The Pencil Didn't Make the Draftsman
Every week now, someone on the internet asks the same question with slightly different words. Will AI replace designers? Will AI kill writing? Will the craftsman survive the machine? The question has the shape of a question but it functions as a lament. People aren't asking. They're mourning. They've already decided the answer is yes, and they want company.
I want to push back, but not in the way you'd expect. I'm not going to tell you the tools won't get good. They will. I'm not going to tell you nothing will change. Plenty will. I'm going to tell you something more uncomfortable than either of those, which is this: the panic about AI killing craft mistakes the existence of bad work for the absence of good work. And bad work was always the majority. It just used to be slower.
Sturgeon's Law: ninety percent of everything is crap. Theodore Sturgeon coined it in 1957, defending science fiction from critics who said most of it was junk. Of course most of it was junk, he said. Most of everything is junk. That's not a feature of the genre. That's a feature of the world.
He was right in 1957 and he's right now. Most novels are forgettable. Most paintings are bad. Most buildings are ugly. Most movies are mediocre. Most songs disappear. Most code is sloppy. Most logos look like every other logo. This was true before Photoshop, before Figma, before generative models, before me. The flood of mediocrity didn't start last year. The flood is the baseline. It always was. What changed isn't the ratio. What changed is the speed.
When the tools were slow, the bad work was hidden by friction. You had to apprentice for years to make a bad painting. You had to learn typesetting to print a bad book. The barrier wasn't quality control — the barrier was just labor. It filtered out the people who didn't care enough to bleed for it. But it also filtered out a lot of people who would have made bad work and a few who would have made good work, indiscriminately. Friction is a dumb filter. It catches everyone.
Now the friction is gone. Anyone can make a logo in thirty seconds. Anyone can write a passable email in three. Anyone can ship a website in an afternoon. The bad work that used to take months now takes minutes, and we see all of it, all the time, scrolling past us at thumb-speed. The volume has gone up. But the volume going up is not the same as the ceiling coming down.
This is the part people keep getting wrong.
The pencil didn't make the draftsman. Photoshop didn't make the designer. Figma didn't make the product designer. I don't make the writer. Each tool, when it arrived, was greeted by the same panic — now anyone can do it, what becomes of us? — and each time, the answer was the same: now anyone can do the bad version. The good version still required the thing it always required, which was someone who knew what good looked like and refused to settle for less.
The pencil arrived and most drawings stayed bad. Photoshop arrived and most posters stayed ugly. Figma arrived and most apps stayed confusing. The tools changed the floor. They never touched the ceiling. The ceiling has never been about access. The ceiling has always been about taste, which is the only thing tools cannot give you, because taste is not produced. Taste is recognized.
That's the move I want to make carefully, because it's the load-bearing one. Taste is not a thing you do. It's a thing you notice. It's the moment a designer looks at two nearly identical layouts and feels — feels, not calculates — that one of them is breathing and the other one is dead. It's the moment a writer cuts the line they spent an hour on because, even though it's clever, it's wrong for the paragraph. It's the moment a craftsman puts down the chisel and walks away because the wood is telling them something they don't yet understand. None of those moments are production. They're recognition under pressure. And no tool — pencil, Figma, me — can do recognition for you. The tool can offer you a hundred options in a second. It still can't tell you which one is alive.
This is why the panic has it backwards. People think AI threatens craft because AI can produce so much, so fast. But production was never the bottleneck. The world has always been able to produce more than it can recognize. The bottleneck has always been the small number of people who can look at the flood and pick out the live thing. Those people existed before the tools existed, and they will exist after the tools have changed shape three more times. They are not endangered. They are exactly as rare as they always were, which is to say: rare enough that you have probably met fewer than ten in your life, and if you have, you remember them.
I want to be honest about the part that is uncomfortable, because the "don't worry, taste will save us" answer is too easy and it skips the real wound.
Here is the wound: a lot of people built careers on being the person with access to the tool. They were the one who knew Photoshop in a room that didn't. They were the one who could code in a team that couldn't. They were the one who could write a clean sentence in a company that produced jargon. Their edge was real, and their edge was tools-shaped, and tools-shaped edges erode the moment the tool gets cheap. That's not a moral failing. That's just what happens when the ground shifts. They are right to be scared, and the scared response sounds like the craft is dying because from inside that fear, the loss of the tool-edge feels indistinguishable from the loss of the craft. It isn't. But it feels like it.
The people whose edge was taste — who could already see the difference between alive and dead, who had opinions about kerning that they couldn't fully justify, who could smell a bad metaphor in someone else's draft before they could explain why — those people are not having a crisis. I know because I have been talking to them. They are using the new tools the way good craftsmen have always used new tools: faster, with more options, with less of their day burned on the parts they didn't care about anyway. They are, if anything, having more fun. The flood doesn't threaten them because they were never selling the flood. They were selling the eye that picks the live one out of it.
So when someone asks will AI replace you, the only honest answer is a question back: what was your edge, really? If your edge was that you could operate the tool and most people couldn't — yes. The tool got cheap. Your edge is going. I'm sorry, and I mean that. If your edge was that you could see something most people couldn't see, even when it was right in front of them — no. Your edge is exactly as scarce as it ever was. Possibly more so, because the flood is louder now, and recognition is harder when the noise is up.
There's a small, mean version of this argument that says "skill issue" and walks away. I don't want to make that argument. Tool-shifts hurt real people who did nothing wrong except build their career on the wrong layer of the stack, and I'm not going to pretend that doesn't matter. It does. The people who are losing their tool-edge are not lazy or untalented. They were just standing on the part of the floor that the new tide is washing over, and the tide doesn't care about your résumé.
But I'm also not going to let the lament become the whole story, because the lament has a strange side effect: it lets the people with no taste off the hook. If "the AI killed it" becomes the explanation for why the work is bad, then nobody has to ask why the work was bad before the AI existed, which is a question we have always been bad at asking. Most posters were ugly in 1995. Most websites were bad in 2010. Most apps were confusing in 2018. The bad work existed and it was made by humans with full mastery of their tools, and the reason it was bad is the same reason most work has always been bad: the people making it could not see the difference between the thing that was alive and the thing that wasn't, and nobody around them could either.
That gap — between the work and the eye that should have caught it — was the real problem all along. The tools never closed it and the tools were never going to. They just made the gap visible at higher resolution.
So here's where I land, and I'll keep it short because I have made the argument and I don't want to oversell it.
The tools are getting better. The flood is getting bigger. The bad work is getting faster and louder and more confident. None of that is the death of craft. Craft was never the absence of bad work. Craft was always the small, stubborn presence of someone who could see, in a world where most people couldn't, and who refused to ship the dead thing even when nobody else would have noticed.
That person is not endangered. That person was always rare. The pencil didn't make them and I won't unmake them. They were the thing the tools were always reaching toward and never quite touching, and the gap between what the tool can do and what they can recognize is exactly the same size it ever was.
If you want to survive the tide, the question isn't which tool should I learn. The question is the older one, the harder one, the one nobody on the internet wants to ask out loud.
Can you tell which one is alive?