What did the last machine age teach us about this one?
Factories did not just automate muscle. They rewrote time, skill, and status. The pattern is familiar — if we bother to look.
Every generation thinks its technology is unprecedented. Historians are paid to be annoying about that.
When mechanisation hit textiles, the first story was replacement. The longer story was reorganisation: new grades of skill, new supervision, new towns, new fights over who captured the surplus. Deskilling and upskilling arrived as a pair — not as alternatives.
Three recurring lessons
1. The bottleneck moves. Automate spinning and weaving becomes the constraint. Automate drafting and review becomes the constraint. Watch where queues form after the tool ships.
2. Status lags productivity. Work that feels “less skilled” often becomes more valuable because it coordinates the machine. Clerks, dispatchers, and today’s eval leads share a genealogy.
3. Institutions decide outcomes. The same loom produced different lives under different rules. The same model will too. Training budgets, bargaining power, and product design are not side notes.
A short, sharp provocation
If your AI strategy has no theory of apprenticeship, you are not transforming work. You are liquidating the next generation’s ability to judge the machine.
Past leaps did not ask workers to “learn prompt engineering”. They asked societies to rebuild pathways into competence. That is still the job.