从人类偏好中学习：安全AI新方法
Why we picked it
该方法解决了AI对齐中的核心难题——如何让AI理解人类真实意图，对构建可控AI系统具有里程碑意义。
OpenAI与DeepMind合作开发了一种算法，通过比较两种行为的好坏来推断人类期望的目标。该方法避免了手动编写复杂目标函数可能导致的危险行为，为构建更安全的AI系统提供了关键思路。
Industry Coordinates & Source Verification
This report is categorized under AITOP's "Papers" desk. Key entities and technical topics under continuous tracking include ai-safety, reinforcement-learning, human-preferences, alignment.
Primary source verification: published directly by OpenAI Blog with first-party authority.
