The Luddites Never Actually Hated the Machine · Issue #173

Opening

Subscriber, there was an awkward scene at a University of Arizona commencement this past May. Former Google CEO Eric Schmidt stood at the podium, speaking about AI, and the students booed him. Not once — repeatedly. Schmidt eventually said, “I know how you feel. I can hear you.”

Schmidt has a long-running metaphor he likes to use: if someone offers you a seat on a rocket, don’t ask which seat — just get on. It’s the line that made him famous when he gave that advice to Sheryl Sandberg back in 2001. It didn’t land at a 2026 commencement ceremony.

Let me give you the conclusion up front. The weavers of 1811 and the graduates of 2026 aren’t reacting out of hatred for machines. Both groups are asking exactly the same question: who is bringing in this machine, for whose benefit, and under what conditions? The answer has changed across 200 years. The question hasn’t.

What 65% Miss Isn’t Having Less to Do

Let’s start with the numbers — what’s actually happening right now.

In March 2026, the consulting firm Adaptavist surveyed 2,500 knowledge workers across the UK, US, Canada, Germany, and Spain. 65% said they regularly miss the way they worked before AI became widespread.

So far, predictable. But the breakdown tells a different story.

42% said they now spend more time verifying AI output than the time AI actually saves them
52% said colleagues regularly have to fix AI-generated output
31% believe AI erodes human creativity so thoroughly that it would be better to eliminate it entirely
46% feel frustrated that work which once required real expertise can now be done by almost anyone

So this nostalgia isn’t “I wish I had less work.” It’s dissatisfaction with how the nature of the work itself has changed. People have shifted from creators to fixers, from writers to checkers. I think the precise term for this is a verification tax¹ — a levy collected quietly, off nobody’s budget line.

There’s evidence this isn’t just a subjective impression. A study published in Harvard Business Review in February 2026 directly observed a single US tech company with roughly 200 employees over 8 months — visiting the site 2 times a week, tracking internal messaging, and conducting in-depth interviews with about 40 people. The title says it all: AI didn’t reduce the work — it intensified it.

Three specific patterns emerged. First, job scope expanded: PMs started writing code, researchers started doing engineering work, and people began doing tasks in-house that would once have been outsourced. Second, boundaries dissolved: “just one more prompt” before stepping away erased the pauses that used to punctuate a day. Third, the number of simultaneously open tasks increased.

And people at this company felt more productive. At the same time, they reported being as busy as before, or busier. That both things showed up together is the key finding.

There’s a further twist. In the Adaptavist survey, 42% of Gen Z respondents said they preferred the pre-AI era, compared to 26% of Gen X. It’s the digital natives who miss it more. That figure makes it hard to sustain the interpretation that this is just “an older generation failing to adapt.”

In 1811, They Didn’t Smash Just Any Machine

Now let’s go back 200 years.

Today, “Luddite”² is shorthand for a technophobe. When a newspaper calls something “a Luddite idea,” it means backward resistance to progress. But the actual history is quite different.

The movement that began in Nottingham, England in 1811 wasn’t made up of ignorant mobs. By the standards of the time, these were highly skilled craftsmen — people who had gone through long apprenticeships to master their trade, who supported families and held standing in their communities through that skill.

What they smashed were knitting frames. But here’s the important part: they didn’t smash just any machine. They left alone the machines of employers who paid fair wages and produced goods of proper quality. Their targets were specific factory owners who used machines to break existing wage agreements, to churn out low-quality mass production, and to replace skilled workers with cheap, unskilled labor.

In other words, what they opposed wasn’t technology — it was the terms under which technology was introduced. That’s the central point Brian Merchant makes in Blood in the Machine. The Luddites didn’t hate machines; they asked whose pocket the profits the machines generated would end up in.

The British government’s answer to that question tells you everything you need to know. In 1812, Parliament made machine-breaking a capital offense. They chose the gallows over negotiation. The poet Byron, in his maiden speech in the House of Lords, opposed the bill, arguing these were people who had lost their livelihoods, not criminals. It made no difference.

The image we’ve inherited — “Luddite” as a synonym for technophobia — is, in other words, the summary written up by the winning side after the argument was already over.

