Why complex reasoning models could make misbehaving AI easier to catch
In a new paper from OpenAI, the company proposes a framework for analyzing AI systems' chain-of-thought reasoning to understand how, when, and why they misbehave.
In a new paper from OpenAI, the company proposes a framework for analyzing AI systems' chain-of-thought reasoning to understand how, when, and why they misbehave.
