Meta AI and the University of Illinois Chicago researchers have created a new system that helps AI models respond more efficiently to queries. As reported by Ben Dickson in VentureBeat, the technology enables AI to determine whether a question needs deep reasoning or a quick response. The system, called Inference Budget-Constrained Policy Optimization (IBPO), uses reinforcement learning to teach models how to adjust their reasoning process based on query complexity. This approach solves a common problem where AI models spend unnecessary time on simple questions. Tests show that IBPO-trained models outperform traditional systems while using computing resources more efficiently. The researchers found that this method works better than supervised fine-tuning, as it allows models to develop self-correction capabilities automatically.