Where Two Uprisings, 200 Years Apart, Overlap Exactly

Now let’s lay the two pictures on top of each other. I think they overlap in precisely three places.

First, the ones resisting aren’t the unskilled — they’re the skilled. The weavers of 1811 were the professional class of their era. The group most uncomfortable with AI adoption in 2026 is the same kind of group. In the Adaptavist survey, 33% of general knowledge workers said they’ve considered changing industries — but among C-suite executives, that figure was 46%. This anxiety isn’t spreading from the bottom up. It’s actually concentrated more heavily at the top.

Second, technology doesn’t reduce labor — it redistributes it. The knitting frame didn’t eliminate labor. It converted skilled labor into unskilled labor and shifted the difference in value to the factory owner. Today’s AI isn’t eliminating work either. It’s converting the labor of creating into the labor of verifying. The problem is that this verification labor doesn’t show up on anyone’s performance metrics. Individuals absorb it quietly.

Third, backlash erupts at the exact moment control shifts from the individual to the organization. This part is especially interesting right now. Until quite recently, companies were encouraging employees to freely experiment with AI. But once AI companies shifted to token-based billing³ and costs became visible, those same companies started tightening usage instead. From the employee’s perspective, they received the exact opposite instruction within half a year — from “use your own judgment” to “don’t use it.”

The technology didn’t change. What changed is who holds the decision-making power. That’s exactly the point the Luddites were angry about too.

There’s one more thing to add here: the creativity problem. This isn’t sentiment — it’s a measured result.

In a 2024 study published in the journal Science Advances, 300 writers were given AI-generated ideas, and 600 evaluators scored the resulting output. Stories from the group that used AI most heavily scored 8.1% higher on novelty and 9% higher on usefulness. Writers who started out less creative saw novelty rise as much as 10.7% and usefulness as much as 11.5%. A clear gain for the individual.

But in that same data, similarity among writers rose by 10.7%. Individuals got better. The group got more alike.

A brainstorming experiment by Wharton researchers, published in Nature Human Behaviour in 2025, is even more dramatic. Across 45 statistical comparisons, idea diversity was significantly lower in the AI-assisted group in 37 of them. In a task asking participants to design a toy from bricks and a fan, 94% of ChatGPT users produced overlapping concepts, and 9 different people, working independently, each named their toy “Build-a-Breeze Castle.” In the human-only group, no such overlap occurred.

So the instinctive complaint that “AI harms creativity” — at least at the collective level — turns out to have real evidence behind it.

But Nostalgia Has a Trap Built Into It

If I stopped here, this piece would only be half true. Nostalgia, as an emotion, is inherently in the business of prettifying the past.

Was the pre-AI content industry really that diverse? Big studios and publishers had been repeating proven formulas long before AI ever showed up — sequels to sequels, remakes of remakes. There was never a golden age when everything created was an experimental masterpiece.

The same applies to concerns about younger workers’ learning. It’s a fair worry that if AI does all the foundational work, what will junior employees actually learn? But was the alternative we’re implicitly nostalgic for — the apprenticeship model where law firms and banks had new hires repeat the same rote task a thousand times — really the best way to learn? Was it kept around because it was the only proven method, or because it was cheap and customary?

If we don’t answer that question honestly, what we recover isn’t what we lost — it’s just what we’re used to. Those are two different things.

The Luddites weren’t exempt from this either. The apprenticeship system they were defending was itself high-barrier and closed off. History sides with them not because that system was perfect, but because the cost of change was billed exclusively to them, and that was unjust.

Oz’s Lens

I see this not as a failure of adoption, but as a failure of design.

There’s a scene I’ve watched repeat itself over and over while building GTM strategy. When an organization brings in a new tool, it always runs 2 calculations: how much money it saves, and how much faster things get. Few organizations ever ask a third question: who does the new labor this tool creates?

I’ve almost never seen an AI adoption plan where the word “verification” appears next to an actual person’s name. The time saved goes up on the company slide deck. The new verification labor gets absorbed into someone’s evening. When 42% say they spend more time checking than they save, it’s not because the tool is bad — it’s because that labor has no owner.

That’s why I think what’s needed right now isn’t AI training, but AI work redesign. Teaching people to write good prompts is easy, and it shows. But writing down, in a document, “who is ultimately responsible for this team’s AI output, and how many hours per week are allocated to verifying it” is hard, and it doesn’t show. The hard one is the real work.

