Prompt Engineering Is a Misnomer — Managing Ambiguity in AI

For years we've called it "prompt engineering" — but what if that framing is fundamentally wrong? This episode explores a provocative idea: that the real skill in working with language models isn't precision, but the art of managing ambiguity. Drawing on hundreds of episodes of real-world prompting, we examine why over-specifying your prompts actually caps output quality, how voice dictation preserves crucial signals that typed text strips away, and why the most impressive AI results come when you deliberately leave doors open. We break down the spectrum between tight specificity and productive vagueness, and show how moving fluidly between them — rather than picking one mode — is where the real power lives. Whether you're debugging a network or brainstorming creative work, the calibration between constraint and creativity is the skill that transfers.