Will you lose your job to a robot?

The Upshot surveys some optimistic and pessimistic views.  I thought it would be useful to restate my views in a single, simple blog post, here is the enumeration:

1. The law of comparative advantage has not been repealed.  Machines take away some jobs and create others, while producing more output overall.

2. That said, some particular kinds of machines increase the relative return to skilled labor.  If the new jobs require working with computers, and working with computers effectively is hard, reemploying lower-skilled workers at good wages may be difficult.

3. Smart software, factor price equalization, and better measurement of value have all boosted income inequality.  Returns to working for low-skilled workers have fallen or stagnated in many regions (not North Dakota).  Returns for many higher skilled workers have risen, but most of them were working and working hard already.

4. Lower returns to unskilled labor mean (on average) that low-skilled laborers will work less.  This effect may interact with government benefits but sometimes people decide to work less or search less hard for a job for reasons unrelated to benefits.  These decisions may produce feedback which weakens pro-work norms in the broader culture.

5. The employment to population ratio will be lower than it otherwise would have been, because of “robots” but not only robots.  The natural rate of unemployment will be higher too.

6. Many of the new service sectors jobs will be better suited to women rather than the most unruly men.  Physical strength will matter less, conscientiousness and teamwork will matter more, and much of the burden of these adjustments will fall on lesser educated men.

7. Facebook makes it easier to get sex and keep friends without having a job.

8. There is good evidence for each of these propositions, although it may be questioned how great is their combined import.  In the meantime, yes robots may lower employment, although the catchphrase “robots are destroying jobs” is misleading rather than illuminating.


Better machines use to reduce working hours not the number of jobs. Why do you think that will not happen with robots.

Better machines could reduce working hours, all else equal. Or, the firm could lower price, increase demand and increase hours.

Time overhead. Nobody is immediately and thoroughly productive from the moment they clock in at the start of their shift, to the end of the working day. Even fast-food workers will have to change into their uniforms and do a quick cleanup of their work area, and they'll usually have to listen to the manager blathering on about something somewhere during the day. And if that's all there is to it, probably no big deal, but the sort of work that is most resistant to automation is the high-complexity sort that comes with much higher overhead levels. Training and retraining, mandatory and otherwise. All the emails from everyone else working on your project, that you have to read before you can make a contribution of your own that won't be wasted effort, and all the emails you have to send them to make sure they don't step all over your work. Meetings and more meetings. Etc, etc, etc.

When we were talking about whether assembly-line workers should put in 60 hours a week or only 55, this made little difference. We're now at the point where people are nominally working 40-hour weeks, with 10-20 hours of overhead included.

So, I'm an employer who hires two people to work 40 hours a week, each optimistically 10 hours overhead and 30 hours productive work - 60 hours total productive work to me, at the end of the week. I buy some robots and cut my labor demand in half; now I only need 30 hours.

Plan A: fire one guy, keep the other. I get 30 hours a week of productive labor, and I pay 40 hours' wages.

Plan B: keep both employees, reduce their shifts to 25 hours/week. Still getting 30 productive hours a week, but now paying for 50.

Plan C: Naively say that half demand means 20-hour shifts for everyone. Now I need to hire a third person, and I'm paying 60 hours a week.

How about four 8.5 hour days. Less beginning of the day overhead, less overall hours. There is still overhead per employee from HR standpoint.

That only helps if the overhead is a fixed penalty at the beginning of the day. Which it maybe is for assembly-line workers and fast-food workers, but not for the sort of high-complexity jobs that will be most resistant to automation. If I need to sort through three hundred emails this week, take an hour of mandatory sexual-harassment training, and spend two hours meeting with colleagues to coordinate our joint strategy on X, it does not matter whether I do that over three days or four or five, whether those days are six or eight or ten hours long. All that matters is, there's maybe 15 hours of that sort of stuff I have to do, before I can do my first hour of actual productive work.

Well, we could end up with 3 day weekends. Less work, but solves a lot of overhead issues.

Good thing our elites let in all those tens of millions of unskilled immigrants. That show real foresight.

