OpenAI has successfully trained a human-like robot hand to solve a Rubik's Cube, marking a significant leap in robotics and Artificial Intelligence. The system utilizes two neural networks trained entirely in simulation via reinforcement learning, the same codebase used for OpenAI Five, combined with a novel technique called Automatic Domain Randomization (ADR). ADR allows the AI to generalize knowledge to the physical world without specific real-world training data.

The robot demonstrates unprecedented dexterity and adaptability, handling unforeseen physical disruptions like being prodded by a stuffed giraffe. This breakthrough proves that reinforcement learning can transition from virtual tasks to solving complex physical-world problems. It represents a major step toward general-purpose robots capable of operating in unpredictable environments.