The AI Shift: What can Victorian bootmakers tell us about AI disruption?

金融时报（科技），金融时报科技版块 - 来自金融时报的最新科技新闻与分析。关注

This article is an on-site version of our The AI Shift newsletter. Premium subscribers can sign up to get the newsletter delivered every Thursday. Standard subscribers can upgrade to Premium here, or explore all FT newsletters

Welcome back to The AI Shift, our weekly newsletter about technology and the labour market. John is on holiday this week, so I’m joined again by my excellent colleague Joel Suss, a former Bank of England economist. We thought we’d do something a little different today: a deep-dive into a fascinating story of labour-saving machines from 19th-century England. Why? Because while people often draw comparisons between the AI age and the upheavals of the Industrial Revolution, this particular tale is a reminder that technological change doesn’t always play out in the way you might think.

Sarah writes

Bootmaking was the single largest craft industry in 19th-century Britain, employing nearly a quarter of a million people. So when a bootmaking sewing machine appeared in the 1850s which enabled one person to do the same amount of work as four people without the machine, people were understandably worried about mass displacement. Indeed, when a manufacturer in Northampton introduced one of the first machines in 1857, he reported that workers went on strike: “I was followed to and from my home by several hundreds of people daily, which swore they would drive me and my machine out of town together.”

But the sewing machines took off, and the industry transformed. Widespread unemployment wasn’t the outcome, though. By 1911, the English bootmaking industry employed roughly the same number of people as it had prior to mechanisation in 1851.

Why? According to a careful study of full count UK censuses by Hillary Vipond, a postdoctoral researcher affiliated with the Complexity Science Hub in Vienna and the London School of Economics, 153,000 artisanal bootmaking jobs did disappear over this period. At the same time, 140,000 new specialised jobs emerged: sewing machinists, riveters, operators, factory foremen and managers.

But the artisans didn’t retrain as machine-enabled bootmakers. Instead, the adjustment happened gradually over a generation. Young people stopped entering the artisanal bootmaking profession, while the older artisans for the most part stayed in their occupations and gradually died out (sadly, the census records don’t tell us what happened to their incomes). The younger generation, meanwhile, were also the ones who entered the new types of jobs.

As Vipond writes: “In the Victorian bootmaking industry, adjustment on both the destruction and creation sides of the ledger pivoted on young workers. There was a collapse of entry into artisanal bootmaking, as young people stopped entering. Simultaneously, young people took up the new and more specialized jobs . . . Some doors closed and some doors opened, and they did so for the same cohort.”

Joel, this is a somewhat gentler account of technological change than the better-known tale of the Luddite weavers. Do you think it can help to inform how we think about AI and automation today?

Joel writes

Thanks Sarah, yes, there are a surprising number of lessons from the experiences of English bootmakers for today’s AI-induced transformation.

The first is on considering all margins of possible displacement. The conversation around AI tends to focus on possible ‘exit’ from existing jobs; people are afraid of being made redundant and widespread technological unemployment. But entry — the career choices and prospects of the young — may turn out to be the dominant margin of transformation, as it was with bootmakers during the second industrial revolution.

There are tentative signs that this is happening in the AI age — studies show little sign that AI is displacing existing jobs. But there looks to be a clear shift in what students are now studying, with moves away from degrees leading to ‘AI-exposed’ careers, such as computer programming. Anticipation of the new economy is driving different patterns of entry, in other words.

Of course, new job creation is necessary for the young setting off on careers, as it was for the young bootmakers entering the industry back then. What will be the new jobs of today? It is too early to say (and if we knew we might be in the wrong line of business), but our colleague Soumaya Keynes explored some interesting new research on job creation under technological change. One conclusion that chimes with Vipond’s bootmaking study is that occupations persist — using Swedish data the researchers found that 70 per cent of employment today is in occupations that already existed 140 years earlier.

Another lesson from the bootmakers is that Jevons’ paradox is real and powerful. Coined by the English economist William Stanley Jevons during the period in which bootmaking was being transformed, the paradox describes how new, labour-saving or resource-efficient technologies expand demand. Jevons cited coal specifically, the usage of which skyrocketed after a more efficient means of burning the fuel was introduced, but this also applied to boots, which were made much cheaper thanks to the machines, enlarging the pool of people who could afford to buy them. And now, with labour-cost savings due to LLMs, the services of ‘AI-exposed’ professionals such as lawyers and consultants may see a similar dynamic. Even ghostwriters are seeing a surge in demand.

Demand for ‘artisanal’ services or products may also hold up in other ways. In her study, Vipond shows how “bespoke demand” for artisan bootmaking persisted, especially in wealthier English parishes. The old way of doing things is still doing business today in traditional cordwaining centres like Northamptonshire and Florence, aimed at the already well-heeled.

Many point to the speed of diffusion as a key concern with how AI is developing. But it was not as if new bootmaking technology diffused slowly. Given the economic logic of far higher output at lower cost, opposition from labour to impede adoption proved futile. A letter by the self-proclaimed ‘introducer’ of the boot sewing machine to England in the Northampton Mercury attests that by 1865 there were already “upwards of 1500 [machines] in the town” despite a strike lasting over a year.

Today, while AI chatbots have seen rapid usage after the release of ChatGPT in 2022, companies have been slow to put LLMs to use in meaningful business activities. This is due primarily to organisational barriers, as noted in this newsletter on adoption. If you believe the doomsters, it may be more likely that AI brings about the demise of the human species before widespread business adoption. Instead, ‘native AI’ companies might slowly replace incumbents.

So far, so reassuring when it comes to the prospects of widespread technological unemployment. But there are some more difficult lessons to take from Victorian-era technological change. While from a macro perspective everything turned out fine for English bootmaking — the numbers of jobs lost were replaced by new jobs in the industry — the adjustment was plainly painful for many.

Vipond documents how the new bootmaking industry became highly geographically concentrated (nearly half of the new jobs were generated in Leicestershire and Northamptonshire alone), a big change from the dispersed nature of the artisan trade. Some places lost out. What are the spatial implications of the AI transformation? The technology is aimed especially at knowledge workers, who have concentrated in ‘superstar cities’ over recent years. Could AI cause even further concentration, or a dispersal of knowledge work? This seems like a subject we should return to.

The geographical concentration of bootmakers also went side by side with vertical scaling from cottage industry and small workshops — larger companies were made possible, altering the balance between capital and labour income.

Mechanisation was also fundamentally ‘deskilling’, Vipond points out. The new tasks in the bootmaking industry did not require the same level of training that apprentices required to learn the old craft. Many modern studies show that the use of AI can lead to ‘cognitive offloading’ and atrophied critical thinking. On the other hand, for some tasks like software engineering, AI is sort of upskilling for those already possessing skills. And unlike artisanal bootmakers, senior developers seem to be enthusiastically embracing new technology.

There is another kind of transformation that it is worth considering when it comes to new technology. The census records, while rich and detailed, are less clear on how the quality or meaningfulness of bootmaking work changed. How did going from artisans working with their hands as ‘cloggers’ or ‘cordwainers’ to the ‘clerks’ and ‘operators’ of the new bootmaking industry affect the lived experience of workers? Perhaps something like going from a poet to a prompt engineer?

The AI Shift: What can Victorian bootmakers tell us about AI disruption?

Sarah writes

Joel writes

Recommended reading

Recommended newsletters for you