The artificially intelligent “agents” promised by tech titans like Sam Altman seem destined to replace workers, but they also sound incredibly useful. It sounds like a hi-tech dilemma, but it’s one that harkens back to the earliest days of job automation.
Like most technological leaps, cutting-edge generative AI applications like OpenAI’s Sora fascinate everyday people just as they do company owners and managers. Their outputs can be eerie and flawed, but they’re also undeniably captivating. Similarly, new AI-based inventions like the Rabbit R1 rely on sophisticated software built on generative AI to perform tasks for us. These concepts (and that’s really all they are for now) can have a dazzling, gee-whiz effect on us, and while some may remain unimpressed, it’s only natural for some individuals to want to apply these task automation tools to their difficult lives.
AI is not new in this regard. Not every device capable of automating a task extracts a human being from the labor pool, and replaces them with a chunk of metal.
For instance, it was surely part of a farmer’s workday to regulate the temperature of Dutch chicken incubators in the 1600s, so when Cornelis Drebbel invented the first thermostat in 1620 (over a century before the Industrial Revolution) it replaced that job function with an “intelligent” agent. A thermostat possesses a very limited form of intelligence — it “knows” a certain temperature is reached, shuts off a heater, and then turns it on again when necessary. A farming duty was certainly replaced when a thermostat was installed, but not necessarily a “job.”
This incubator example points to one reality that’s obscured when we talk about AI “replacing” humans. Sometimes workers want some other entity to step in and do part of their work. What they don’t want is for the people with economic power over their lives to chuck them in the proverbial bin.
If you imagine a slider with “workers receive handy new tools” on one side, and “total job apocalypse” on the other, automation’s real effect on labor doesn’t look like one of the two extremes, but — as has been documented by the economists Daron Acemoglu and Simon Johnson — the overall effect of automation technology throughout history isn’t pretty for workers, and small fish in general, no matter what the techno-optimists like to tell themselves. Machines are morally neutral, but their rollouts tend to squeeze the many for the benefit of the few.
Without a doubt the most illuminating period in history we can study to understand this ambiguity is the Industrial Revolution. Brian Merchant is a tech journalist and author of books that wrestle with deceptive narratives in tech history. His latest, Blood in the Machine: The Origins of the Rebellion Against Big Tech is a look at the industrial revolution through a contemporary frame. It’s about the circumstances that led to a rebellion in England around 1811, in which former textile workers known as the Luddites sabotaged some of the machines that were being used to degrade their way of life.
It’s not a simple story about factory workers being replaced with machines; for one thing, this was the time when the modern factory came to be invented in the first place.
Mashable talked to Merchant about the popular narrative of worker replacement by AI, and how to better understand what working people are up against.
Mashable: Obviously things in 19th century England were different than where we are now, but also the same. Set the scene a bit.
So, some important context here is that cloth production was the largest industry in England — there were hundreds of thousands of weavers, knitters, cloth finishers, and so on. The largest industrial worker base in the nation.
They had made the country rich! And these cloth workers had resisted exploitative uses of automation technology for decades, primarily by using informal trade groups to push back (actual unions were illegal), and by calling on Parliament to protect their trade with petitions and letters.
They argued, quite correctly, that entrepreneurs were using machinery as a way to get around long-held trade rules and regulations.
Think of Uber arguing it’s not a cab company, but a tech company, so it doesn’t have to play by the laws that govern local taxi companies. The workers were ignored by the government, as workers mostly are today.
[S]ince running the machines required less skill to operate, entrepreneurs could justify lowering wages, or employing child labor. So pay went down, the work became unpleasant and subordinated, and hand-spinning grew rarer.
Was there a big precipitating event?
Around 1810, there was a big economic depression. And when things got tight, it became very clear to the cloth workers who was using machinery to undercut their wages and take their work. And yeah, they got mad.
After peacefully pushing the government to enforce trade rules and to enact a safety net, after politely asking the entrepreneurs to consider their families and so on, and then holding mass protests, they were out of good options.
