Everyone reaches for the printing press analogy when discussing AI and labor, then immediately apologizes for it — as if the comparison is too obvious, too worn, too undergraduate. This is a mistake. The scribal parallel is not a lazy rhetorical crutch. It is, on careful examination, one of the most structurally precise historical analogies available, and the details most people skip over are exactly the ones that explain what’s happening in the labor market right now.

Let’s start with what actually happened to scribes after Gutenberg.

The common telling compresses everything into a single act of displacement: the printing press appeared, scribes vanished, end of story. But the transition took roughly 60 years from the 1450s to the early 1500s, and within that period the scribal profession didn’t simply die — it bifurcated, then stratified, then partially survived in forms that would have been unrecognizable to a monk in a pre-Gutenberg scriptorium. Some scribes became typesetters. Others pivoted to notarial work, to court documentation, to the kind of handwritten personal correspondence that printing couldn’t serve. A small premium tier moved into manuscript illumination, where the human hand became the explicit luxury signal — expensive precisely because it wasn’t automated.

Sound familiar? It should.

The first year of data on AI and white-collar employment tells a nearly identical bifurcation story. The 2025-2026 Bureau of Labor Statistics figures (along with comparable data from the UK Office for National Statistics and Germany’s Federal Employment Agency) show that the categories experiencing the sharpest employment contractions are not “AI-adjacent” in the abstract. They are specific: entry-level paralegal work, first-draft journalism at wire services, basic financial modeling roles at mid-tier accounting firms, and junior-level UX copywriting. These are precisely the roles that resemble what a medieval scriptorium would have called its journeymen — skilled enough to produce reliable output, not specialized enough to command irreplaceable status.

The master illuminators survived. The journeymen didn’t.

What’s telling about the 2026 data is the occupational list that surprised almost everyone. Radiologists were supposed to be among the first casualties — AI diagnostic imaging has been “18 months from displacing radiologists” since approximately 2016. In practice, radiology employment is flat to slightly up in 2026, because the liability architecture of medical diagnosis requires a credentialed human signature on every read, and because the actual workflow burden of reviewing AI outputs created new coordination roles that absorbed the productivity gains. Translation: the technology arrived, the displacement didn’t, because the institutional scaffolding hadn’t collapsed yet.

Scribal guilds performed an identical function in the 15th century. They didn’t prevent the printing press; they slowed the workforce transition long enough for new institutional forms to emerge. This is not a heroic act of resistance — it’s a structural observation about how labor markets actually process technological shock, which is always slower and messier than the technology itself.

The more interesting occupational surprise in 2026 runs in the other direction. Vocational therapists, home health aides, and licensed electricians are seeing wage growth that exceeds tech-sector averages for the first time in a decade. This is not because AI can’t assist these roles — it can and does. It’s because these roles require physical presence, embodied judgment, and the kind of contextual improvisation that language models handle badly. A therapist reading a patient’s body language in a difficult session. An electrician troubleshooting a 1960s panel box with non-standard wiring in a house that has been renovated four times by four different owners. The AI is useful as a reference tool in both cases. It is not the worker.

The scribal analogy maps to this too. The Gutenberg press handled text reproduction. It did nothing for the wax seal engraver, the parchment tanner, or the bookbinder. Those trades experienced their own disruption later, from different technologies, on different timelines. The mistake is to assume that a single technological event eliminates everything in its vicinity simultaneously. It doesn’t. It eliminates specific functions, reshuffles adjacent ones, and often creates demand spikes in the trades that the new technology depends upon (early printers needed ink, type, presses, and the craftsmen who built and maintained them).

There’s a more uncomfortable part of the analogy that economists have been reluctant to discuss plainly.

The scribal transition was broadly good for society and broadly bad for scribes. Literacy rates eventually rose. Book prices collapsed. Scientific and religious ideas propagated at speeds previously impossible. None of this helped the journeyman copyist in Mainz in 1470 who had spent 12 years developing a skill that was now worth substantially less than it had been. The aggregate welfare gains were real. The distributional pain was also real, and it fell on specific people, in specific places, in ways that aggregate statistics obscured.

The 2026 labor data shows the same distributional signature. Aggregate productivity in knowledge work is up — significantly, in some sectors. The workers experiencing the sharpest wage compression are concentrated in the 25-40 age cohort, in mid-tier cities, in service sectors where the cost of AI tooling is now genuinely low enough that small firms can substitute it for junior headcount. These are not the workers that economic models of “AI complementarity” were designed to describe. They are closer to the Mainz journeyman: skilled, not replaceable yesterday, noticeably less necessary today.

One thing the scribal analogy gets wrong, or at least incomplete, is the speed of institutional adaptation. The 60-year scribal transition happened in a world where communication itself was slow, where knowledge of new techniques spread through physical demonstration and guild networks. The current transition is happening in a world of near-instant information diffusion, which means the displacement is faster but the policy response is also — in theory — faster. In theory. In practice, every retraining program study I’ve read for 2025-2026 shows a consistent pattern: the programs are well-designed, the completion rates are low, the wage premiums for completers are real but modest, and the people who most need them are least able to take 18 months off to attend them.

Gutenberg didn’t offer his displaced scribes a retraining program either, but at least the 15th century had the excuse of not knowing it should.

The scribal parallel has one final dimension that almost nobody mentions, because it requires sitting with an uncomfortable thought. The monks who copied manuscripts weren’t just doing a job. They were, within their own understanding, doing something sacred — preserving and transmitting human knowledge as an act of devotion. The destruction of that role by a machine (which is essentially what Gutenberg’s press was) carried a psychological charge that went beyond economics. The work meant something beyond its market value.

A substantial portion of the knowledge workers now experiencing AI-driven compression would describe their work in similar terms. The junior journalist who believed in the social function of local reporting. The paralegal who understood that attention to documentary detail protects people’s rights. The copywriter who thought language was a craft. Whether or not these self-descriptions are economically defensible, they are psychologically real, and they shape how displacement is experienced and whether retraining is embraced or resisted.

The printing press gave us literacy and the Reformation and the scientific revolution. It also gave us 60 years of disrupted lives in the trades it touched. The honest position on AI and labor is that both halves of that sentence are probably true again, and we are, right now, inside the first half of the 60-year arc.

The scribes who survived — who became typesetters, notaries, illuminators of luxury manuscripts — didn’t do so by fighting the press. They did so by identifying what the press couldn’t do and becoming very good at exactly that. The data suggests this is still the viable strategy. It has just never been a comfortable one.

Get the best articles from Think Different

One email a week with practical Mac workflows, useful comparisons and honest tech recommendations.