Better communication through prompting

There's an old idea that good writing is good thinking. Being forced to put an argument on paper — building the case, anticipating the objections, finding the holes — doesn't just communicate a thought. I think it actually clarifies one. You often don't know what you believe until you've tried to write it down.

Prompting, I think, does something similar, and in some ways the feedback loop is even more useful. When we write for human readers, they tend to fill in our gaps. They infer intent, give us the benefit of the doubt, maybe ask a clarifying question if something's unclear. LLMs don't do any of that. They take your request at face value and execute on exactly what you said, not what you meant. If your ask was vague, you find out immediately.

What I've noticed is that this trains something genuinely useful over time. We start thinking about how our requests could be misread before we send them. We get more precise about outcomes — what we want, and just as importantly, what we don't. And I don't think those habits stay confined to how we work with AI tools. They start showing up in how we write emails, give direction, and communicate with the people we work with. Maybe that's an underappreciated upside of all the time we're spending learning to prompt — we might just be getting a little better at communicating in general.