Britain's craft apprenticeship schemes are receiving hundreds of applications for one or two places. The training pipeline that could absorb that interest was dismantled in the 1980s.
Eryn Lawley leans into the foil-block press at Ludlow Bookbinders in Shropshire and listens to the metal settle against the cloth. Foil-blocking is the process of pressing metallic or colored foil onto a book cover, one slow stamp at a time. Lawley, 27, used to work in admin after a politics degree. "Staring at a screen all day was bad for me," she says. The work at the press is repetitive enough to be quiet in a way the office never was.
She is an apprentice at a Shropshire bookbinders in a British trade that has shrunk to fewer than 50 working artisans, by the count of the shop's chief executive, Paul Kidson. That pipeline, he says, "dried up in the 50s, 60s, 70s and collapsed in the early 1980s." No formal apprenticeship pipeline produced a generation of replacements. The trade Kidson now runs, with a partnership from the Folio Society on a limited-edition Game of Thrones release, sits on the far side of that collapse.
Interest, meanwhile, is running the other way. A single National Theatre technical apprenticeship in London drew 728 applications in its most recent round, up from a baseline of about 100. Cox London, a metalwork studio that launched its apprenticeship scheme in 2025, received more than 160 applications for two places in 2026. The trade bodies that still exist cannot, on their current scale, train anywhere near the people asking to join.
The mechanism behind the mismatch is the one the Stanford Digital Economy Lab's August 2026 "Canaries in the Coal Mine" report tries to size. AI is hollowing out entry-level work in the occupations that used to absorb new graduates. Accountants, software engineers, junior lawyers, and entry-level creative workers are feeling it first. The lab's Canaries dashboard tracks the early employment effects across the labor market.
Generative AI now produces entry-level creative work in-house, and AI tools are being deployed into the entry tiers of accounting, software, and law. The bookbinding press is a refuge that exists because the destination jobs are narrowing. Lawley's former trajectory runs from a politics degree through admin to the press. The press is what she reached for after the screens stopped feeling like a career.
There is no national bookbinding training infrastructure, and there are fewer than 50 handcraft bookbinders in Britain. The Folio Society, which has publicly rejected generative AI in its creative processes, is partnering with a craft shop that has no formal way to grow.
The feature that profiled Lawley also captured counter-claims. Sarah Ray-Dobson, 27, a prop-maker at the National Theatre, says her practice has not been affected by AI at her level. Nazifa Tabassum, an apprentice metalworker at Cox London, frames AI as a tool that assists without replacing skilled manual work. Both voices push back against the "AI-proof" framing the original article puts in quotes. The craft migration is real, but so is the gap between trade rhetoric and labor-market arithmetic.
Interest in non-AI manual work is rising faster than the training infrastructure that was dismantled in the early 1980s can absorb. Hundreds of applicants chasing one or two places is a feel-good migration story only if you stop reading at the number of people who want in. The bookbinders who could train the next cohort are, by Kidson's own count, fewer than 50 nationwide. That is the structural answer to whether bookbinding will stay "AI-proof." It cannot grow large enough to be one.