Looking for evidence of technological unemployment? This older post from Andrew McAfee at M.I.T. looks at the total...

Looking for evidence of technological unemployment? This older post from Andrew McAfee at M.I.T. looks at the total…

Reading Time: 1 minute

Looking for evidence of technological unemployment? This older post from Andrew McAfee at M.I.T. looks at the total output of manufacturing in the U.S. (in blue) and compares it to the total number of employees in manufacturing in the U.S. (in red). When you see output going up and labor going down`that means that work is happening, but that it’s just not being done by humans. 

Source: http://andrewmcafee.org/2013/01/mcafee-technological-unemployment-us-manufacturing-econom/



  1. Taken in isolation, this can seem like a bad thing. Certainly for those who were displaced it is. Overlaying an all-sectors employment line may show that other jobs were created in the process (such as engineering and design to build the robots). A challenge for any given economy is whether these new jobs are created within the same country.

  2. Shaker Cherukuri China?  Who has the cash?  IDK.

  3. Well workers have shifted from manufacturing to the service industry. As stated above, new jobs are created as machines take up old jobs.

    What is interesting to me is that for some reason there is a preconception that automation will only disrupt blue collar jobs. I see no reason why machines can’t develop software, diagnose disease, trade stocks, or even write articles. Indeed the latter is much more lucrative.

  4. Gideon, thank you for this analysis.  Could you please update this post and share the same currently..  Thanks.  

  5. Gideon Rosenblatt – since technological advances will only continue, every society will have to find ways to employ people.

  6. Rather than lament this, it should be seen as a good thing and an opportunity for people to move from humdrum jobs and do more with their lives. Machines should be our servants, and we should be more than just cogs in the machine.

    I would prefer to call it technical liberation, rather than technical unemployment.

  7. As is unfortunately usual, the graphic displays a gratuitous failure of having the vertical axis start at 0 — well, even more gratuitous than usual I’d say.

  8. Paul Wooding the correct term in Economics is in fact “technological displacement”, which recognises that new jobs are created elsewhere (although not typically for the same workers).

  9. That graph does not show however, that the jobs are not moving somewhere else. Another thing is investment in R&D has still not risen to 2009 levels, if job loss and tech were so closely anti-correlated you’d expect their time series to shift in opposite directions (neither is there any lag evident).

    Instead the unemployment rate has dropped since 2009 and there seems no real shift in the percentage of the unemployed with only secondary vs tertiary education. Long term unemployment has risen though so I’m more inclined to think this is a capital/economic and not technological issue. For now.

  10. I agree with Tau-Mu Yi – not only should we not assume that blue collar jobs are most at risk – we have reason to believe, via the Moravec Paradox – that it’s the bulk of white collar jobs that are most at risk (a refrain I find my self repeating). So there’s this fat middle between blue collar and subtle expertise that AI can displace. Anything with straight forward rules that involves juggling lots of variables is fair game. That happens to intersect with most midskill white collar jobs.

    There’ll likely be a push down and out effect (some that would have typically gone white collar go blue, many go out of the market completely).

  11. Greg Anderson “liberating” from a social perspective. 😉

  12. Eli Fennell careful. We need the poor and uneducated to do all the mundane jobs to support the rich and well educated.

  13. Boris Borcic, the chart is indexed, not absolute. If you look at it’s starting point on the left, it is effectively 100, which for an index is effectively what you are calling “zero.”

  14. Just to be clear, I’m not saying that automation is only going to affect manufacturing or blue collar jobs, Tau-Mu Yi and Deen Abiola – this chart and the data behind it just happens to be focused on manufacturing. It’s pretty clear that any kind routinized tasks will be the first types of work that will be the most susceptible to automation, and thus the first to go.  Many of those will be white collar and in fact there is much reason to assume that those will be particularly hard hit. 

  15. And Greg Anderson, you nailed in your parenthetical insert – the jobs that do appear later in the creative part of creative destruction are very often not jobs that can satisfy the employment losses of those who have experienced the destructive part of innovation. 

    This is one of the things that complexity economics emphasizes, which is that the equilibrium assumptions of classical economics and the idea that the market is magically optimized ignores the existence of time. Sure, over the long haul, things adjust, but it takes time and there is plenty of turmoil and local imperfections that happen on the road to that equilibrium. 

  16. While I agree white collar jobs are also on the line (and are already being replaced), I believe they will be less affected than they could be. Managers could very easily replace me, for example, with a well constructed SharePoint site but there are other issues (rivalry between departments, not wanting to do the other BS that is never part of a position description, etc) that will buy white collar workers a little more time.

    Also, corporations will overstep, like what Foxconn is planning (http://mobile.extremetech.com/latest/221646-foxconn-is-attempting-to-replace-its-human-workers-with-thousands-of-robots), and there will be a backlash.

    Won’t matter in the long run but may give people a little more time to adapt.