That’s when they truly rose up and became the Luddites — when they ran out of ways to impress upon those in power that they were human beings deserving of things like dignity and enough income to feed their families.
Let’s talk about the Spinning Jenny. How does it relate to the present day “worker replacement” issue?
The Spinning Jenny is an interesting case, because it automated the spinning of yarn, a task that was done largely by groups of women, by hand, before the Industrial Revolution kicked into high gear.
The Jenny didn’t necessarily replace those hand spinners, but it did mean more yarn could be produced with less labor, and it meant that the yarn would often need to be spun in the confines of a workshop or a factory, under the direction of the man who owned the machinery — who would, as it happens, take a large cut of the profit for himself.
So spell out for me what the Spinning Jenny did, if not simply “replace” workers.
It transformed the nature of work, who profited from it, and how much people made by doing it. Spinning, like weaving, used to be tightly linked to the home, where it was comparatively nice to work, and offered freedom and flexibility. You were working with your family! You could sing, or take walks in the garden if you felt like it.
Not so much in a factory. The machine could produce many times more yarn than a human—though, notably, its quality was a lot worse, especially at first — but since demand was also skyrocketing, you still needed a lot of people to run them.
But since running the machines required less skill to operate, entrepreneurs could justify lowering wages, or employing child labor. So pay went down, the work became unpleasant and subordinated, and hand-spinning grew rarer.
Wouldn’t it be better to try to do what the Luddites really wanted, which was to give working people a real say in how their futures were constructed, in how technology was deployed in society—and to receive a share of the economic benefits from said technology?
So, less pay, less enjoyment, less pride in one’s craft. Is that what you expect from AI?
Yes. Especially where production of a higher quantity of goods or services is important. ….
In a lot of contexts, if ChatGPT does spread and is adopted by entrepreneurs and executives en masse, it could squeeze and transform vocations, give employers more control over the work process, and, again, drive down wages.
Even if the chatbot doesn’t produce better output, it offers management an opportunity to wrest more control over the process, and to subordinate the worker, who’s reduced to checking larger quantities of text output for errors, or feeding a narrow set of terms into the machine. Less a matter of job replacement, and more of job degradation.
We’re myth-busting, so: Most people think of the Luddites as provincial technophobes who compulsively smashed machines, as if they just were Neanderthals standing in the way of progress because they were ignorant. And that’s basically how we use the term now. What were they really doing?
The Luddites smashed the machinery that entrepreneurs and factory owners were using to strip them of dignity and economic security — and only that machinery. They left the rest intact. It was a highly organized and strategic operation, and their demands were clear and precisely targeted.
They were machinists themselves, and had no hatred of machinery in general.
I know you don’t have a crystal ball, but is AI about to bring about another Luddite uprising?
I think the idea that the rise of AI companies alone will lead to conditions that are so bleak that we have a reprisal of the Luddite uprising — in the sense of an organized campaign of mass sabotage against employers’ tech anyway — is pretty unlikely, at least for the foreseeable future.
We have a relatively high employment rate now, we have more tools at our disposal to agitate for change than the Luddites did, and the problem is less that there are no jobs, but that so many of the available jobs are not good ones.
That said, there have been outbursts of classic Luddism over the last ten years — for instance, when Uber moved into European markets, the company was met with outright hostility from cab drivers. In France, taxi workers smashed Uber cars that were cutting into their turf. In San Francisco, activists threw rocks at Google buses.
In that case what should be the takeaway for The Powers That Be as we enter the age of AI?
Wouldn’t it be better to try to do what the Luddites really wanted, which was to give working people a real say in how their futures were constructed, in how technology was deployed in society—and to receive a share of the economic benefits from said technology?
Merchant’s book, Blood in the Machine: The Origins of the Rebellion Against Big Tech is now available from Little, Brown and Company.
Topics
Artificial Intelligence
