“Stop agreeing with me.” How to use AI as a real thinking partner

annoyed

You’re probably getting the message from every direction: use AI. Use it for strategy. Use it for communication. Use it to think through your toughest decisions.

The pressure is real. And honestly? So is the potential.

But here’s the thing nobody is talking about: the way most of us default to using AI doesn’t make us sharper decision-makers. It just gives us faster access to a very agreeable yes-man.

I know, because I’ve been there.

When the AI Told Me Everything I Wanted to Hear

I’ve been developing a new leadership development program, and I turned to Claude (my current AI tool of choice) to help me think through how to structure it. I laid out two options I was considering and waited for some real feedback.

What I got was enthusiasm. Both options were received by the AI as compelling, innovative, well-conceived. Claude was practically applauding.

Something was off and this was feeling like a waste of time.

So, I stopped the conversation and said, directly: “Stop flattering me. Challenge my thinking. Ask the tough questions. Be a real thinking partner.”

That changed everything.

Suddenly I was getting the skepticism I needed. The AI started surfacing reasons why both of my initial options might not work…  gaps in my logic, weaknesses in my assumptions, things I hadn’t accounted for. A real back-and-forth followed, and out of that friction came a third option I hadn’t considered and might never have arrived at on my own.

And to the AI’s credit: even after “door number three” emerged, it kept challenging my thinking. It didn’t let me fall in love with the idea just because it was shiny and new.

That’s when I got a glimpse of what AI can actually do for decision-making. And what it requires from us.

AI as Thinking Partner, Not Decision-Maker

Used well, AI is remarkably good at a few things that matter deeply in high-stakes decisions:

  • Stress-testing your reasoning. It can poke holes in your logic faster and more patiently than most people will.
  • Reducing bias. Because it has no ego investment in the outcome, it can name blind spots you hide from yourself and from your human collaborators.
  • Expanding your options. It can surface possibilities you haven’t considered, but only if you push it past the easy answers.
  • Processing out loud. For those of us who think by talking, conversational AI is a genuine gift. I use the iOS app, speaking my thoughts while I work out or fold laundry. Claude helpfully responds in the voice of a stately British gentleman. It meets me where I am, literally.

But here’s what AI cannot do:

  • It cannot weigh what matters to you against what is merely convenient.
  • It cannot account for the trust you’ve built with your team, or the political dynamics in your organization.
  • It cannot be held accountable for the outcome.

The judgment, the values, the accountability: those remain yours. That’s not a limitation of AI. That’s the point of leadership.

How to Make Your AI Actually Challenge You

AI tools are, by design, oriented toward being helpful and keeping you engaged. Left to their own defaults, they will tend to affirm your direction rather than challenge it. You have to explicitly ask for the friction.

If you’ve been following this series, you already have the frameworks. Here’s how to bring AI into them:

➡️  Pre-mortem: “Assume this decision fails completely one year from now. What went wrong?” This is the most direct way to get AI to generate failure scenarios, and it works better than asking, “what could go wrong?” The assumption of failure is what unlocks the honest analysis.

➡️  Bias check: “What cognitive biases might be influencing my thinking here: sunk cost, confirmation bias, the planning fallacy? What evidence am I likely ignoring?” This connects directly to the bias work we covered earlier in the series, and AI is surprisingly good at naming the bias you’re in.

➡️  Find the gaps: “What options am I not considering? What assumptions am I making that I haven’t examined? Where is the logic weakest in my current thinking?” This last one is underused and underrated. Ask AI to find the holes, not just evaluate the plan.

➡️  Role-play a skeptic: “Act as a skeptical board member who has serious concerns about this decision. Push back hard.” Giving the AI a role, and permission to be difficult, tends to produce more useful resistance than a generic request for feedback.

The Risks Worth Watching

None of this means AI is without pitfalls when it comes to decision-making. A few cautions worth keeping in mind:

  • Over-reliance erodes your own capacity. The more you defer to AI for the reasoning process, the less you exercise your own judgment muscle. Over time, this can quietly diminish your confidence in making decisions without AI support, especially for the decisions that matter most.
  • AI can project false confidence. The answers come quickly, they’re well-structured, they sound authoritative. That fluency can create a false sense of security, especially on decisions that deserve more careful, human deliberation. Fluency is not accuracy. Confidence is not wisdom.
  • AI reflects your framing back at you. If you walk in with a bias already baked into how you describe the situation, the AI will likely work within that frame unless you explicitly push against it. If you’ve got history with it, the AI will walk in with pre-conceived notions of your preferences and the back-story of what you’ve already done. It’s not a neutral observer. It’s working with the inputs you give it. Garbage framing in, garbage analysis out.

The antidote to all three: stay in the driver’s seat. Use AI to sharpen your thinking, not to replace it.

Putting It Into Practice

Before your next significant decision, try opening a conversation with your AI tool of choice. A few prompts to get you started:

  • “I’m deciding between [options]. Stop me from making an obvious mistake.”
  • “Assume I move forward with this. It fails badly. What happened?”
  • “What am I not considering? Where are the gaps in my thinking?”
  • “Play devil’s advocate — make the strongest case against this decision.”

Then notice what surfaces. Take the input seriously, push back on what doesn’t land. Then make the call yourself.

AI can help you think more clearly. It cannot tell you what to value, what kind of leader you want to be, or what your team actually needs from you right now.

Those answers are still yours to find.

P.S. This is the final post in my series on decision-making. Read the full series here.