How a violent 1800s uprising against machines could foreshadow our future with AI

• AI could upend white-collar work, making some people better at their jobs and hindering others.
• I talked to the author of "Blood in the Machine: The Origins of the Rebellion Against Big Tech."
• He shared some lessons from studying the Luddite clothworkers' uprising against machines.

AI may be a great leveler for the workforce, making people who aren't very good at their jobs suddenly become a lot more productive. A crummy programmer can use GitHub's Copilot to write proficient code, while a subpar writer could create perfectly fine marketing copy or seem eloquent over email.

But, as Business Insider's Aki Ito reports, this isn't all upside. If suddenly it's easy for anyone to make decent code, it becomes hard to justify paying huge salaries to highly talented coders, or paying big bucks to a marketing copywriter with more education and experience.

This, of course, is not the first time new technology has radically transformed our lives and careers. The Industrial Revolution of the 19th century radically changed human labor, with a lot of good things (steam trains!) coming at the cost of bad things (child labor!).

Brian Merchant is the author of "Blood in the Machine: The Origins of the Rebellion Against Big Tech," which tells the history of the Luddite rebellion of the early 1800s, when clothworkers in rural England banded together and smashed the new weaving machines that factory owners were introducing.

I asked Merchant for his perspective on how history informs how we should think about AI and labor.

My colleague Aki Ito wrote about how AI will ultimately harm the most talented people in white collar jobs, because it will make it harder for them justify sky-high wages.

Is that sort of what happened  to the clothworkers in the early 1800s when machines were introduced (taking into account that when we're talking about harm in coding jobs, people aren't exactly going to lose an arm or get maimed)?

I like what Aki is getting at here, because I think that too often, economists who look at the macro picture miss just how brutal this process can be — and what stands to be lost to it. 

Take the clothworkers: skilled weavers made really high quality cloth and framework knitters made stockings that were nice and durable. The stuff would last, and the workers took pride in making it. When factory owners started using machines that could emulate on some level that production, they still needed human operators. But the work could now be done by children who could be paid a lot less, if anything at all.

Just about everyone loses except the factory owners, who profit from churning out more shoddily made stuff at a rate that the skilled workers can't compete with. 

Now, a few of the highest skilled weavers and clothmakers were just fine — there was still a market for the finest artisanal goods, and machine-made stuff couldn't touch any of that. I expect this is a similar dynamic we'll see play out with AI and a lot of jobs involving things like writing, coding, illustration, and graphic design.

In the short term, yes, you might see more productivity because people who weren't as quick at writing marketing emails or web copy or power point slide illustrations can use generative AI services to whip that stuff up. And the highest-skilled, most in-demand, and best known writers, artists and coders will continue to do just fine.

Based on your understanding of history, what are the chances that AI helps people by lowering the gap between lower skilled and higher skilled workers?

Now, as then, the vast, vast majority of the time, AI absolutely cannot replicate a good writer or worker's output. It can produce output that may be passable to managers looking to tell their managers that they saved on labor costs this quarter by firing Ian and Deborah in marketing or whatever, and teaching an intern how to use ChatGPT, which, by the way, they just purchased enterprise tier for.

But that's the end game, and what needs to be understood as the beating heart of all this AI talk; it's ultimately being sold to every firm that adopts it as a labor-saving and cost-cutting tool, period. 

Something that people say a lot about AI is that it might get rid of some jobs, but make new other ones; a net even. That wasn't exactly true for the clothworkers of the 1800s right? Like, jobs didn't change, they got actively worse (more dangerous, cheaper, harder to make a living etc).

And you're exactly right — the Industrial Revolution didn't cause a net job loss. But when people say this, especially about the Industrial Revolution — that some old jobs were lost but new ones were gained — it always blows my mind a bit. Because the jobs it "created" are literally some of the worst jobs in human history. Child laborers toiling in windowless factories inhaling cloth fibers and getting their fingers smashed in churning machinery? Not great!

I expect the net impact of AI, if it continues to be deployed in the workplace in the contexts we've seen it deployed so far — by managers and executives, with little to no input from workers themselves — then the result will be the erosion of the quality of people's jobs and an accelerating inequality as all those productivity gains get funneled to the top.