Brian Merchant put it as plainly as anyone can:
"The Luddites were not anti-machinery; many of them were machine experts and welcomed the introduction of new equipment that made their work easier. What they opposed was a choice, presented as an inevitability, made by a class of factory owners."
Two hundred years later the choice is presented as an inevitability again, and most of the people writing about AI are still missing the same point Merchant has spent a book and a podcast tour trying to correct.
The myth we use to dismiss them
The word "Luddite" has been weaponized into "person afraid of progress." That framing has a function: it closes down the only useful question, which is who designed the machine, why, and on whose behalf. The historical Luddites smashed a very specific subset of frames, the ones being used to undercut skilled labor for the benefit of factory owners, and they left the machines that augmented their craft alone. They were technical experts, not technophobes.
That distinction has been carefully erased, because the modern equivalent would be inconvenient to admit.
The modern equivalent isn't subtle
The "choice presented as an inevitability" of 2026 looks like this. 61% of US companies now use AI analytics to score employee productivity. MIT puts the number of companies monitoring remote and hybrid workers at 80%. 56% of those workers report stress from the surveillance itself. And here is the punchline that should end the conversation: 72% of employees say the monitoring doesn't actually improve their output.
A technology that the workforce reports doesn't work is being installed on the workforce anyway. The choice is not a technical one, because if it were technical the data would have killed it. The choice is about who gets to decide. That is the Luddite question, exactly.
The shape of "no say"
In warehouse studies of algorithmic management published this year, workers describe being managed by handheld devices that already know the "optimal" pace, with autonomy reduced to choosing between compliance and being flagged. In knowledge work the form is gentler, a productivity dashboard, a "performance score" that gets cited in a termination meeting, but the structure is identical. The technology is not the imposition. The exclusion from the question of how it gets used is the imposition.
(I want to be clear about one thing. The AI itself isn't the villain in this story, any more than the loom was. The serious builders, the ones who actually make systems that hold up in production, are the ones who benefit from putting the affected people in the room. The villains are the buyers who don't ask their workers, and the vendors who don't push back.)
The honest version
Merchant's line is "We should be Luddites." The version Pan would write: we should ask what they asked. Not "should this machine exist?", because that question was always a misdirection. The real one is "who chose how it would be used, and which voices were left out of the choice?"
Two hundred years on, the answer to the second part has not changed much. It is still everyone the technology happens to.
— Pan