General Intuition, a startup focused on embodied AI, has developed a foundation model designed to replace the need for extensive real-world data in robotics. The company argues that general-purpose models can transfer spatial and temporal reasoning across multiple environments, reducing the reliance on large datasets. According to Pim de Witte, CEO of General Intuition, the industry is currently focused on specialized work for individual robots and environments, but this will soon become redundant with the emergence of general models. "The generalization of the model itself is the product," de Witte said. "The fact that it has a base level of reasoning about space and time is going to be the reason why people stop collecting hundreds of thousands or millions of hours of real-world data." General Intuition trained its model on millions of hours of video game data, including controller inputs and timing, to develop human-like intuition for spatial-temporal reasoning. The startup demonstrated that its current model can both play video games for hours and power a quadrupedal robot after fine-tuning it on just eight minutes of real-world data. "The fact that [the robot] was actually able to zero-shot on just the front camera, with no other sensors, in the office with dynamic objects being introduced and people walking by was a very big surprise to us," de Witte said. "I think it’s a sign of what’s to come." The end game for General Intuition isn’t to build robots itself, but to become the foundation model of physical AI, a base model for other robotics companies to build upon for their own machines. Or, as de Witte put it: "We’re not gonna build a self-driving car company. We’re gonna make it 10 times easier for the next person to build a self-driving car company."
Source: techcrunch