I think TC *does* think it will happen with robots. That's why he wrote

1. The law of comparative advantage has not been repealed. Machines take away some jobs and create others, while producing more output overall.

I agree with this as a statement about the real world. But, it can't be turned into a universal truth. At some point, robots really are different from other machines.

I can imagine as civilization of machines with no need for humans. Clever, adaptive creations might roam the world, striving in a Darwinian struggle to replicate the information that creates them. That world would have an economy, it would have consumers, producers, preferences and probably even joy and sorrow. It just wouldn't have any humans. I can also imagine all kinds of middle grounds between that economy and the human-centered one we now live in.

It didn't in Japan, doesn't in China, and didn't in the US from 1800 to about 1980.

Its an old story. Surely you grew up with the fable of John Henry, the Blue Ox, etc. Or were you born late enough to grow up since 1980 when labor was a drag on the economy and the war on labor began. Wall Street and Republicans are clearly anti-labor - workers chose to be worthless. They deserve to be paid 0.1% to 0.01% of those who chose to be wealth and job creators. Treating workers and CEOs as simply workers, employees, who aren't that different is pure communism.

If the driving public policy is creating jobs and higher incomes, then labor saving technology, automation, robots will create more production going to more people because more people can afford more.

More might mean better in some fashion that is taken for granted today. Public water systems are high technology that saved the labor of hauling water from the river or well, and that enabled indoor toilets that meant you didn't need to prepare for the bad weather going to the outhouse, and the labor of digging a new hole when the old one filled. Unpaid labor in the house was externalized to create jobs that employed people in the household to pay for connecting to the water system. Jobs making pipe, digging trenches, making pumps, building treatment plants, making the labor saving machines for digging trenches, etc.

In the US since about 1980, investing in new capital has been opposed unless it generates really high immediate rates of return. To the degree that this can be done, the rates of return on old capital have been jacked up, and making investment to improve existing decaying capital is thus opposed because it returns too little compared to a few isolated technologies that can generate high returns.

The job losses or lack of growth has come from all capital being required to be in the 1% of exceptional high returns. The best solution is to make existing capital labor saving by allowing it to decay.

Oil pipelines that are privately owned have benefitted from labor saving neglect. That is a business innovation. Use political influence, lawyers, public relations to reduce labor cost. Lawyers and lobbyists are great labor saving technology because the costs of oil pipeline leaks is much less than using the great new technology of construction equipment and technology and better equipment technology to deliver century life capital that is cheaper to build and use.

For example, metal and coating technology can make pipe far more resistant to corrosion and to shear and impact stress so the pipe will require less energy to move fluid and also require less maintenance over a longer time. New construction technology and equipment allows installing pipe without cut and cover which destroys the integrity of existing pavements.

The private and public policy has been to put up with increasing pollution by decay to save labor instead of replacing or refurbishing old infrastructure with high tech solutions because using "robots" requires paying workers. Horizontal drilling is a robotic excavation method, but the equipment is very expensive because of the large amount of skilled factory labor, plus the skilled labor force on the job site which can install probably five times the length pipeline per day with less disruption to others. But it is cheaper to not replace the old pipeline and spend on marketing to sell the idea that oil industry is clean and that government is the cause of leaking corroded pipelines.

Economists and business experts in the US argue that technology is desirable for creating high profits and thus pumping up the price of the derivatives of capital, not for creating consumer or worker benefits. Sure, economists will extol the great consumer surplus of the iPhone which gives you a supercomputer and global communication for a mere 100 hours of labor plus a mere 10 to 20 hours of labor per month, and you have the benefit of watching video off the Internet as you walk to your minimum wage job lasting 4 hours which you found using your iPhone. You are so much better off than in the 60s when you would have worked for three times the minimum wage while easily buying newspapers and calling around using the easily afforded landline for the union job that would double your income, give you benefits, and job security for a couple of decades.

How can you have a consumer surplus when technology makes you worse off by private and public policy.

In China, public policy is to make workers better off, so smart phones help entry workers find better jobs, which is forcing employers to put on the golden handcuffs of training for career advancement and higher wages to cut down on worker churn, something possible because government policy is committed to increased demand for production.

