The Art of the Better Question: How to Get AI Advice Worth Taking

The Question That Changed Everything

A friend of mine spent three frustrating weeks going back and forth with AI tools about whether to accept a job offer. The responses she got were generic, hedging, and ultimately useless — the kind of advice that sounds reasonable but helps no one.

Then she changed her question.

Instead of asking "Should I take this job?" she asked: "Given that I value creative autonomy over salary, have two children under five, and burned out badly at my last startup role, what specific factors should I weigh most heavily in this decision? And what am I likely to be overlooking?"

The difference in response was night and day. Suddenly she was getting counsel that actually engaged with her situation — surfacing considerations she hadn't thought through, challenging assumptions she didn't know she was making, and helping her see the decision from angles she'd missed entirely.

Same AI. Same underlying capability. Radically different results.

The variable wasn't the technology. It was the question.

Why Most Questions Fail

When we bring decisions to any advisor — human or AI — we tend to ask questions the way we think about problems: vaguely, incompletely, and with hidden assumptions baked in.

Consider these common question patterns and why they consistently produce mediocre advice:

The Binary Trap: "Should I do X or Y?" This frames complex decisions as simple either/or choices, when reality usually offers a spectrum of options, hybrid approaches, or entirely different paths you haven't considered.

The Context Vacuum: "What's the best marketing strategy?" Best for whom? With what resources? In what market? Over what timeframe? Without context, any advisor can only offer generic principles that may or may not apply to your situation.

The Validation Seek: "Don't you think I should leave this relationship?" The phrasing telegraphs the answer you want, which biases the response toward agreement rather than genuine counsel.

The Premature Solution: "How do I convince my boss to give me a raise?" This assumes the raise is the right goal, that convincing is the right approach, and that your boss is the right target. A better question might explore what you're actually trying to achieve and whether a raise is the best path there.

These patterns aren't character flaws. They're the natural result of how human minds work under uncertainty. We simplify to cope. We seek confirmation to feel confident. We jump to solutions because sitting with problems is uncomfortable.

But awareness of these patterns is the first step toward asking questions that actually unlock useful guidance.

The Anatomy of a Better Question

Good questions share several characteristics. They're specific without being narrow. They provide context without drowning in detail. They invite challenge rather than seeking validation. And they leave room for the advisor to surface what you might be missing.

Here's a framework for constructing questions that get you advice worth taking:

1. State Your Situation Concretely

Don't just describe the decision — describe the landscape around it. What's your role? What constraints are you operating under? What resources do you have? What's the timeline?

Weak: "I'm thinking about starting a business."

Stronger: "I'm a 34-year-old software engineer with $80,000 in savings, a working spouse, and a product idea in the developer tools space. I'm considering leaving my stable job to pursue this full-time."

The second version gives any advisor — human or AI — enough terrain to offer genuinely relevant counsel.

2. Name Your Values and Priorities

Decisions always involve tradeoffs. The right choice depends entirely on what you're optimizing for. Make this explicit.

Weak: "Which city should I move to?"

Stronger: "I'm choosing between Austin and Denver. I prioritize outdoor access for mental health, a strong professional network in tech, and affordability since I'm still paying off student loans. I'm less concerned about nightlife or cultural institutions."

Now the advice can actually help you weigh what matters to you, rather than projecting generic preferences onto your situation.

3. Share Relevant History

Past patterns predict future outcomes. What have you tried before? What worked? What didn't? What do you know about your own tendencies?

Weak: "How do I get better at managing my time?"

Stronger: "I've tried time-blocking, Pomodoro, and GTD over the past three years. Time-blocking worked well until I got a new role with constant interruptions. I tend to overcommit and then feel paralyzed by my to-do list. Mornings are my peak focus time but I keep scheduling meetings then."

This kind of context allows for advice that accounts for what's already failed and what your actual patterns look like.

4. Invite Challenge and Blind Spots

Explicitly ask for perspectives you might be missing. This counteracts the natural human tendency to seek confirmation.

Add phrases like:

• "What am I likely overlooking?"
• "What would someone who disagrees with my current leaning say?"
• "What assumptions am I making that might be wrong?"
• "What questions should I be asking that I'm not?"

These invitations give permission for the kind of honest counsel that's actually valuable, rather than the comfortable agreement that feels good but helps no one.

5. Specify the Type of Help You Need

Are you looking for options you haven't considered? A framework for evaluation? Help anticipating consequences? Emotional processing? Different needs require different responses.

Weak: "Help me with this career decision."

Stronger: "I've already identified three options and done initial research. What I need help with is thinking through second-order consequences — what each path might lead to in 5-10 years that I'm not currently seeing."

Assembling Your Council of Perspectives

One of the most powerful applications of better questions is using them to gather diverse viewpoints on the same decision. Rather than asking a single question and accepting the first answer, you can systematically explore a decision from multiple angles.

Consider asking your question through different lenses:

The Pragmatist: "Setting aside ideals, what's the most realistic path forward given my actual constraints?"

The Long-term Thinker: "If I optimize for where I want to be in 20 years rather than next year, how does this change the calculus?"

The Devil's Advocate: "Make the strongest possible case against my current preferred option."

The Values Guardian: "Which choice best aligns with the person I'm trying to become, regardless of external outcomes?"

The Risk Analyst: "What's the worst realistic scenario for each option, and how recoverable would I be?"

This approach — assembling a council of perspectives around a single decision — transforms AI from a single advisor into a thinking partner that can help you see your situation from angles you'd never explore alone. Tools like thonk are built around exactly this principle: that better decisions come from structured exposure to diverse viewpoints, not from seeking a single "right" answer.

The Questions Behind the Question

Sometimes the most powerful move is to question your question itself. Before diving into "Should I take this job?" pause to ask:

• Why am I framing this as a yes/no decision?
• What am I actually trying to achieve or avoid?
• What would need to be true for this to be obviously the right choice?
• Am I asking about the decision I need to make, or am I several steps ahead of where I should be?

Often, the presenting question isn't the real question. Someone asking "How do I tell my business partner I want to dissolve our company?" might really be grappling with "How do I have difficult conversations without destroying relationships?" or even "Is there a way to save this partnership I haven't considered?"

Good advisors — and good questions — make room for these deeper explorations.

A Practice for Better Questions

Here's a simple exercise to upgrade your question-asking immediately:

1. Write down the question you're initially inclined to ask.

2. Read it back and identify what's missing: Context? Values? History? Invitation for challenge?

3. Expand it using the framework above. Aim for 3-5 sentences minimum.

4. Add an explicit request for blind spots: "What am I likely missing or assuming incorrectly?"

5. Consider whether you need multiple perspectives. If so, craft 2-3 variations that approach the decision from different angles.

This takes an extra few minutes. Those minutes will save you hours of follow-up clarification and produce dramatically more useful responses.

The Humility in Good Questions

There's something worth noticing about what makes questions powerful: they require us to admit what we don't know.

Asking "What am I overlooking?" acknowledges that our vision is limited. Providing context admits that our situation is particular, not universal. Inviting challenge accepts that our current thinking might be flawed.

This isn't weakness — it's wisdom. The best decision-makers aren't those who project certainty; they're those who systematically seek out what they're missing.

As we explore regularly on thonk, the quality of our decisions tracks closely with the quality of counsel we seek. And the quality of counsel we receive depends almost entirely on the questions we ask.

So before your next important decision, pause. Look at the question forming in your mind. And ask yourself: Is this the question that will get me the advice I actually need?

The answer might change everything.