The next literacy is externalizing judgment

This article argues that the most important skill for leveraging large language models (LLMs) is not prompt engineering but learning to externalize judgment. It uses the analogy of hiring a designer to illustrate how vague terms like 'premium' or 'engaging' cause miscommunication. LLMs expose ambiguity in human thinking, forcing users to translate intuition into explicit criteria. Unlike prompts, which are task-specific, judgment is transferable across tasks. The article provides principles for articulating taste, such as challenging conventional wisdom early and introducing concrete examples before abstractions, to improve LLM outputs and personal clarity.