In China, you can probably afford your own 10-15 square metres to live in now that minimum wages are double what they used to be.

It's a little early to start romanticizing anything about working and living life in China. It is getting better, and that's what matters.

Nope. The jobs will be far more complex and demanding, and the cost of training two people to work 30-35 hours a week compared to training one person to work 60-70 is far less. Plus there isn't twice as many people available to train.

The people who know how to produce, design, etc. those robots will be called upon at very high price to work many, many hours a week.

But I wish it could be the way you suggest and we cold evenly distribute the gains so we could all work 6 hour days or 4-day work weeks. That isn't likely to happen without significantly rejigging redistributive mechanisms.

The short answer is it depends - primarily, it depends upon whether the incorporation of automation technology allows both the volume of business and the business' revenue to grow enough with respect to the cost of doing business to create new employment opportunities for work that cannot be automated.

But all bets are off if the cost of doing business changes so much that those gains are offset.

Doesn't "Returns for many higher skilled workers have risen, but most of them were working and working hard already." combined with #4 imply more growth in servant-type jobs? Busy 2-career professional couples who benefit from the robot economy and earn high wages will employ the unemployed at lower wages while the professionals work ever longer hours. I see this happening already and it will only increase. The refrain "It's just so hard to find good help these days" will make a comeback. However, we will dignify it by having everything be arms-length contracts instead of "live-in" servants. More young people are going to make a career out of TaskRabbit style odd-jobs services, nannies, personal-chef services, house cleaning, home repair, etc.for the wealthy classes. Also personal assistants, tutors, coaches, fitness trainers, dog-walkers, shoppers. Not sure if robots will take these jobs also, not for a while at least.

The issue is externalities, ostensibly classless societies don't breed the sort of respect that placing underclasses into high trust that many service positions require.

re #7, is that your personal experience, or are you speculating? No, this is not a snarky question.

Yeah that one's a little out of left field.

If Facebook is changed to a broader "internet" then its a reasonable speculation.

The internet in general offers people another way to have social interactions, outside of work and school. That probably matters in the sense of making not working something different from being socially isolated and friendless.

They always were different.

I think it's a nod to Roissy's "Drop out, tune in and turn on" message.

Money matters less and less in terms of sexual market value, but TC couldn't say that.

Okay, asking as the maintainer of http://lesswrong.com/lw/hh4/the_robots_ai_and_unemployment_antifaq/: If you phrase it as "automation increases relative returns to high-skilled labor", why did we not see fewer low-skilled people working as a result of earlier automation? Is the idea that only the New Automation requires high-skilled operators that require new levels of IQ in order to get the required college credential? That the New Automation advantages computer programmers who genuinely require rare mathematical talents not just for credentialing purposes? On the surface of things, it doesn't seem to me that we can plausibly tell a story where computer programmers and STEM graduates are capturing enough gains to mass-unemploy unskilled labor.

If the main message of the post is that new service jobs require politeness instead of the ability to shut up and work in a factory line, therefore disadvantaging unruly men whose particular work was automated away, then I can see how our society may have previously employed unruly men especially at the jobs most subject to automation (those requiring no public facing or social skills, just physical dexterity or rote repetition and the ability to focus in exchange for money).

We did see fewer non-skilled workers as a result of earlier rounds of automation. Far, substance farming, rag-picking and other such things were ground-state jobs that anyone with enough backbone could do if they were desperate enough. The industrial revolution created demand for turners-and-fitters, secretaries and so on.

Eliezer Yudkowsky: "...why did we not see fewer low-skilled people working as a result of earlier automation?"

I think we have been seeing that for decades, though at a very slow rate of change. I think we don't have to worry about the loss of jobs to automation at this moment, but the near future is another story, because the change in machines is going to be more qualitative, as you yourself point out at your link.

Already, automation may underlie a small portion of the tendency to "jobless" recoveries we have seen for decades now. This is not the first "jobless" recovery; but the big problem at the moment is not automation.