And if this diagnosis is right, then the graduates who booed weren’t anti-technology. Before getting on the rocket, they asked whether the seat had a seatbelt. I think that’s an entirely reasonable question.

Closing

Let me sum up. The 65% who miss the pre-AI era aren’t missing a time when there was less to do. They’re reacting to a situation where the work of creating became the work of checking, and nobody has acknowledged that shift. It’s the same structure as the Luddites resisting not the machine itself, but the terms of its introduction. So the question we should be asking now isn’t “should we use AI or not” — it’s “whose ledger does the new labor AI creates get written into?”

Here’s one thing I’d suggest trying this week: count how many minutes you actually spend reviewing AI-generated output. If that number is bigger than the time you believe you’re saving, that’s not a tool problem — it’s a design problem.

Have you had a moment recently, using AI, where you thought, “this wasn’t originally my job”? Tell me in the comments which task that boundary quietly crept across. I’ll gather readers’ examples of where verification labor is actually piling up the most, and map it out in a future issue.

💬 Tell me in the comments about a moment when you thought, “this wasn’t originally my job.” I’ll bring it into a future issue. 📨 If you know a colleague who’s worn out from cleaning up after AI output lately, send them this piece.

References & Further Reading

Primary sources

The Adaptavist Group, Understanding the Human Cost of AI Transformation, March 2026. Link ··· The backbone of today’s piece. Surveyed 2,500 knowledge workers across the UK, US, Canada, Germany, and Spain — the generational cross-tabs are especially worth a look.
Aruna Ranganathan & Xingqi Maggie Ye, “AI Doesn’t Reduce Work, It Intensifies It,” Harvard Business Review, February 2026. Link ··· Its strength is eight months of field observation rather than survey data — though as a single-company case study, generalize with caution.
Anil R. Doshi & Oliver P. Hauser, “Generative AI enhances individual creativity but reduces the collective diversity of novel content,” Science Advances, 2024. Link ··· Quantifies how individual creativity and collective diversity move in opposite directions. I’d recommend starting with the similarity analysis in Section 3.
Lennart Meincke, Gideon Nave & Christian Terwiesch, Nature Human Behaviour, 2025. Link ··· The experiment where nine people independently named their toy the same thing. Verifies the drop in diversity across 45 comparisons.
“Former Google CEO Eric Schmidt booed during graduation speech about AI,” NBC News, May 2026. Link ··· The source for the opening scene. Includes the full remarks from the event.

Background

Brian Merchant, Blood in the Machine: The Origins of the Rebellion Against Big Tech, Little Brown, 2023. Link ··· Recovers the Luddites as a labor movement rather than a case of technophobia. The historical section of today’s piece leans on this book.
Kwangseob Ahn, People Who Outsource Their Thinking: Homo Brainless, Jpub, 2025. Link ··· Covers verification labor and the delegation of judgment at greater length.

A past issue worth reading alongside this one

OZ Talking, “The People Who Switch Off Their 20-Watt Brain”

📝 Glossary

The author, Kwangseob Ahn, is a professor of business administration at Sejong University and lead consultant at OBF (Oswarld Boutique Consulting Firm). He teaches statistics and data analysis, including business data management and business analytics, while leading GTM and AI strategy consulting in the field, designing the seam between technology and business. He has published academic research on a memory architecture for AI dialogue systems (HEMA) and runs Daily Arxiv, a daily curation of global AI papers. He holds a master's from Korea University's Graduate School of Technology Management and a KMBA. He is the author of Homo Brainless: The People Who Outsource Their Thinking.

Footnotes

Verification tax: The time and effort spent confirming and fixing AI-generated output. Not an official term — it’s used here to describe a cost that quietly eats into the time supposedly saved. ↩
Luddite: A movement of skilled English weavers who destroyed knitting frames between 1811 and 1816, named after signatures attributed to a fictional figure called “Ned Ludd.” Today the word means technophobe, but it originally described a fight over wages and employment conditions. ↩
Token-based billing: A pricing model where AI services charge based on the volume of text processed rather than the number of users. Because costs rise with usage, this gives companies an incentive to manage how much employees use AI. ↩