Sean Percival

Essays · October 2026

The Absolute Boredom of Losing Your Job to AI

Oslo
An Edvard Munch-style painting after Melancholy: a dark-haired man in a black coat rests his cheek on his hand at a rocky Norwegian shore at dusk, while in the distance three figures on a jetty load boxes into a rowboat, one of them carrying a laptop

Everybody is worried about AI taking their job. I have looked at my inbox and I am not sure AI would want it.

Picture a frontier model, trained on the entire sum of human knowledge, sitting through a 45-minute call with the word "alignment" in the title. Alignment on what. Nobody knows. Imagine the vast computational power of these models trying to decipher the meaning of a "synergy workshop," a "stakeholder touchpoint," and a "deep dive." If AI ever does become sentient, this is the part where it quietly unplugs itself.

Now this isn't just an essay to complain about the drudgery of corporate life. I haven't even mentioned the expense reports and never-ending flow of email! Mostly because those are two jobs I have already given to my AI to manage. Whether it wants them or not. So AI has already taken some of the boring parts of my job, but it's also taking some of the fun parts. Basically the job itself.

The job is the AI now

The more honest version is that my job isn't being taken by AI. My job is AI. Pretty much all the work I do that moves anything happens in Claude now, with a pile of MCP connections hanging off it, and I avoid every other tool as much as I'm allowed to.

And it's getting a little boring.

I didn't expect that. The output is better. More of it, more accurate, and I can actually show my work. The old way, I'd spend a day pulling exports from four systems, stitching them in a spreadsheet, and presenting a number I was maybe 80 percent sure about. "I think it's right" was a legitimate sentence in a leadership meeting. Now the number comes with its own sources and I'm bored by how right it is.

Someone built a chairlift on my mountain

The best way I can describe it: somebody installed a chairlift on my job.

For twenty years the work was the climb. You had to dig into systems, argue with a CSV file, call someone on the phone, figure out why the numbers from two tools disagreed by exactly 11 percent every month. It was slow and dumb and most of it was not "strategic." But when you got to the top, you'd earned the view. You learned something on the way up. You could feel the altitude.

Now I sit down with a problem that used to take hours, sometimes days, and I'm at the summit in five minutes. Still in my boots. Haven't broken a sweat. The view is the same view, arguably a nicer one, and yet I feel almost nothing. The payoff was always proportional to the climb, and the climb is gone.

The weird part is the fun hasn't disappeared, it's just moved. The lead-up is still great. Figuring out how to make it work, which tools to connect, rewriting the prompt out of spite until it one-shots the thing. That's the good drama. Then it's done, and all that's left is tweaking. You stand there at the top of the chairlift adjusting the font.

Nobody rides up with you

I wrote before that the excitement of building anything with AI is non-transferable. Everyone walks in for the last four seconds and sees a table with data. The interesting part does not survive the trip from your machine to someone else's eyes, because everything that made it exciting happened before the part they're looking at.

As things get more automated or "agentic," the ride is even lonelier. Now things are just happening in the background, much like a dishwasher. Once my automation is dialed in, I'm not even along for the ride! The thought, the tweaking, the eleven attempts to get the query right, all of it is baked into a thing that now executes silently while I'm asleep. Some of my automations have gotten so good they just fix themselves when something goes wrong. I automated the climb, and then I automated the summit, and now I check my inbox like a man reading his own obituary. Accurate, well-sourced, mildly boring.

So what does this mean

I'm doing more interesting work thanks to AI, if you measure interesting by what gets produced. More intelligent work, in the sense that the thinking is denser and the manual grind is thinner. And it is less satisfying, almost in direct proportion.

I don't think this is a me problem. I think it's what work feels like once the chairlift is everywhere. The people who are good at their jobs will produce staggering amounts of excellent output and feel faintly flat about it, and the people who aren't will still be in those synergy workshops, which is the one place the chairlift refuses to go.

Maybe the answer is bigger mountains. Harder problems, the kind that still take a week even with the lift running at full speed. I'm trying. Most days the problem in front of me is a report, and the report is done before my coffee is.

AI may or may not want my job. But it keeps helping itself to the good parts.