If Being Concerned About AI-Driven Joblessness Is Extreme, I’m An Extremist

Dissecting the discourse about the likely impacts of automation, again

Image for post
Image for post
source: Moltec

The other day I encountered a post by Sam Lessin at The Information, entitled The Future of Work, which caught my attention since I think of the future of work as my beat.

Here’s his opening:

We are entering a period where machine learning is going to dramatically change the nature of work across a broad swath of service and knowledge industries.

Extremists see a world where all human work evaporates rapidly, driving a conversation about human dignity and equality in a world of technological unemployment. A more sober group believes we’re on the brink of eliminating a class of repetitive white-collar knowledge work — not unlike transitions we have gone through in the past.

My view is that this debate is several steps ahead of reality. Those that concentrate just on the elimination of some specific job categories underestimate the impact of new technology.

Oh, boy. Here’s the first paragraphs of comments I left there¹:

My hackles raised when you wrote ‘Extremists see a world where all human work evaporates rapidly, driving a conversation about human dignity and equality in a world of technological unemployment. A more sober group believes we’re on the brink of eliminating a class of repetitive white-collar knowledge work — not unlike transitions we have gone through in the past.’

You seem to be simply repeating the techno-utopian conventional wisdom, which you consider as ‘more sober’ in contrast to those who disagree — those who are deeply concerned about the potential for AI to lead to joblessness — those you castigate as ‘extremists’.

Notably, your piece is relatively free of facts or citations of others research that might support your opinions. As Daniel Moynihan observed, everyone’s entitled to their own opinion, but not their own facts. You’ve penned yet another fact-lite post along the lines of so many others published in silicon valley techland recently.

My recommendation to your subscribers is that they look to the news of what’s happening on the front lines instead of softsoap op-ed like this: the hedge funds that are laying off quants as their deploy AI as fund managers and firing human financial advisors and roll out AI in their place, the companies that are increasingly reliant on AI for hiring, and the coming deluge of basic innovations like hamburger-cooking robots, AI radiologists, and driverless taxis.

Lessin, a former Facebook guy, is deeply embedded in a start-up/entrepreneurial community that pooh-poohs AI as a job destroyer, a community that often makes the following argument:

We’ve had technology revolutions before where some jobs — horsewhip making and newspaper typesetters, for example — are eliminated, but the workers shifted to other, new sorts of jobs created by the new technology. AI is just another cycle of that sort of disruption.

The problem is that this jumps to a conclusion, avoiding the possibility that AI is different from earlier technological revolutions, like agriculture and industrialization, since those supplanted human physical labor. AI, on the other hand, is supplanting cognitive skills, which is a completely different game.

‘Aha!’, the techno-optimists like Lessin might respond, ‘if that is the case, how come unemployment isn’t higher? Where are the unemployed millions?’

Well, there’s a complex answer. In the US, at least, a very large proportion of formerly employed industrial workers have been sidelined by a shifting economy. As the Economic Policy Institute points out,

In today’s labor market, the unemployment rate drastically understates the weakness of job opportunities. This is due to the existence of a large pool of “missing workers”–potential workers who, because of weak job opportunities, are neither employed nor actively seeking a job. In other words, these are people who would be either working or looking for work if job opportunities were significantly stronger. Because jobless workers are only counted as unemployed if they are actively seeking work, these “missing workers” are not reflected in the unemployment rate.

The US labor participation rate is around 63% meaning 80 million or more possible workers are currently not seeking work, presumably because a/ wages are not high enough to motivate them, or b/ they are actually unemployable. (The latter category includes millions who would fail drug tests if they applied for work.) But still, millions are on the sidelines. Where are the jobs?

In manufacturing, we have been on a slide since the late ‘70s:

The manufacturing jobs delusion | The Economist

In 1979, the high point for American manufacturing jobs was reached at 19.5m. The subsequent recession of the early 1980s caused that number to fall but there were regularly 17m-18m jobs in the 1980s and 1990s. From the turn of the millennium, however, the total fell pretty remorselessly, with the 2008–09 recession proving the coup de grace. The low was just under 11.5m in early 2010. As the economy recovered, some jobs returned and a peak of 12.3m was reached early last year. But since then, the numbers have been drifting down again.

In 1979, the high point for American manufacturing jobs was reached at 19.5m. The subsequent recession of the early 1980s caused that number to fall but there were regularly 17m-18m jobs in the 1980s and 1990s. From the turn of the millennium, however, the total fell pretty remorselessly, with the 2008–09 recession proving the coup de grace. The low was just under 11.5m in early 2010. As the economy recovered, some jobs returned and a peak of 12.3m was reached early last year. But since then, the numbers have been drifting down again.

And one of the factors in manufacturing job decline is automation. Yes, it is the case that adding robots to a manufacturing plant does not always translate to layoffs there. However, something more complex happens. Let’s look at Germany, where significant increases in robots hasn’t led to a decrease in employment. But what does happen?

Jill Petzinger | Germany has way more industrial robots than the US, but they haven’t caused job losses²

While industrial robots haven’t reduced the total number of jobs in the German economy, the study found that on average one robot replaces two manufacturing jobs. Between 1994 and 2014, roughly 275,000 full-time manufacturing jobs were not created because of robots.

“It’s not like jobs were destroyed, in the sense that a manufacturing robot is installed and then the workers are fired because of the robots — that never really happened in Germany,” study co-author Jens Südekum, from Düsseldorf Institute for Competition Economics, told Quartz. “What happened instead is that in industries where they had more robots, they just created fewer jobs for entrants.”

“Typically around 25% of young labor market workers went into manufacturing and the rest did something else, and now more workers have immediately started in the service sector — so in a sense the robots blocked the entry into manufacturing jobs.”

So, aside from the extreme near-term, deploying more robots into manufacturing — and by extension, AI into other industries — may not translate into immediate layoffs. However, even in those cases the cumulative effect will be to divert workers out of those industries where robots and AI are deeply integrated, since exponential scale of production is provided by AI and robots, with at most a linear growth of human workers.

And others argue fairly conclusively that robots do lead to job losses:

George Dvorsky | Robots Are Already Replacing Human Workers at an Alarming Rate

Each additional robot in the US economy reduces employment by 5.6 workers, and every robot that is added to the workforce per 1,000 human workers causes wages to drop by as much as 0.25 to 0.5 percent. Such are the conclusions reached by MIT’s Daron Acemoglu and Boston University’s Pascual Restrepo, who published their findings at the National Bureau of Economic Research.

Going back to the observations from the Economist, we have a large number of prospective workers — a great deal of male blue collar workers who formerly worked in the merry-go-round of industrial trades, mining, manufacturing, construction, and so on — who have been displaced, and can’t easily shift to a services economy.

These are only a few data points, but there are dozens more, like this post about AI replacing fund managers at BlackRock, FoxConn’s plans to replace nearly all workers with robots, and the predictions of the devastating impact of driverless vehicles on the millions of truckers, delivery and taxi drivers.

So, as the title says, if being concerned about AI-driven joblessness is extreme, I’m an extremist. Call me what you want. But please offer up some research, some data, some economic analysis to support your claims, not just the rewarmed conventional wisdom of those deep in the Silicon Valley bubble, a community that identifies with the technologists and investors behind the greatest economic disruption of our time, and not those who are always the first to bear the blunt force trauma involved.

Written by

Founder, Work Futures. Editor, GigaOm. My obsession is the ecology of work, and the anthropology of the future.

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