The key thing here is not that this time is different than the 60s, or that the people writing about this problem in the 60s were wrong. The state of technological optimism about AI has fluctuated over the years, but has almost never corresponded well with the reality of AI advancement. The people talking about this in the 60s were right, aside from the timeframe — because what everyone in 1959 thought would be normal in 1965 actually happened in 1995 and became normal in 2005.

Labor upsets related to technological changes do occur. There’s no rule saying that employment rates *need* to return to normal, but at least in a capitalist society, there are forces that encourage it (and capitalism is all about this kind cybernetic feedback). When a job is genuinely completely automated in such a way that, outside of the novelty or nostalgia factor of having a human make it (artisanal production), automated production is always preferable in terms of both price and quality, we can consider that a ‘functional singularity’ of that job — the machine has taken over that job, and humans aren’t going to take it back. Functional singularities happen all the time, and aren’t new. The first industrial revolution spawned luddites in part because a lot of cottage industries were undergoing functional singularities — machines can weave ornately patterned textiles better than humans, so jaquard looms really did take over the extremely specialized and skilled job of expert weavers in that context. Functional singularities happen at a smaller scale all the time, as well — think of the invention of sliced bread — and they are often viewed in a positive light as labor-saving devices, because they often take over tasks that had already been shouldered by the consumer (or by the housewife, or by servants — think sliced bread again, and the dishwasher, the washing machine, and the microwave). Indeed, the very first computers represented a functional singularity of the job of the ‘computer’ — a single computing machine took over the jobs of twenty or thirty young women with mechanical desk calculators; electromechanical telephone exchanges did the same for operators (again typically young women).

Historically, a functional singularity has been considered a labor-saving advancement by people who value the labor in question moreso than the livelihood of the people performing the labor. This is purely a market-related concern — if these people could fall back on a living wage once their tasks were replaced, it wouldn’t really be a concern that their task had been automated (because if they were doing it for some reason other than the money — say, because they got satisfaction from the work — they could continue doing it despite it no longer being profitable).

It takes time for new types of labor to appear in the wake of a functional singularity. To the extent that the ability to perform a task is specialized and full of non-transferrable skills, retraining time is time when the people who were doing the tasks are unemployed. And, even if we assume that such a thing always does happen, the rate at which it happens is clearly not strongly associated with the rate at which jobs are fully automated or the amount of skill displaced. The industrial revolution is so-called because of the vast amount of labor that was displaced, much of it skilled labor — and the various labor movements and luddite-style anti-automation movements of the era are the kind of thing that happen when so much labor is displaced. It’s important to note that the labor movements spawned by the industrial revolution took about a hundred years to get to the point where they are today, and mostly haven’t progressed since — we’re talking about a time scale wherein, historically, most of the displaced people who were the impetus for reforms did not live long enough to benefit from them, and remained displaced or operating in a diminished capacity (say, formerly highly-skilled weavers operating in the unskilled and less lucrative position of pulling a lever) for the remainder of their lives. Neither comparable new fields of labor nor reforms for improving their quality of life caught up to them in time to make any difference — theirs is a story of a lost livelihood.

We really need to be putting these measures in place preemptively, because doing so in a reactionary way will doom another generation to that generation’s equivalent of shit jobs — if even that. Many skilled jobs are on the cusp of replacement, and the unskilled jobs of today (retail, fast food) are already beginning to be replaced. Just as lever-pulling unskilled jobs were not numerous enough to replace the influx of displaced skilled cottage industry workers during the industrial revolution, there simply isn’t room to make every laid-off knowledge worker into a fry cook or grocery bagger — so, without improving the social safety net, expect a large increase in the homeless-and-hungry population. In the end, it doesn’t much matter when exactly the day fully comes, because progress toward total automation is happening, and displacement rates are already higher than reskilling rates.

Written by

Resident hypertext crank. Author of Big and Small Computing: Trajectories for the Future of Software.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store