  17. For those who don’t believe in the possibility of technological unemployment, let me ask this: 

    Is it possible that the real questions here are questions of learning and seeking entrepreneurial opportunity? To find the new opportunities that the automation opens up in the economy, humans need to be able to spot the economic opportunity and then learn how to do what is needed to fulfill that opportunity. And if that is true, what makes you absolutely sure that machines wont (eventually) become much faster and better than humans in finding those opportunities and learning how to fulfill them? 

    This is the real meaning of the race with the machines. 

  18. Gideon Rosenblatt No, the 100 is the reference unit, not the reference zero, and it makes it particularly wrong to not show the corresponding height in full on the axis. The whole point of showing the whole height is exactly the same as that of calibrating the value at 100 at some point: to express the current value as a proportion of the reference value. While depending on how the curves behave with time, it may reveal impractical or uninformative to have the axis start at zero, but this is clearly not the case here, and what we get is a gratuitous dissonance.

  19. Gideon Rosenblatt I do believe that someday technological unemployment will be a thing. And that it’s looking ever more likely, just it’s not happening now. And for a while yet it will be displacement and augmentation, not elimination. So things become cheaper more democratized hence more people can start companies and become owners of capital – which benefit most from automation. Quickbooks did not eliminate accountants it simply shifted their focus. Elevators did more for cities than just removing the difficulty of climbing steps – lots of intelligent tools will be augmentation in a similar manner.

  20. Boris Borcic, so let me ask you this – what year would the chart have to go back to in order for zero to be relevant for total manufacturing employment or output?

    This is why you set indices around some common point of reference that most people recognize.

    Sure, you could show the full height of the chart for scale reference, and if that’s all you’re arguing, then fine, sure, yes you could do that. Point taken.

  21. Gideon Rosenblatt Boris Borcic based on many similar charts I have seen that track productivity/employment/median wage there are really only two marker points that are applicable to today. Just after WWII (1948ish) and the mid-1970s. Anything further back is too out of context for the effects of recent technological advances. What Gideon posted here is inline with everything else I have seen. 

  22. I’ve just come across this article from July which is related – Lawrence H. Summers on the Economic Challenge of the Future: Jobs – WSJ – http://bit.ly/1p6JYgy

  23. That’s a good one, Paul Wooding. Thanks. I like the focus on income, purchasing power and dignity. 

  24. Deen Abiola, in my view, saying that you believe in technological unemployment, but just not now is equivalent to seeing it as some end state rather than an ongoing process that is accelerating. Technological unemployment is an ongoing process that creeps from one economic sector to another as the creative destruction of technological innovation destroys this one even as it then opens up opportunities in the next. It is happening now, just locally. I think the question that then emerges is when does creative destruction and local technological unemployment constitute a kind of global technological unemployment where humans lose the learning and opportunity discovery race against the machine, and destruction begins to rapidly outpace creation. That’s my take anyway…

  25. No I do not agree..Majority of instances of job loss currently on going are not due to technology, they’re from economic and financial mismanagement. And where there is a shift due to technology, it’s not a net negative, there is reshuffling not removal. Furthermore as someone working on a project with lots of AI, I can say most examples today are augmentative. Meant to be paired with a human. Sure now a single person can do more than 10 before but the space of opportunity is increased more than 10 fold.

  26. Let me explain it this way. Why would I replace a worker with a robot? Because it’s cheaper in the long run. So what does that mean? If it’s cheaper for me then it’s cheaper for anyone. That’s a lowered barrier to entry – It means that just a few more of those who could not before afford factories and assembly lines are now able to. So sure there are less factory jobs but there are now more people able to start factories selling all kinds of new developments. Each building off each other.

    Each new improvement lowers the barrier and since we can expect incremental change, perhaps with some unobtrusive policy we can manage this so the sphere of capital owners is ever increasing.

    The same for lawyers and doctors. Sure there are less doctors but now the average person can easily have access to world class expertise. Meanwhile the remaining doctors will be augmented or might shift to research etc.

    I believe that someday AI will be better at all things but at that point no one will have a job. No rich no poor. I also leave open the possibility of a dystopia and doomsday scenario but weight the possibility of such ever occurring to be very low.

    There’s a kind of scale free nature to the argument too you know. In the same way that automation can help ever poorer people accomplish more, it can aid third world countries catch up more quickly with first world nations.  

  27. Well, I’d put that slightly differently. There’s an ill-knotted interaction of financial logic, robotization, and globalization (=reaching the scale of the finite Earth) driving the system to an impasse at ever accelerating speed. All is not to say, “this is no more than a coordinate singularity”, there’s immediate need to “refresh” financial logic as we know it, to help the human planet drive itself to some kind of landing.

  28. Eli Fennell re:”if everyone were well educated”, that’s actually likely false, among other things the stats get skewed because the well(/over)-educated tend to hide their un/under-employment through solopreneurship/”consulting”/asf.

