I was listening to episode 51 of the Danish podcast Så må du have held og lykke med det… with Christiane Vejlø and Kristina May – the one with the gloriously unpromising title "Tissekonekage og Elons flyvende prutbil" – when a phrase stuck with me: AI shame.
Christiane described three kinds of it, and I recognised all three.
Three kinds of AI shame
The ones who haven't got on board. Everyone around you seems to be using AI for everything, and you haven't really started. Asking how now feels like admitting you've been left behind, so you don't ask.
The ones who got on board early. You've used it from the start, and now you worry people think you're cheating – that everything you produce is just the machine, and you're along for the ride.
The ones who got fooled. You hear a song or read something, and it gets to you. You get that little lift of liking something. Then you find out it was AI-generated, and the good feeling turns sour. Christiane compared it to discovering that your partner was a con artist all along.
A fourth kind, at work
At work we have started marking the things AI has made. I understand why. But somewhere along the way, using AI has started to feel slightly wrong, as if the label were a confession.
One colleague makes a point of marking work as "made by a human". And honestly, the human-made version is often the weaker one.
That is not a dig at my colleague. It is a point about the label. "Made by a human" tells you how something was made. It tells you nothing about whether it is any good. Neither does "made with AI". Something is not bad because AI was involved, and it is not good just because a person wrote it from scratch.
How this post was made
When I write a blog post, I start with an idea and describe it loosely: half sentences, things I've noticed, the point I want to land. Then I have Claude structure my thoughts. Then I go through the output, push back on what's wrong, and rewrite the sections that don't sound like me.
This post was made exactly that way. The result is better than what I would have written on my own in the same time, and it is still mine: the idea, the examples, the opinion, and the decision about what goes out under my name.
Where the shame belongs
There is a kind of AI use that deserves a little shame, and it is not the one people worry about. It is the unread output. Text nobody checked, code nobody ran, a summary nobody held up against the source. That is AI slop, and it wastes everyone's time – because then the reader has to do the work the author skipped.
So the line is not between AI and no AI. It is between reviewed and not reviewed. If you use AI, you owe the reader the effort: read it, question it, fix it, cut it, and make sure it says what you mean.
Would you put your name on it?
That is the test I use. Not "did AI touch this?" but "would I put my name on it?"
It also takes some of the sting out of the three kinds of shame. If you haven't started, ask – nobody learnt this without asking. If you started early, be open about how you work; the process is easier to defend than the secrecy. And the con-artist feeling comes more from not being told than from the AI itself. Being upfront about how something was made, and standing behind it, takes care of most of it.
What I would like us to stop doing is treating the label as the verdict. Look at the work.