Who’s gonna fix the plumbing?
A single comment under my note about AI adoption sent me down this whole line of thinking.
My first reaction was: you are so damn right.
The world doesn’t revolve around AI, nor around the “future of work” conversation. Most people don’t care about posting on Substack or LinkedIn — it’s too abstract a concern when you’re still battling first-line problems. What they cared about first was having their basic human needs met, and feeling taken care of.
This is the part of the AI conversation nobody in the AI conversation seems to remember: most of the world is NOT in the AI conversation.
We talk about “the future of work” as though work were one thing — happening in one kind of room, affecting a type of person you can easily describe: someone sipping coffee at a desk, taking Zoom calls, most likely 9 to 5, glued to a small or big screen, whose part of the job can be performed by a model update.
For a narrow, loud, extremely online slice of humanity, that’s true. But step outside that slice and the picture changes fast. Agricultural workers don’t ask an AI agent when it’s the right moment to begin harvest. The electrician doesn’t diagnose a wiring fault with AI support. The nurse changing bedsheets at 3am isn’t thinking about how leveraging AI could automate her workload, because it can’t.
There are plenty of professionals untouched by the ongoing AI changes, still doing their job like before, or with only minor adjustments — the truth is, there are more people fixing pipes, wiring houses, growing food, and taking care of analog human bodies without a gram of AI than there are people worrying about whether a chatbot will take their job. We just don’t hear from them, because they’re busy — with their hands, in the physical world, on problems that don’t resolve with a better prompt.
Here’s the shift most future-of-work thinking refuses to make: it treats visibility as universality. If a type of work is well-documented, frequently tweeted about, easy to bring up over coffee with people who do the same kind of work, it starts to feel like the kind of work — the default, the center. Everything else becomes “other,” background noise nobody bothers to describe in detail.
Fewer people share their story about reinventing themselves and learning how to lay tiles than people who pursue entrepreneurship, digital business, e-commerce, or marketing — anything that can scale off the back of globalization, anything with a deck.
The background is where your pipes actually get fixed by someone crouched under a sink at 7am. It’s where a roof gets patched before the rain comes, where a farmer checks the soil with their hands instead of a sensor, where a nurse sits with someone at 3am because a machine can’t hold anyone’s hand.
The center of the discourse and the center of the economy are not the same place, and mixing them up is how you end up asking whether AI will “replace work,” as if work were one uniform, screen-shaped thing instead of a thousand different rooms most of us will never set foot in.
Now, the part I have to say to you directly, since you’re the one reading this:
I’m writing to you from inside Pandemonium.
So are you (most likely) or you wouldn’t have made it this far into an essay about AI and work written in August.
That’s not an accusation. It’s just where the light is. We built ourselves a small, loud, well-lit room and started mistaking its glow for the whole world, and I don’t think either of us is entirely innocent of that.
Now ask yourself:
When was the last time you saw someone panic about robots coming for the person sweeping your street at dawn or trimming the hedge outside your building?
You didn’t — and it’s not because their jobs are safe by accident. It’s because, somewhere underneath all our noise, everyone already knows. The economy has a floor, and that floor is made of people nobody’s rushing to replace. It’s still cheaper, and frankly still better, to hire a human for the things that require a human. Taking care of other human beings was never going to be automated out from under us. It’s just easier to write essays about the ceiling than to notice who’s holding the whole thing up — and easier still to write them from the ceiling, which is exactly where I am right now, typing this.
We are building ourselves an exclusive hell on earth, and on purpose, we forgot to invite most of the world to burn in it with us. Oops. How millennial of us, to build a whole civilization’s worth of anxiety around a problem most people don’t have the luxury of worrying about, and call it universal because everyone we know is worried about it too.
So before the next think-piece about which jobs AI will hollow out — including (possibly) this one — go find whoever’s actually holding the wrench, the mop, the syringe, the bag of soil, and ask them what they think about any of this.
You already know what they’ll say.
They don’t have time to worry about it. They’re too busy doing the thing no model can.
No essay, including this one, was ever going to do it for them.
Fin.
While you’re here:
Not sure I agree. In India, small models are being explored and piloted in agriculture, rural energy services, sexual reproductive healthcare provision, etc.
AI application is not limited to knowledge work
Love this reminder, Lucy, and I totally agree. AI may be finding its way into most industries, but that doesn’t mean it’s reaching every level of work, or every person doing it. I think those of us who spend a lot of time thinking, talking and writing about AI can very easily forget how small our little corner of the conversation actually is. There’s something very human about that :)