When you type instructions into ChatGPT or Claude, you’re not having a conversation. You might feel like you are because the interface looks like a chat window, the AI responds with “That’s a great question!” and adjusts its tone based on yours but what’s actually happening bears little resemblance to human communication.Prompt engineering is constraint optimisation dressed up as dialogue. Understanding this distinction matters deeply for anyone building AI products, setting user expectations, or trying to understand what these systems actually do.What Communication Actually RequiresThe Asymmetry of InstructionWhy LLMs Feel Like ConversationLinguistic Frameworks We Already UseThe Product Development ProblemThe Monetisation of UnfamiliarityWhat This Means for AI ProductsThe Deeper QuestionThe human capacity for communication, bidirectional, negotiated, grounded in mutual agency, remains distinctive. Recognising that distinction helps us use powerful tools like LLMs more effectively, without confusing their constrained optimisation processes with the richer, messier, genuinely collaborative process of human dialogue. This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit thestrategiclinguist.substack.com/subscribe

Ask this episode

“What did the council decide on item 42?” — answered with the speaker and the moment.

Transcript

Related