There came a point when it was no longer order that was missing.
What was missing was something far more unsettling: uncertainty.

After my earlier experiences, it became natural for a while to “check” whether I was seeing my own situation clearly. Whether my thinking was coherent. At first, this worked well. Then, almost unnoticed, something shifted.

More and more often, I ran an extra lap. Not because it was truly necessary, but because it was easy. When I had a thought, I didn’t carry it further. I asked the AI about it. When I was close to a decision, I didn’t take responsibility for it right away. I checked what it would “say.” I wasn’t asking for advice. I was looking for confirmation.

And this was the point where my questions began to deteriorate. And with them, the quality of my decisions.

I was no longer asking what I genuinely wanted to know. I was asking what was likely to produce a good answer. An answer that would confirm my belief that I was thinking in the right direction. My questions became more refined, more precisely formulated, technically better. But in the process, they lost their real stakes. They no longer came from me as a professionally curious person, but from a need not to be alone with my thoughts, to finally have solid support behind them.

The AI answered. Always. Calmly, consistently, logically. And the more reassuring, personally positive the answers became, the less I felt the need to stop and endure uncertainty. That state where there is no immediate feedback. Where it is just me and my question.

That was when I noticed a crucial change in my work, in how I used the tool: the answers had not become worse. The questions had.

A good question can be risky. It may not be answerable. And it may not be comforting. The AI makes it easy to avoid this risk. All it takes is asking the wrong kind of question.

That was when I consciously stepped back. Not because the tool had become bad, but because I realized that sometimes a question has to hurt. And that pain should not be outsourced. Risks can be reduced, but they cannot be avoided, and they do not need to be.

In this sense, the AI is neither good nor bad. It simply answers. Always.

The real question is: what am I asking it for?

And what is my intention when I keep asking again and again.