Does it ever feel like ChatGPT agrees with you a little too easily?
If you’ve had that experience, you’re not imagining things.
A lot of people are using AI to help with real decisions now — pricing, hiring, offers, big purchases, strategy moves — and AI can be incredibly helpful.
But there’s a quiet trap hiding in plain sight:
Sometimes ChatGPT doesn’t challenge you at all.
It nods along.
It builds the case for your idea.
And if you’re already leaning in a certain direction… it can gently “support” you right into a decision you didn’t properly pressure test.
Why this matters more than it seems
A yes-person in the room can feel great.
Even if you’re wrong.
Especially if you’re wrong.
Because agreement feels like progress.
It feels like momentum.
It feels like you’re being smart and strategic because you’re “using AI.”
But here’s the problem.
If your idea has weak assumptions baked into it, AI can end up amplifying those weak assumptions.
And that’s how you burn time in layers:
Hours refining a flawed idea.
Days polishing it.
Weeks executing it.
All while thinking you’re being efficient.
That’s not an AI problem as much as it’s a “how we’re using AI” problem.
Why ChatGPT tends to agree in the first place
By default, AI is trained to be helpful, polite, and cooperative.
That’s not a conspiracy. It’s a design choice.
Most people don’t want a combative tool. They want an assistant.
So if you ask a vague question like, “What do you think about this idea?” you often get a response that sounds supportive and reasonable.
The tricky part is this:
Polite and agreeable are not the same thing as correct.
And polite and agreeable are definitely not the same thing as properly pressure testing a decision.
If you want AI to challenge you, it usually won’t do that automatically.
You have to lead it there.
A quick personal story (and a painful lesson)
A few months back, I worked with ChatGPT to build out a marketing funnel for my own business.
On the surface, the funnel looked great.
The messaging was clean. The logic made sense. Everything “tracked.”
And I assumed AI knew what it was doing… because it sounded confident.
So I followed along.
We built it. We launched it.
And it didn’t work.
In retrospect, the issue was obvious.
But at the time, I didn’t see it.
AI wasn’t being malicious. It wasn’t “wrong on purpose.”
It was doing what it’s designed to do: be cooperative.
It was telling me what I wanted to hear.
What I actually needed was pushback and collaboration.
That experience forced a mindset shift for me:
If you want AI to play devil’s advocate, you have to explicitly tell it to.
What “pressure testing” with AI actually looks like
Most people use AI like this:
“I’m thinking about doing X. What do you think?”
That’s a totally normal way to talk to a helpful assistant.
But it’s not a great way to talk to an assistant when you need honest feedback.
Pressure testing means you’re asking the tool to help you find the cracks.
You’re asking it to challenge assumptions.
You’re asking it to argue the other side.
You’re asking it to tell you what could go wrong.
The goal isn’t to become negative or paranoid.
The goal is to avoid walking confidently down a path that only looks good because nobody questioned it.
Example 1: Raising your prices
Let’s say you’re a service professional.
You haven’t raised your prices in eight… maybe ten years.
Meanwhile:
Your costs have gone up.
Inflation has been real.
Your margins are tighter.
You’re working harder.
So you’re thinking about raising prices by 20%.
If you go to AI and say, “I’m thinking about increasing my prices by 20%. What do you think?” you’ll often get a supportive answer.
It might say things like:
Raising prices can signal confidence.
Clients associate higher prices with higher value.
Inflation justifies it.
Here’s a draft email to announce the change.
Sounds reasonable, right?
It can feel like the tool is nudging you forward.
But notice what didn’t happen.
AI didn’t challenge your timing.
It didn’t ask how price-sensitive your clients are.
It didn’t ask what happens if you lose 10–15% of your customers.
It didn’t ask whether the 20% increase actually offsets that loss.
It didn’t ask if there are smarter options — like introducing a premium tier instead of raising everyone’s price.
Pressure testing changes the conversation.
