Will Smith eating spaghetti is my favorite benchmark. Rather, I should say it was.
These days, AI generations just look like Will Smith agreed to shoot a silly little video leaning into a meme from 2023.
We’re still in the first third of 2026, and his grainy, contorted face morphing through bastardized clumps of pixelated noodles already feels quaint.
Almost nostalgic.
That original AI video of Mr. Smith dropped about 20 months ago. In baby terms, we still gauge elapsed time in months when it’s less than two years.
Now it’s laughable to think that we could discern the legitimacy of any spaghetti footage, any celebrity footage, any war footage. The tail wagged the dog. The tail is the dog. There is no spoon.
Quite some time ago, I wrote about the fading moment when folks should cherish their confident ire. That momentary window when you could point and laugh. Definitively classify. That this specific thing was made with AI. Silly computer, and silly person for trying to pass this off as your own.
But then, as things tend to do, the technology improved. And the pointing stopped.
Not because the pointers changed their minds. Because the target moved. It got better faster than the ridicule could keep up, and one morning the guffaws just didn’t land anymore.
The Rubicon whizzed by. Silently. Under the mockery. Past the gathered eyes squinted in ridicule.
And what I’ve been turning over is this: I don’t think the mockery phase was ever really about the technology.
---
“Silly computer” was always shorthand for I still matter, right?
The guffaws weren’t aimed at bad spaghetti renders. They were aimed at the implications.
If the thing can do the thing, then what was the value of that thing? Was it ever about the doing? Was it in the value of the thing having been done? Was it about the identity of having done it?
Some of the people pushing back did know what they were fighting AI over. Not a question of “is this good?” but “what did someone mean by this?”
And those people weren’t wrong. They were just early. They felt the divide before anyone had language for it.
Because, to them, it didn’t look like a river to cross. It looked like a fork in the road. And most people took the wider path without noticing it was a choice.
---
The wider path is the one where you use the tool because it’s there. Because not choosing is always easier than choosing. Because energy flows downhill and when the simple option becomes available, nobody needs to justify using it.
Nobody asks why you used the easy thing. The default is invisible. Only the deviation needs a reason.
The narrower path isn’t narrow because fewer people want it. It’s narrow because it now requires declaration.
Before, you were just making things. Now you’re making things by hand, and the “by hand” part is load-bearing. It’s a modifier where there used to be no need for one.
That’s the real indignity. Not that the machines got good. It’s that the humans who do the thing now have to explain that they do the thing.
The craft became its own adjective. And elbowing for room on that path means constantly answering a question nobody used to ask: why this way?
“Because I meant it” is a real answer. But it’s an exhausting one to keep giving.
It’s exhausting because the question never stops.
Every fair, every portfolio, every inbox, every scroll past a generated image that took someone four seconds and looks 80% as good.
The answer has to be re-earned each time. Not once. Not as a manifesto. But Tuesday, and then again Wednesday, and then again when someone asks why you’d bother when the machine can do it faster.
The wider path doesn’t require any justification at all. It just is.
The narrower path demands you keep choosing it, every day, against friction.
The fork didn’t make people simpler. It just made the simpler option widely available.
---
But here’s the thing nobody on the wider path gets: the friction isn’t a cost. It’s the finish.
The resistance is what gives the surface a sheen capable of catching light.
A hand-thrown pot has tooth to it. The lathe marks, the slight asymmetry, the place where the thumb pressed too hard. Those aren’t defects the potter failed to eliminate. That’s the record of a negotiation between intention and material.
The imperfection is the polish.
You can prompt for “slight imperfections, handmade feel.” But that’s decoration performing as declaration. It’s the difference between a scar and a tattoo of a scar.
And the narrower path doesn’t just get narrower. It gets higher resolution.
The people still on that path become more attuned to what they’re actually doing and why. The elbowing then forces an articulation that the unchallenged maker never needed.
Articulation sharpens intent.
You start making things harder on purpose and discover that the hardness was where the meaning lived all along.
Which is a weird gift from the machines, if you’re willing to accept it from that direction.
They didn’t devalue the craft.
They burned away everything that wasn’t the craft and left the core standing naked, obvious, maybe for the first time.
Because for a while there, I think fandom absorbed craft.
The consumption apparatus got so good at celebrating the output that it swallowed the process whole. You didn’t need to understand how a thing was made to have a sophisticated opinion about it. Taste became the skill. Curation became the craft.
And that worked, for a while, because there was still a bottleneck: someone, somewhere, had to actually make the thing. Fandom could orbit the output all it wanted, but the maker was still at the center, even if nobody was looking at them. Gravity still pointed inward.
