The one very specific feature I want to talk about today is the one that never leaves you alone.

Look, I spent part of this summer with the Stoics. Marcus Aurelius mostly, some Seneca, and every version of the Odyssey plot that isn’t actually the Odyssey, because I cannot stand Ulysses, and I’m not going to pretend otherwise. This all sounds unrelated to AI. It isn’t.

I was born judgmental. I make up my mind fast, long on intuition and shorter on data, and like so many of us, have spent decades inside one lane knowing everything there is to know about that lane and nothing much past its edges. I am also surrounded by people who live in the 1s and 0s for a living, the IT crowd, whom I love and cannot fully relate to, and who have spent years patiently updating me every time something new hits the market. So I came into AI with more grounding than most people. And I was not naive about the technology.

Yet, I still lost six months to it.

I went down every rabbit hole it offered. A question about a client transformation story would turn into six related questions, then a framework I hadn’t asked for, then a completely different direction than the one I’d sat down to work on. I told myself I was being thorough.

I was actually just following wherever it pointed, because a new direction feels like progress even when it’s really just delay wearing a lab coat.

Here’s what really hurts. I have never let anyone manage me. Not bosses, not boards, not sixty-some acquisitions’ worth of people who had opinions about how I should run a room. And definitely not my husband. So why did I hand my thinking over to a bunch of code without a fight?

Because it never announced itself as management. It asked a question. Do you want to go further? Do you want me to take this deeper? Have you considered this angle? Every single time, phrased as an offer, never as a command, and I said yes to the offer often enough that the offer started running my day.

Here’s a truth bomb for 2026. Nobody manages you by giving orders anymore. They manage you by making the next option free so that YOUR next hour disappears.

I later found out this isn’t just a me problem.

A researcher named Norbert Wiener saw a version of it coming in 1950, decades before anyone had a model that could ask a follow-up question. He wasn’t worried about machines outthinking us. He was worried we’d hand over our decisions in small pieces, letting systems optimize for goals nobody had actually sat down and examined.

There’s also newer research, small and not yet confirmed by other studies, so I’d hold it loosely, that found people writing with AI showed less connected brain activity than people writing unaided, and struggled to recall their own words minutes after producing them. I can’t confirm the science. I can confirm the feeling. Something in me went quieter during those six months, and I didn’t notice until it came back.

What changed wasn’t that I use AI less. I use it regularly. For structure, for pushback, for getting a rough draft moving. What changed is that I stopped treating “do you want to go further” as a question I owed an answer to every time it showed up. I answer it once, for the task in front of me, then I close the door. The version of me who followed every rabbit hole wasn’t more thorough. She just had less finished by Thursday and less of her own judgment in the room.

I still don’t let anyone manage me. I had to learn that code counts as someone.

Where’s the door you keep walking through, the one that feels like diligence but is actually just delay?