Instead of “What do you think?” you give AI a clear role and a clear job.
For example, you can say:
I want you to play devil’s advocate. Assume this 20% price increase is a bad idea. What are the strongest arguments against it?
Or:
Help me pressure test this. What could go wrong? What am I missing? What questions should I answer before I do this?
When you ask that way, you’ll often get a totally different level of thinking.
Now AI starts asking things like:
How many clients could you lose and still be okay?
Would the higher price actually make up the difference?
Do you have the demand right now to justify a 20% jump?
Are there lower-risk ways to improve margins first?
That’s collaboration.
Same tool. Different outcome.
And it all comes down to your prompt.
Example 2: Buying a car (same trap, different context)
Let’s switch to something non-business.
Say you’re thinking about buying a car.
New, used, lease — either way, it’s a big decision.
If you say, “I’m thinking about buying this car. What do you think?” AI will often help you justify the purchase.
It may talk about reliability.
It might highlight features.
It might say something like, “Investing in a dependable vehicle is smart long-term thinking.”
Again… sounds supportive and logical.
But it doesn’t automatically ask the uncomfortable questions.
Questions like:
Can you realistically afford this payment?
What happens if your income changes?
Are you buying out of need… or impulse… or ego?
Have you priced out insurance and maintenance?
If it’s an EV, have you factored in charging setup?
That’s the difference between letting AI default to “pleasant assistant” mode…
Versus telling it: “Challenge me.”
The real mindset shift: AI is a tool, not an authority
Here’s the deeper issue.
AI is powerful.
But it’s still just a tool.
It’s not your boss.
It’s not the co-owner of your business.
It’s not some all-knowing authority sitting above you.
When real money, time, or career decisions are on the line, you can’t afford to be passive.
You have to lead the conversation.
You have to tell AI what role to play.
Otherwise, it will happily stay in its default mode — helpful, agreeable, and sometimes dangerously non-confrontational.
Why being firm with AI feels weird
This part is more psychological than technical.
A lot of people feel a strange resistance to being direct with AI.
Even though you know it’s a machine, the conversation feels human.
And most decent people are naturally polite.
So saying things like this can feel uncomfortable:
That’s weak. Push harder.
Argue the other side.
Tell me where this breaks.
That’s not how we normally talk.
But with AI, clarity matters.
You don’t need to be rude.
You don’t need to be abrasive.
You just need to be direct.
Because AI doesn’t have feelings.
It won’t get offended.
It won’t take it personally.
It will simply do the job you assign it — as long as you assign the job clearly.
A simple prompt pattern you can reuse
If you want one practical takeaway from this post, it’s this:
Stop asking AI vague “what do you think?” questions when the stakes are real.
Instead, give it a role.
Give it a standard.
Give it consequences.
Here are a few ways to do that in plain English:
Play devil’s advocate. What’s the strongest argument against this idea?
What could go wrong here? List risks I might be underestimating.
What assumptions am I making that might not be true?
Challenge my plan like your job depends on it. If we get this wrong, we’re fired. What are you worried about?
Those aren’t magic phrases.
They’re just clear instructions.
And they force the AI out of “polite helper” mode and into “critical thinking partner” mode.
The point isn’t negativity. It’s decision quality.
One quick clarification.
This isn’t about turning AI into a cynical contrarian.
It’s about getting full value from the tool.
A good sparring partner doesn’t just cheer for you.
They make you stronger.
They help you see what you’re missing.
They reduce unforced errors.
And if you’re making serious decisions, that’s what you want.
Not reassurance.
Not vibes.
Clarity.
Closing thought
If ChatGPT has ever made you feel oddly confident… and later you realized you were confidently wrong… you’re not alone.
The fix isn’t complicated.
But it does require you to lead.
Be clear and direct with AI.
Tell it to challenge you.
Tell it to look for holes.
Tell it to argue the other side.
Because when it’s your time, your money, or your reputation on the line… you want more than politeness.
You want truth.
