The ChatGPT Revolution in Robotics: General Intuition Lays the Foundation for a New Era

While the field of AI has made a giant leap in processing text and images, robotics has long remained confined to narrowly specialized models. However, the startup General Intuition has announced that the era of universal foundation models, similar to ChatGPT, is beginning in this field as well. Instead of training robots for each individual movement, the company aims to give them a general "mind" that understands spatial and temporal logic. This is reported by Techcrunch.com reports .

Previously, companies had to build models from scratch for every specific task and train them with millions of hours of real-world data. In an interview with TechCrunch, General Intuition CEO Pim de Witte noted that this approach is now becoming obsolete. Just as OpenAI's GPT-3 model revolutionized natural language processing, the new generation of models can give robots the instinct to navigate in diverse environments.

The Eight-Minute Learning Miracle

General Intuition used millions of hours of video game data to create its foundation model. This data includes not only the visual input but also when and how a human pressed controller buttons. As a result, the system has mastered the connection between action and consequence, a logic similar to human intuition.

One of the biggest achievements of the company is that their model learned to control a quadruped robot based on just eight minutes of real-world data. Without any additional sensors, using only a front-facing camera, the robot was able to successfully navigate through dynamic obstacles and moving people in an office. This result came as a surprise to industry experts.

Future Goals and Investment

General Intuition sees itself not as a robot manufacturer, but as a platform creating a technological foundation for other companies. According to Pim de Witte, they are not building self-driving cars, but providing software that makes it 10 times easier for the next entrepreneur to create such a vehicle.

This promising idea was also highly valued by investors. The startup recently raised $320 million in funding, reaching a total market valuation of $2.3 billion. The project's lead investor, Vinod Khosla, also believes that motion data is the key to AI performing perfectly in the physical world.

In conclusion, General Intuition's approach is expected to completely transform the robotics industry. If it previously took months to teach a robot to walk in a single room, this process can now be accomplished in a matter of minutes. This will significantly accelerate the integration of robots into our daily lives, manufacturing, and service sectors.