The output gap or depression we are currently experiencing was initiated by the sharp fall in housing spending and the concomitant consumer credit crunch, then it was propagated from that moment by the fact that monetary policy is almost useless at the zero-lower bound (except to prop-up the structure of ownership in the financial subsystem). There was a paltry fiscal response (which, in all prior recessions back to Reagan and further, had been increased adequately to compensate) + a political failure to writedown debt across the board, both of which would have got more employment going. And we also suffer from the fact that about 3% of income is still going into the trade deficit, so the consumer demand that ought to create more domestic incomes continues to be slowly sucked away. (This is Dean Baker's macro analysis, more or less.)

But in the long term it looks to me like automation will cause much more unemployment in the future, and may finally overturn capitalism due to the Marxian-Schumpeterian psychological argument. The rate of automation may not need to accelerate.

The reason will be because the QUALITY of the automation is going to change in two very different ways: 1. to bipedal robots which can invade more low-skill work areas and supply better brute force, finally without the need for management, and with easy parts and software repair; and 2. to expert/intelligent systems which will surpass many human and cognitive functions, if only by mass-computation algorithms which simply try all combinations of things until something clicks on a problem.

Finally I think that Tyler Comen's "comparative advantage argument" (his point #1) will fall way, if only because humans have an advantage over machines only in non-algorithmic, Gödelian, contextual ways. But these ways are not necessarily economically valuable, although there may be a brief market hierarchy of pure artists, entertaining goofballs, and trendy fascinators. Finally, though, smart pills and genetic re-engineering will enable everyone to have the same skills and talent.

Speaking of unruly men, what does automation mean for China, which depends on relatively cheaper to drive its export economy?

Not only will chinese companies be tempted to automate, but foreign companies in particular as they look to manufacture at home and save on shipping costs.

Throw into the mix the 35 million excess males due to selective abortion and the prospect of Unruly Men in China is a little frightening.

Automation will probably be driven by China quite soon, given that wages are rapidly rising there (in no small part due to rapidly rising official minimum wages).

So you write a whole post proving that automation destroys jobs, but just for the hell of it, you conclude: "the catchphrase 'robots are destroying jobs' is misleading rather than illuminating."

Except in that it's effectively true. As you've just conceded.

I seriously don't believe there's a more self-deluded economist on the face of the earth. Markets in everything means we have to live with the consequences of markets, but we don't have to price them into our individual- and enterprise-level economic choices. Robots will definitely mean more unemployment and fewer desirable jobs so long as we leave the process up to markets.

We could decide to share the advantages of innovations across the entire workforce. But markets aren't interested in that, and we're only interested in markets. Markets markets markets.

"I seriously don’t believe there’s a more self-deluded economist on the face of the earth. Markets in everything means we have to live with the consequences of markets, but we don’t have to price them into our individual- and enterprise-level economic choices. Robots will definitely mean more unemployment and fewer desirable jobs so long as we leave the process up to markets."

See #5: I'm not sure what delusion you think he is under.

Also, "fewer desirable jobs" depends on what jobs are desired. Having worked both in an office and on an assembly line, I'm perfectly happy ceding to robots the assemby line work.

If you think all that's being automated is rote work, think again. Lawyers, copywriters—more and more intellectual workers are getting replaced.


Even if the skilled classes aren't being 100% replaced by robots, it still makes a difference if computers and networking replace the majority of those workers. e.g. How many fewer copywriters do we have now that we have advanced word processing programs that edit mistakes and quickly format your work for print? If a piece of software doubles the productivity of your workforce, that can mean that you produce twice as much work, or it can mean that you halve your workforce.

If we could get you a job shoveling up the shit behind horse-drawn carts, you would take it?

1. Automation has been destroying jobs for centuries.

2. Today, there are more jobs than ever.

This. "Robots" is just a new word for advances in technology.

Yes, the returns are greatest to those who can best master the technology; but even a sub-70 IQ can drive a car or dial a cellphone. Cellphone engineers are reasonably well-paid, but not obscenely so. Competition between skilled workers keeps salaries modest: I don't see that changing.

"Today, there are more jobs than ever."

Well that settles it. There are more people, and there is a lower proportion of well-paying jobs, but as long as there are more today than there were 100 years ago, everything must be fine.

