 8 minutes of real-world data was enough to help a robot navigate the office cold. Why General Intuition reportedly turned down an acquisition offer from OpenAI to maintain its independence, and why building a generational company requires investors who support its mission. How the company is trying to get ahead of AI job losses by building Nerve, a marketplace that connects gamers to data labeling and remote control tasks.
Subscribe to Equity on YouTube, Apple Podcasts, Overcast, Spotify, and all casts. You can also follow Equity on X and Threads at @EquityPod.
