K - How to Ask AI the Right Questions
How to Ask AI
the Right Questions
The quality of what you get back starts with the quality of what you type in. Here's how to ask well.
Why it matters
Vague in, vague out
AI doesn't read your mind — it continues whatever pattern you start. A vague prompt gives it a vague pattern to continue, and a vague answer is exactly what comes back.
This is the single most useful thing to internalize about working with AI well. The model isn't holding back a brilliant, specific answer and waiting for you to unlock it with a magic phrase — it's genuinely working with whatever information you gave it. Ask a broad question, and it has no choice but to guess at what you actually meant, filling the gaps with generic, safe assumptions. Ask a specific, well-framed question, and it has real material to work with, producing something far closer to what you actually needed.
It helps to picture the AI as a very capable, very fast new colleague who has never met you, doesn't know your project, and can't see what's in your head. Handed a one-line request like "write something about marketing," a talented new hire would still have to guess at the audience, the tone, the length, and the goal — and would likely hand back something safely generic to avoid getting it badly wrong. Give that same colleague the actual brief — who it's for, what it needs to say, how long it should be — and the work that comes back improves dramatically, without their underlying skill changing at all. AI responds to that same shift in exactly the same way.
Side by side
The same request, two different ways
"Write something about marketing."
"Write a 150-word Instagram caption announcing a 20%-off weekend sale for a small bakery, upbeat tone, with two relevant hashtags."
"Help me with my resume."
"Review my resume for a mid-level marketing role. Focus on making my bullet points more results-oriented, and flag anything that sounds generic."
"Explain machine learning."
"Explain machine learning to a curious 12-year-old, using a sports or gaming analogy, in under 150 words."
Notice what changed in each pair: not the topic, but the amount of context, audience, and desired shape provided up front. That's the real skill behind "prompting" — it's less a magic trick and more just good, clear communication, the same kind you'd use briefing a new colleague on a task.
It's also worth noticing what stayed the same across each pair: the underlying subject. You don't need to learn special vocabulary or arcane syntax to get better results — you just need to say the ordinary things you'd naturally tell a person if they were about to do the task for you. If you wouldn't hand a task to a colleague with only three words of instruction, that's usually a sign the AI needs more too.
The four ingredients
What a strong prompt usually includes
You don't need all four in every prompt — a quick factual question needs none of them — but for anything you actually care about getting right, running through this checklist takes a few extra seconds and noticeably raises the quality of what comes back.
Task is the verb of your request — write, summarize, explain, compare, debug — stated plainly rather than implied. Context is anything the AI would need to know that isn't obvious from the task alone, like who the audience is or what's already been tried. Format covers the shape of the answer: how long, what tone, bullet points or paragraphs, formal or casual. Constraints are the boundaries — words to avoid, information to definitely include, a hard word limit. Missing any one of these usually isn't fatal, but including the ones that matter for your specific task is what separates a passable answer from a genuinely useful one.
Beyond the first message
Treat it as a conversation, not a vending machine
-
Start reasonably specific
Include the task, key context, and desired format up front, rather than a bare one-line request.
-
Read the first draft critically
Notice specifically what's missing, wrong, or off-tone — vague dissatisfaction is harder to act on than a precise gap.
-
Give targeted feedback
"Make the second paragraph shorter" or "use simpler vocabulary" works far better than just "try again."
-
Break big asks into steps
A complex task — like planning an entire event — often goes better as a sequence of smaller requests than one giant prompt.
-
Provide an example when you can
Pasting a sample of the tone or format you want is often faster and clearer than describing it in words.
Worth remembering
Questions worth double-checking
Facts and figures
Specific statistics, dates, quotes, or citations should be checked against a reliable source before you rely on them.
High-stakes decisions
Treat medical, legal, or financial answers as a helpful starting point for research, not a final professional opinion.
Good prompting gets you a far better first draft — it doesn't remove the value of applying your own judgment to what comes back, especially on anything where being wrong actually costs something.
One more habit
Let the AI ask you questions too
Prompting doesn't have to be a one-way street. If a task is genuinely complex or you're not sure what details matter, it's often faster to say so directly — "ask me any clarifying questions before you start" — rather than trying to anticipate every relevant detail yourself. Many AI tools will happily respond with a short list of questions, and answering two or three of them can save you several rounds of back-and-forth revision later. Treating the interaction as a two-way conversation, rather than a single perfect command you have to get right on the first try, tends to produce better results with less total effort.
The takeaway
AI answers the question you actually asked, not the one you meant to ask — so the clearest path to a genuinely useful response is giving it the task, the context, the format, and any constraints up front, then refining from there. Better prompts aren't a trick; they're just clear communication, the same skill that makes any request to a person easier to fulfill well.
Comments
Post a Comment