A US-based startup, General Intuition, is making waves in the robotics sector, proposing a revolutionary shift in how intelligent machines are developed. The company believes that the industry is on the cusp of a 'ChatGPT moment', where general-purpose 'foundation models' for physical AI, akin to those used in natural language processing, will become the standard. This approach could significantly streamline the creation of more capable robots by reducing the need for vast, task-specific real-world data.
General Intuition's CEO, Pim de Witte, argues that the current method of building specialised robot models from scratch, each requiring extensive real-world datasets, is inefficient. Instead, he advocates for focusing on higher-quality datasets that can produce foundation models capable of transferring fundamental intuition about movement and interaction across diverse environments. The company has developed its own such model, trained on millions of hours of video game data, including human controller inputs, which they believe is key to developing human-like spatial-temporal reasoning.
The startup recently secured a significant investment, raising $320 million last month at a valuation of $2.3 billion, underscoring investor confidence in their thesis. General Intuition has already demonstrated the capabilities of its current model, showcasing its ability to play video games for extended periods and, remarkably, power a quadrupedal robot. The latter was achieved after fine-tuning the model with a mere eight minutes of real-world robotics data, a stark contrast to the hundreds of thousands or millions of hours typically required.
This development has profound implications for the robotics industry, potentially making the development of advanced robots more accessible and faster. De Witte envisions General Intuition becoming the foundational layer for physical AI, providing a base model upon which other robotics companies can build their specific applications. This would mean they would not build robots themselves, but rather enable others to do so with unprecedented ease and efficiency, potentially accelerating innovation across sectors from logistics and manufacturing to autonomous vehicles.
The emergence of such general-purpose models also brings into focus regulatory considerations. In the UK, the Information Commissioner's Office (ICO) is actively monitoring AI developments, while the EU's AI Act, set to come into full effect in the coming years, will establish comprehensive rules for AI systems. These frameworks will likely play a crucial role in shaping the responsible deployment and ethical considerations surrounding increasingly autonomous and intelligent physical AI systems, ensuring safety and accountability as the technology advances.