Researchers have found an equivalence between two reinforcement learning algorithms: policy gradients and soft Q-learning. This means that the two methods, which are often used separately in AI systems, can be combined or replaced with each other without significant changes to the overall system. The discovery was made by OpenAI researchers, who published a paper detailing their findings. --- Why it matters: This equivalence matters because it allows researchers and engineers to choose between policy gradients and soft Q-learning based on specific needs of their project, rather than being locked into one approach. This flexibility can lead to more efficient and effective reinforcement learning systems. Source: https://openai.com/index/equivalence-between-policy-gradients-and-soft-q-learning