Readers of this blog — and pretty much every other platform on the internet — are hearing a lot these days about the rise of artificial intelligence and what it’s teaching us about the contemporary workplace. One thread of this field I’ve found particularly intriguing is the sudden popularity of the term Prompt Engineering. In short, Prompt Engineering is the ability to craft prompts for Large Language Models (Claude, Gemini, ChatGPT, etc.) that produce more relevant, useful outputs. I’m no expert on the subject, but from what I’ve gathered providing context, uploading reference materials, offering details on what kind of answers would be most useful to you, and actually pursuing clarity through negation — letting the LLM know what you don’t want — all raise the probability its answers will be more satisfying.
While interesting, better AI prompting is not what this blog is about. What has really “lit my wick,” if you will, about this subject is how relatively uninformed and under-practiced we still are at offering each other, in human-to-human dialogue, decent prompts for our own conversations. Two of our most common mistakes:
- Closed Questions. Questions that can be answered with a simple yes or no, one-word answer. We don’t realize we’re doing this, until the ride home from school when our child hits us with a series of “Goods” and “Fines” and we feel like a cast member on Survivor trying to make fire with flint and stone. We wonder why our conversations die on the vine and assume there’s nothing we can do to avoid this fate.
- Leading Questions.Essentially, these are suggestions or opinions disguised as questions. You can often hear these coming: “Have you ever thought of…?” These narrow the aperture of conversation. Your partner will reply but it will likely be to agree or disagree with your statement, rather than offer a fresh response from whole cloth.
Genuine inquiries, as opposed to leading ones, allow us to hear responses that teach us something about what’s in the other person’s head, as opposed to just publishing what’s in our own. Open questions elicit richer and more detailed responses. Beyond that, there‘s a vast landscape of (human-to-human) prompts available to us, more diverse than we might have previously imagined. From the visually evocative, “Paint a picture for me…,” to the numerically oriented, “Tell me three things that…” to the solicitation of personal preference, “What was your favorite (or least favorite) part of...”
The fact is there is no script or formula for an enriching conversation. But if we think of bringing a design or engineering lens to the questions we ask, and accept that our questions will elicit different responses based on how they’re designed, we’ll raise the likelihood that our daily exchanges will become more stimulating, surprising and meaningful.
Start paying attention to the questions you ask, and are asked. Not to evaluate them, but to begin listening to what sorts of responses they organically produce. Inevitably you’ll start refining the questions you pose in order to get closer to the sorts of responses you’re hoping for. Your exchanges will grow more interesting, and your conversation partners will become more compelling.
That’s right: just as we can tee AI up better, we can set each other up for more conversational success.
Give it a try. Your next coffee date will thank you.