In his “Own the Outer Loop” keynote, Addy Osmani frames cognitive surrender as blindly accepting what an AI agent gives you rather than applying independent judgment to it. Because delegated work is only nominally the agent’s — it carries the delegator’s name, reputation, and responsibility — the risk is not that the agent’s output is sometimes wrong, but that a person stops checking it well enough to catch that it is wrong.
Usage
The source cites a Wharton study on this effect: it found that when an AI’s advice was wrong, nearly three-quarters of people accepted it anyway, and felt more confident in the decision than they would have without the AI’s involvement — a comforting result when the AI is right and a concerning one when it is not. It is presented as one of three hidden costs of agentic delegation, alongside cognitive debt and the orchestration tax.