The prompting technique that forces you to look at your problem from three angles at once

You’ve been there, right? You ask ChatGPT, “Hey, explain how AI is affecting advertising,” and it spits out three paragraphs straight from a corporate brochure. The classic “on one hand… on the other hand…” routine — polished, polite, and utterly useless.

Maybe you’ve tried the “be a skeptic vs. optimist” trick. It’s fun, sure, but still pretty shallow.

Now imagine this: instead of asking for an analysis, you tell the AI to stage a real conversation between two professionals who see the world completely differently.

That’s what I did — and honestly, it changed the way I think about this whole industry.

How I set up the experiment: two guys, one tense chat

It was simple. I asked Claude to play two characters:

Fernando: lifelong creative director, 25 years in agencies. To him, AI is the enemy of true creativity.

Martín: digital director, 15 years, navigating the post-internet era. He sees AI as a tool that amplifies what you already do well.

The key: I didn’t want a staged debate. I told them to talk like real people — interrupting, hesitating, admitting when the other had a point, even changing their mind mid-conversation.

The result? Way better than any “general analysis” you’ll ever get.

Why it works (and no, it’s not magic)

When you ask AI to talk “in general,” it gives you its average answer. Like searching Google for “the most typical page about this topic.” Safe, bland, and forgettable.

But when you make it play two concrete characters — with backgrounds, biases, and pressure — it shifts gears. The AI has to:

• Stay consistent: Fernando can’t sound like Martín. That forces it to build stronger arguments.
• Create real tension: It can’t end with “well, both are right.” It has to negotiate the conflict, like humans do over a beer.
• Explore nuance: The best moment is when Fernando goes, “You know what, you’re right.” Not surrender — evolution.

In my case, we moved from “AI will destroy agencies” to a real debate on why 86% of brands are happy with their in-house teams (Martín’s data), which made Fernando wonder if that’s because teams got better — or because clients simply lowered their expectations.

That’s actual thinking.

What I discovered (and didn’t expect)

The dialogue revealed tensions no normal analysis would catch:

• The problem isn’t AI — it’s how we use it. The fear is mediocre output. The truth: most people use it mediocre.
• The real 80/20 rule. Fernando, the skeptic, thought agencies were doomed. Martín got him to admit maybe 20% of the work — strategy — remains deeply human.
• Evolution, not apocalypse. They ended up agreeing: the industry might shrink, but it’ll get better.

How you can do it (in 5 steps)

1. Define your “fighters.” Give them clear biases, not neutrality.

2. Add context (and pressure). Make the setting realistic and charged with stakes.

3. Ask for sparks. Tell the AI to make them challenge each other and change opinions if it fits.

4. Demand references. Cite concrete sources or say when unsure — keeps hallucinations in check.

5. Watch for turning points. That’s where learning happens.

What it’s good for

•  Anticipating internal objections.
• Understanding complex problems deeply.
• Exploring future scenarios with richer insight.

A fair warning

This isn’t real life — it’s enhanced reality. AI still tends to be too reasonable. Think of it as a sparring partner: tireless, honest, unoffended when you say, “That’s nonsense.”

Start today

Open ChatGPT or Claude. Create two characters with opposite opinions on something that bugs you at work. Give them context. Let them talk. Then notice the exact moment you change your mind.

That’s thinking. And that’s what real AI is for — not automating what you do, but amplifying how you think.