Imagine controlling a robot not with complex code, but with simple paragraphs. That's the surprising premise of new research exploring how Large Language Models (LLMs) can revolutionize robotics. Researchers have created a system where multiple LLMs communicate using plain English, enabling robots to perform a variety of tasks with minimal human intervention. This approach makes robots easier to understand and control, even for non-experts. The system consists of interconnected LLMs focused on vision, audio, and action planning. These LLMs exchange information through a “natural language data bus,” allowing humans to observe the robot's “thinking” process in real time. Amazingly, this system works effectively even with a slow processing speed, similar to that of the human brain. One fascinating observation was how LLMs embodied their assigned roles. For instance, an LLM prompted to act like a dog exhibited dog-like behaviors, even to the point of ignoring collision avoidance data because it wanted to “smell” the obstacle. This underscores the importance of understanding how LLMs interpret and react to their environment. Another intriguing aspect is the use of blockchain technology to establish and enforce rules for robot behavior. These “guardrails,” written in natural language and stored on a blockchain, ensure the robot operates within defined boundaries. This research highlights the potential of LLMs to make robots more accessible, adaptable, and trustworthy. Imagine a future where anyone can customize a robot’s behavior with simple instructions, opening up a world of possibilities for personal and professional use. While challenges remain in coordinating multiple LLMs and ensuring robust safety measures, this research points towards a future where robots are no longer complex machines, but collaborative partners readily understood and controlled by humans.