🔍 Ask This Before You Ask AI

Good morning. I used to fire off research questions to AI with zero setup and wonder why the answers came back mushy. These days I scope the question first, sometimes asking AI itself to sharpen it before I even run it. Today's lead is about that one habit.

I used to fire off research questions to AI with almost no setup: a vague topic, hit enter, and wonder why the answer felt thin or generic. It took me longer than I'd like to admit that the problem wasn't the AI. It was the question.

Deep Research tools built into ChatGPT, Gemini, and Claude can genuinely dig: scanning dozens of sources, cross-checking claims, building a real report instead of a guess. But they answer the question you actually asked, not the one you meant. Ask something too broad and you get a mile-wide, inch-deep summary. Ask something too narrow and you miss the context that would have changed your answer.

The fix I use now: before running a research prompt, I tell AI what I already know, what I'm actually trying to decide, and what would make an answer useless to me. Sometimes I go a step further and ask AI itself to sharpen the question first, a trick called meta-prompting that's exactly what it sounds like.

The catch: a better-scoped question still needs a skeptical reader. AI can hand you a beautifully organized answer to the wrong question, and you won't know the difference unless you already understand roughly what you're looking for.

This is where years of judgment do more work than any prompt trick. You already know what a real answer should sound like in your field. Bring that skepticism to the question, not just the answer.

Go further: What Deep Research tools actually do ◆ How to have AI sharpen your own prompts

FROM OUR PARTNER

AI is moving fast and most people are falling behind. 

The Rundown AI keeps you ahead of the curve. 

It's a free AI newsletter that keeps you up-to-date on the latest AI news, and teaches you how to apply it in just 5 minutes a day.

Plus, complete the quiz after signing up and they’ll recommend the best AI tools, guides, and courses — tailored to your needs.

Sign up to start learning.

AI TUTORIAL

One AI can sound completely certain and still be wrong. I've learned that the hard way on business and investment decisions, which is exactly why I stopped trusting a single answer for anything that actually matters.

✓ Pick a decision that's big enough to be worth the extra time: a major purchase, a career move, an investment, a business move.

✓ Write one clear prompt describing the decision and the real factors you're weighing.

✓ Run that exact same prompt, word for word, through three different AI models.

✓ Compare the three answers specifically for where they disagree, not just where they agree.

✓ Ask each tool why it landed where it did, and what it thinks the others are missing. That follow-up question is where the real value shows up.

✓ Sometimes I take the response from two of the AIs and paste them into the third and ask it to review and compare to its own answer. That can be pretty interesting.

If it does not work: If all three tools agree completely, that's not proof you're right. It usually means the question had an obvious answer, or you gave leading information without meaning to. Try removing one detail that might be steering the answer and run it again.

What this doesn't fix: This is a second-opinion generator, not a fortune teller. None of the three tools knows anything about your specific situation beyond what you type, and none of them is accountable if the decision goes wrong. A decision big enough for this exercise is often big enough for a human advisor too: a lawyer, a financial advisor, an accountant.

ASK A FAMOUS MIND
🍳 What would Julia Child tell someone who hates cooking but wants to eat better?
Ask her yourself in Ask a Famous Mind. Always free. 👉 Try it here

NEWS YOU CAN USE
This Matters at Home
Short, practical summaries to help you stay updated on recent AI news—and what it means for everyday life.

📚 ChatGPT's New Teen Mode Wants to Coach, Not Cheat. OpenAI's new teen version has a Study Mode that nudges kids toward answers instead of handing them over. Parents still need to watch for workarounds.

🩺 Chatbots Are Getting Better At Spotting Suicide Risk, Not Perfect. A nonprofit tested 50,000 simulated conversations and found today's models are far less likely to encourage suicidal thinking than older versions, though they can still miss the signs behind a task-based request.

🍳 This App Turns Your Saved TikTok Recipes Into an Actual Grocery Order. Remy can now pull a recipe straight from a social video and load the ingredients into your grocery cart. One less excuse to order takeout instead.

REAL WORLD
How We Used AI This Week
Real people, real uses—here’s how AI is quietly making life better, one small win at a time.

🔍 Jiwoo K., Seattle, WA. Her parents had a nearly 100 page lease for their store and one specific answer buried inside it. Instead of reading line by line, she uploaded the whole document to ChatGPT and asked it to find the relevant section, then had it translate that answer into Korean so her parents could understand it. The whole thing took less than ten minutes. A long search through dense paperwork became one quick question, in the language that mattered.

👉 Have a story of your own? Tell us how you used AI this week.

READER POLL

If today's tutorial has you thinking about running a big decision through more than one source, Boldin is worth adding to a retirement or investment call specifically. It models different scenarios (retiring earlier, spending more, a market downturn) so you can see the actual numbers before you decide, instead of guessing. Try it here.

PASS IT ON
Know someone who would like this newsletter? Send them here.

Missed an issue? Read the archive.