MIT and NYU Introduce FLEX System for Robots to Generalize from Conversational Corrections

Researchers from MIT and New York University have developed FLEX, a new AI-assisted learning platform that enables robots to interpret and generalize from natural language commands beyond literal instructions. This 'Flexible Feature Editing Pipeline' allows humans to provide conversational corrections, helping robots infer broader preferences (e.g., avoiding all fragile objects instead of just a specific mug), leading to more intuitive and interactive robotic systems.