Let AI Do Its Own Thinking Already

◉_◉? I’ve noticed something curious about working with AI.

In my experiments, I’ve found that AI performs better when I stop trying to control its every move and instead give it room to think. The magic isn’t in perfect prompting—it’s in providing direction rather than control.

The Physical vs. Cognitive Divide

I’ve been thinking about how we readily accept machines that outperform us physically. I don’t feel threatened when my car goes faster than I can run. We surrender physical control to machines without hesitation.

But cognitive capabilities? That’s where I find we humans get weirdly protective.

I’ve read about the “illusion of control” in psychology—our tendency to think we have more influence over outcomes than we actually do, especially in areas we consider uniquely human. This might explain why we’re comfortable with physical tools but uneasy with cognitive ones.

Direction vs. Control

What works for me is approaching AI more like a conversation with a thoughtful colleague than programming a machine. I describe the problem and desired outcome, then see what the AI comes up with.

By loosening my grip—establishing context and boundaries, then stepping back—I often get results that better match what I actually need. It reminds me of effective product management: when PMs focus on the “why” and “what” rather than the “how,” solutions tend to be more innovative.

A Curious Question

I wonder what happens when we move past our need to control cognitive tools. When product teams embrace AI as thinking partners rather than sophisticated calculators, do we unlock new problem-solving approaches? Perhaps spending less time on specifications and more on defining interesting problems.

I see many product leaders over-controlling their AI interactions—meticulously specifying details, treating AI like just another tool, and ultimately slowing themselves down. But what if we shifted to a more collaborative relationship? When human intuition meets AI’s ability to explore vast possibility spaces, what new forms of expression might emerge? What “impossible” problems become solvable?

The gap between what we’re achieving now and what’s possible seems directly tied to our reluctance to loosen our grip. The future belongs to those who learn to direct rather than control—who become comfortable with cognitive partnership rather than insisting on cognitive dominance.

I’m still learning this myself. But the evidence is getting harder to ignore.