    The trend has been called “the precariat”:


    Check out this take by Brad Feld (VC/Angel and head of Foundry Group, a well-known incubator/accelerator that put Boulder, CO on the startups map):

    “…Capital remains incredibly cheap, so it’s flowing into wages. But that’s only at the high end of the market around technology jobs.

    [see another link below about what the “high end” means to tech founders/VC-types today!]

    At the other end of the spectrum, we have the famed jobless recovery with the elimination of massive numbers of jobs that previously existed, especially in industrial and Fortune 5000 companies.

    While this is happening, we have an entirely new class of [“]entrepreneurs[“], or self-employed, being created by companies like Uber. [And other signs of the “Precariat”?!]

    Yeah – this [stuff…] is super complicated and it plays out over a long period of time. In fact, it might only be really possible to understand what is happening in hindsight.”


    Here’s that promised other link:


    But when you say “many”, keep in mind that e.g. all of WhatsApp, now serving 600M Monthly Active Users, was built/run by only about 50 engineers until recently!

  29. Deen Abiola some good thoughts,  but re:”perhaps with some unobtrusive policy we can manage this so the sphere of capital owners is ever increasing.” — what about this though:


    Crowdfunding is in full swing, but Crowd-INVESTING (AKA capital building) is still being blocked by the SEC despite the JOBS Act… I wonder why…

    Consumerist society up to now is specifically predicated on the consumer consuming and NOT building capital/productive assets. Mostly only sham/make-believe assets.

  30. Does this technological unemployment/displacement rely on or relate directly to the capitalist (or any) economic system?

    ie it is an inevitable consequence, or would the only way to avoid it require a different economic or political system?

  31. You raise an interesting point Alex Schleber, that the status quo will resist any change that forces them to relinquish their advantage. I have two thoughts on this.

    1) I’m hopeful they won’t be that stupid. A starkly polarized world would be much too unstable for the kinds of conditions that are optimal for smooth economic growth. Plus, surely they’d realize that it’s better to have some of a lot than a lot of some. The developments would only take away their ability to charge excess rent, quality of living would still be up. As has been the case for centuries…they can’t fight that.

    2) The system would just work around them. Moving to digital currencies and Distributed corporations and collectives. Toys till its too late.

  32. Paul Wooding That is the thesis of Joseph_Schumpeter. At its most basic, ‘creative destruction’ (German: schöpferische Zerstörung) describes the way in which capitalist economic development arises out of the destruction of some prior economic order

    Creative destruction can cause temporary economic distress. Layoffs of workers with obsolete working skills can be one price of innovations valued by consumers. Though a continually innovating economy generates new opportunities for workers to participate in more creative and productive enterprises (provided they can acquire the necessary skills), creative destruction can cause severe hardship in the short term, and in the long term for those who cannot acquire the skills and work experience

    Schumpeter’s theory is that the success of capitalism will lead to a form of corporatism and a fostering of values hostile to capitalism, especially among intellectuals. The intellectual and social climate needed to allow entrepreneurship to thrive will not exist in advanced capitalism; it will be replaced by ‘laborism’ in some form. He points out that intellectuals, whose very profession relies on antagonism toward the capitalist structure, are automatically inclined to have a negative outlook toward it even while relying upon for prestige. There will not be a revolution, but merely instead a trend in parliaments to elect social democratic parties of one stripe or another. He argued that capitalism could collapse from within will come about if democratic majorities vote for restrictions upon entrepreneurship that will burden and destroy the capitalist structure. He also emphasized non-political, evolutionary processes in society where ‘liberal capitalism’ was evolving because of the growth of workers’ self-management, industrial democracy and regulatory institutions

  33. As can be seen from above, the worry over technological unemployment is a redressing of old and important Marxist ideas (arguable it stretches all the way back to Adam Smith). It is no surprise then, that aspects of the solution have strong socialist slants. 

    But I think it’s actually human nature not creative destruction that leads to this constant churn. The right combination of laziness, intelligence and creativity (the kind that makes you spend 20 hours to save 2 minutes of tedium).

    But it’s not all bad when I can print out solar panels, download some plans and print out circuits for a wind turbine and some robotic arms. Buy some Baxters 5.0 and start exploring the space innovation that is far larger than the imagination of any collection of human beings.

    Or maybe I’ll buy a sequencing toolkit and thanks to robot mass pippeting, ‘chemputers’, cheap storage and bandwidth; small collectives of us can do breakthrough biotechnological research. Sure technology eliminates jobs but jobs, massive corporations are anomalies which go against the human psyche.

    This would be a return to how things should be. And unless there is some ugly social policy that makes sure everyone remains poor during the gradual decline of prices and that only the rich have access to the amplification tech, I find this kind of enabling of the individual and collectives to be far more likely than people living like rats in slums. I’m more worried about unsustainable energy and resource use policies.

  34. Deen Abiola I think you forgot the part where you eat… asf. 🙂

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Sign up here for the latest articles. You can opt out at any time.

Subscribe by email:

Or subscribe by RSS:

%d bloggers like this: