Someone asks a question in Slack. It’s the kind of question that has a one-line answer: where a setting lives, whether a job runs nightly, who owns a service. A few minutes later the reply shows up, and it’s four paragraphs long.
It opens by restating the question. Then it explains some background nobody asked about. Then there’s a list of considerations, a couple of them reasonable, most of them generic. Somewhere around the third paragraph is the actual answer, and after it comes a closing line offering to help further. At the bottom, in small gray text: “Sent using ChatGPT”.
I’ve seen this enough times now that I can’t unsee it, and the first few times it genuinely stopped me. Damn. I get using these tools to help. I use them all the time. But this wasn’t help. As far as I could tell, nobody had read it before it went out.
The footer isn’t actually the part that bothers me. Plenty of people use AI to tighten a reply or check their reasoning, and you’d never know, and that’s fine. What gives it away is the “shape”. A colleague who knows you answers differently. They know what you already understand, so they skip it. They know you’re in the middle of something, so they keep it short. “It’s in the config, check the retry setting.” That reply has a model of you inside it. The four paragraphs don’t. They’re written for a generic person asking a generic version of the question, because whoever forwarded them never told the tool who was asking or what you already knew.
It’s correct. It’s thorough. And it isn’t addressed to anyone.
I’ve spent most of this year writing about what AI is changing in this industry: how developers get formed, who stops getting hired, what happens when everyone ships fast. This isn’t that. This isn’t about code at all. It’s about the experience of being answered by a person, which I never thought of as something you could lose, because for most of my career it was just what a reply was.
And I miss it. Not in a “kids these days” way. In the way you miss something so ordinary you never noticed it was there until it wasn’t.
The Stoics get accused of being against feeling things, and it’s mostly wrong. When Lucilius lost his friend Flaccus, Seneca didn’t tell him to get over it. He wrote: “Let not the eyes be dry when we have lost a friend, nor let them overflow.” Grief is allowed. What Seneca wouldn’t allow was grief that moved in and stayed. “I should prefer you to abandon grief,” he told Lucilius, “rather than have grief abandon you.”
Then, in the same letter, he admits he didn’t follow his own advice. He wept so hard for his friend Annaeus Serenus that he counted himself among the examples of people overcome by grief. The man giving the advice failed it, and said so, in writing, to the person he was advising.
I have my own version of that.
When one of those walls of text lands on me, I’ve handed it to an LLM and asked it to boil the thing down. More than once. It’s a reasonable thing to do with information, and I’d do it again. But follow it all the way through. Someone asks a question. A model writes the answer. A person forwards it. I receive it, and a model reads it for me and tells me what it says. There were two people in that exchange, and for a moment there wasn’t a person on either end of it.
There’s a theory that’s been floating around for a few years, the dead internet theory, that most of what’s online is just bots talking to bots. It started on the fringe and it’s mostly a joke now. I’ve never watched it happen in one of our threads. But I’ve been one end of the pipe, which is closer than I’d like.
Amor fati gets misread as acceptance, or worse, as liking whatever happens to you. At the opening of Book 4 of the Meditations, Marcus Aurelius describes the mind as something that “makes a material for itself out of that which opposes it,” the way a fire catches whatever falls into it. A small flame gets put out by what’s dropped on it. A strong one “rises higher by means of this very material.”
That’s what loving your fate actually asks. Not liking it. Using it.
And if I’m honest, this fate has handed me something. For twenty-five years, a short typed reply from a colleague carried no signal at all. Everybody sent one. It meant nothing that a person wrote it, because a person wrote everything. That isn’t true anymore. In a channel full of thorough answers addressed to nobody, the two-line reply with a typo in it, from someone who clearly knows what you’re working on, says something it never used to say: somebody was here. Somebody thought about you specifically, for thirty seconds, and wrote back.
The industry I came up in made that free. The one I have now made it rare. I didn’t choose that, and some days I still miss the old version. But a rare thing carries more than a free one ever did.
So I’m going to keep sending them. Short, a little rough, written to the person who asked. It used to be the least I could do. Now it might be the most.
👉 If you enjoy reading this post, feel free to share it with friends!
Or feel free to click the ❤️ button on this post so more people can discover it on Substack 🙏
You can find me on X and Instagram.
Real person here to say: great essay. When I receive AI-generated replies, I do feel offended because I feel the social contract we agreed to has been broken. When I read something from an individual sent by an individual and authored by an individual, there's an unspoken agreement for both parties to spend their time engaging in communication. When its clear they broke that agreement, then I'm fairly annoyed, because it feels like a reneging on the deal; like we went to shake hands and they pulled away at the last second.
So, I don't. My mantra is: if you can't take the time to write it, don't expect me to take the time to read it!