Generative tools dissolved that bottleneck.
Suddenly fandom didn’t need a center anymore. You could go from taste to artifact without passing through craft at all.
The orbit became the thing.
And for a lot of people that felt like liberation: I can finally make the thing I could always see in my head. Which is real. That’s a real feeling. I’m not here to diminish it.
But what got lost in that collapse is something almost impossible to explain to someone who hasn’t felt it: the resistance changing you.
Not the result of you shaping the material. The material shaping you back.
The pot pushes against the potter’s hands and the potter becomes someone slightly different by the end.
The sentence fights the writer and the writer discovers they meant something they didn’t know they meant.
The chord progression resists the melody and the melody bends into something the composer couldn’t have planned.
That feedback loop between maker and medium is where the actual transformation happens. The artifact is almost a byproduct at that point. A receipt. Mere residual proof of someone’s change in the making of it.
When fandom absorbed craft, it kept the receipt and discarded the transformation. Which looks like a clean trade from the outside. You still get the product, in abundance.
But nobody was sharpened in that process.
Nobody discovered a creative solution to a problem that shouldn’t have had one. Nobody sat with a failure long enough to learn that the failure was the point.
The maker’s perspective didn’t just shrink. It was reclassified.
The maker became an absurd alchemist of ideas. Someone who combines inputs rather than a person with a point of view forged by resistance. And we barely noticed, because the outputs kept coming, and they tasted fine, and they tickled an emotion, and they left without lingering.
---
I started writing this as an argument against the holdouts. The ones who refuse to touch the tools. I wanted to say: you’re missing something, the spark is real, come see.
But the piece mutated on me, as things do when you actually sit with them long enough.
Because the holdouts aren’t missing anything. They’re choosing something.
They’re choosing the narrower path, the daily re-declaration, the exhausting answer.
And in doing so, they’re becoming more themselves. More articulate about why they make what they make. More resolved. More burnished by the friction that the wider path avoids entirely.
This isn’t about output. It was never about output. It’s about what happens to the person making it.
The wider path gives you the thing. The narrower path forces you to become the best maker of the thing.
And I think we’ve chosen to discount the value of the maker so egregiously at this point in capitalism’s decline.
Partly in the name of profit, partly in the name of access, partly just because the wider path looked like the only path. We’ve hollowed out the only perspective that was never supposed to be optional.
So this isn’t a screed. It’s a love letter.
To the ones still on the narrow path. The ones answering why this way for the hundredth time and meaning it every time. The ones whose surfaces are catching light because of the friction, not in spite of it.
The machines didn’t make you obsolete. They made you visible.
Now is your time, too.
Author’s Note
This piece started as a screed. I wanted to write about the people still mocking AI-generated content—the ones clinging to confident ridicule while the technology had already lapped them. I had a take. I had momentum. I had the kind of righteous clarity that feels great for about forty-five minutes.
Then the tools did what this essay argues tools do: they pushed back.
I brought the first draft into Cowork—Anthropic’s document-focused AI workspace—expecting it to help me sharpen and finish. Cowork didn’t do that. Cowork is optimized for efficiency, not for hand-holding, and its clear-eyed response eviscerated my initial position. It didn’t soften the counter-argument to keep me engaged. It handed it to me like a brick. I think that kind of honesty only comes from a tool focused on getting the work done, not one designed around turns and sessions and keeping you coming back.
Cowork didn’t care about my feelings because caring about my feelings wasn’t its job.
That broke the piece. I stopped writing yesterday because I didn’t know what I thought anymore. The screed was dead and I hadn’t figured out what replaced it.
I took the rubble to ChatGPT to recalibrate.
ChatGPT helped me see what I’d interpreted as a failed argument was actually the beginning of a different, better one. It helped me find my footing.
Then I brought the reframed draft to Claude—this chat interface, the conversational one—and we did what we always do. Intellectual patty cake.
Claude helped me find the structural arc: open with elegy instead of indictment, expand the fatigue of the narrower path, separate the fandom collapse from the transformation payoff, and land the ending as a love letter instead of a lecture.
Three tools. Three roles. A strange orchestration bouncing between them, feeling the friction of each surface on the final output.
Cowork broke the piece by being too efficient to be gentle. ChatGPT helped me find my footing in the rubble. Claude helped me rebuild in my own voice.
And the whole process became the essay’s argument.
I set out to write about friction shaping the maker, and the writing process was friction shaping the maker. I came in with a position, the resistance changed me, and the piece I ended up with is proof of someone being changed in the making of it.
This artifact is just a byproduct. A receipt.
Conversations linked below:
Cowork session link
ChatGPT conversation link
Claude conversation link