Every Time I Write About AI, I End Up Writing About People

I did not set out to write a series. I set out to write one post about decks, and then a thing kept happening: every time I sat down to write about AI, I ended up writing about people. Three posts in, the pattern is too loud to ignore, so let me name it.

First I wrote about my twenty decks. AI made me better, stronger, faster at communication, and adoption did not move an inch, because communication was only ever one of five conditions people need before they change. The production got superhuman. The lever stayed human.

Then I wrote about slop. Everyone is receiving AI slop and everyone is furious about it, and the fury is aimed at the machine when most of it belongs to the delegation. Vague requests, no standard, no review, and a first draft shipped to someone else as quality control. The generating got superhuman. The standard stayed human.

Then I wrote about failing fast, which made me laugh, because failing was never the part that needed accelerating. AI compressed the try loop until I could fail more before lunch than I used to fail in a quarter, and it did nothing at all for the climb back out. The attempts got superhuman. The recovery stayed human.

One lesson, three costumes

Line those up and it is the same lesson wearing three costumes.

AI is a production revolution, and it is often mistaken for an adoption revolution, a judgment revolution, or a resilience revolution. It is none of those. It removes constraints on making things: decks, drafts, analyses, attempts. And when you remove every production constraint at once, something clarifying happens. You find out where the work actually was.

It was never in the making. In my field we have said this for forty years in four words: installation is not implementation. You can install a system, a process, a strategy, or now a mountain of beautifully formatted AI output, and none of it counts until people change what they do. The gap between the two is where change efforts have always died. AI did not close that gap. AI paved everything right up to the edge of it and made the drop easier to see.

That is the honest through line of these three posts. Willingness, ability, confidence, and control cannot be generated. A standard cannot be generated. Getting back up cannot be generated. Everything on that list was the hard part before AI, and it is the entire part now, because it is all that is left.

Which means the stakes went up, not down

Here is what I want to leave you with, because it is the opposite of the replacement story. And I do not say it lightly. Friends and family have been calling me about layoffs for over a year now. Lately the pace has been three calls in three weeks, but this is not a recent phenomenon. It is a season of disruption that has persisted for over a year. Some of the hardest calls are from people laid off a year before that, who are only now finding the courage to say it out loud. That is what this disruption actually does to people. It does not just take the job. It takes away the ability to articulate feelings for a while too.

So understand that I am not writing any of this from the cheap seats. I understand the ramifications. AI is affecting our jobs. It is affecting our work. It is affecting our futures, and pretending otherwise would insult the people on the other end of those calls. This series is not an argument that the disruption is overblown. It is an argument about where solid ground is while the disruption is real.

If AI does the production, then the human contribution concentrates in exactly the things this series kept running into: creating the conditions for people to adopt, holding the standard for what is good, and having the stubbornness to get up and go again. Those were always the differentiators. They were just easy to hide behind busyness. A full calendar of deck-building looked like contribution. A pile of drafts looked like progress.

That cover is gone. When the machine handles volume, the only thing you can be is the part the machine cannot do. For some people that is a threat. It should be a promotion. The judgment, the involvement, the reinforcement, the recovery: this is the senior work, and suddenly it is the only work.

So when I say AI is coming for the excuses more than the people, I say it knowing real people are getting real calls, some of them mine. What I mean is this: the surest ground to stand on is the work no machine does. It already took mine. I can no longer blame the deck, the draft, or the pace of iteration for a change that is not landing, because all three are now effectively free. What is left is the question that was always underneath: have I done the human work?

Twenty decks, a thousand prompts, and more failures than I am going to count later, that is the whole series in one sentence. The machines got faster.

The work came home.