Natural Language Reinforcement Learning (NLRL) is a new reinforcement learning framework proposed in 2024 by researchers from University College London, Shanghai Jiao Tong University, Brown University, University of Bristol, National University of Singapore, and University of Surrey.Natural Language Reinforcement Learning".

The core idea of NLRL is to analogize the core concepts of reinforcement learning (RL), including task objectives, strategies, value functions, Bellman equations, and policy iterations, to natural language-based forms, thus opening up a new path to smarter and more natural AI decision-making learning. This approach innovatively redefines RL principles as language counterparts, enabling AI systems to learn through interaction with the environment without relying on any human-labeled data.