The purchasing power of per capita income in the US in 1914 -- the *real* income of the average worker -- was about one eighth of what it is today. That obviously doesn't account for distribution, and there other factors that make the comparison across time less than exact. But there's no way there was a *higher* "proportion of well-paying jobs" in 1914 than there is today.

There was a lot of really dire poverty in the USA a century ago, urban and rural. Average life expectancy was over 20 years shorter.

I'm with you on the underlying principle that maldistribution of income is a serious problem! But there was no golden age, and blaming tech change makes you sound like a loon.

Oh, CD, I didn't mean to imply I thought there was a golden age. We seem to be of similar values. But automation could cause drastic shifts, because markets don't control for such things. "Robots" basically represent capital competing for wages against people. There isn't some endless demand generation machine out there that will always prompt capital to find new ways to put humans to work. This is possibly no better for capitalists than it is for workers, in the long run. With no structural mechanism in place to maintain consumer demand on one hand and enough jobs requiring manual/human labor on the other hand, there's no rational reason to believe we'll make it through. It requires the same kind of faith that capital will innovate our way out of climate change and out of resource depletion. I can't prove that it won't, but it's pretty clear it very well might not.

"There are more people"

Ding ding ding.

I'm constantly amazed at how people, particularly smart people, have so little faith in the ability of human beings to make themselves useful to one another.


No structural mechanism to sustain consumer demand? What about the 15% of the US population below the poverty line? Any extra income goes directly into consumption. Extrapolate to all the people coming out of poverty around the world and those poor robots are going to suffer to keep up with the demand. Some people just want a roof, food, clothes and TV, and they are billions.

Isn't it remarkable, then, that there are a fair number of people who have spent years unemployed and on public assistance? And in very poor countries like Haiti, many, many people who don't have enough work to get them out of truly godawful poverty and misery. I wonder why they haven't managed to make themselves useful to others in some way that would get them some money. Why, it's almost like there might be some kind of social or economic forces at work here, keeping those unemployed poor people unemployed and poor.

Yes some people do a poor job of developing themselves, and some societies do a poor job of providing an environment where people can develop themselves.

Here in the good ole USA, something like 8 or 9 out of 10 people with an objective of gainful employment achieve that objective. Today. This is not the impression you get from listening to whiners and hand-wringers though.

"Anyone can" =/= "Everyone will". We can carry those who can't or won't as long as it's a manageable percentage.

Academics view themselves as carrying out the 21st century version of the white man's burden. They think with enough hand-wringing and ivory tower scribbling, they can save their benighted lessers. Shot through with condescension. It's clear to me that most of these people know few or any of the people they constantly write and worry about. They lay awake at night wondering: what will we do with all these people? Implicit as that these people are helpless and can do nothing for themselves. It's a weird self-important and self-congratulatory vibe and I'm heartily sick of it.

They will destroy jobs unless our material desires are unsatiable.

They are unsatiable. And hence we suffer for need to continuously satiate our material desires.

But let's not all learn too much from Buddhist philosophy too quick or everything will fall apart!

In the meantime, there are hundreds of million of Africans who look forward to satisfying all manner of material want, and can continue to do so on an increasing basis for a very long time before they would be advised to say "enough, I think I'll just try to be content now".

I agree. Maybe what he means by ‘robots are destroying jobs’ is misleading rather than illuminating.” is that some workers will be (much) more affected than others.

"That said, some particular kinds of machines increase the relative return to skilled labor. If the new jobs require working with computers, and working with computers effectively is hard, reemploying lower-skilled workers at good wages may be difficult"

That's why we need to import them by the 10s of millions.

Again, "Average is Over" is the most convincing case for socialism I'm currently aware of.

Again, this is to say that there aren't very convincing cases for socialism out there generally.

"Facebook makes it easier to get sex and keep friends without having a job."

No, no TC, please no, you have to realize how creepy that sounds...

The robot, model CAPLAN2.0, that recently replaced Cowen hasn't been updated with the ANTICREEP software yet.

I don't get the connection between “Facebook makes it easier to get sex and keep friends without having a job.” and "Will you lose your job to a robot?".


This will shift the supply curve for labor to the left. Solve for the equilibrium.

Butterfingers explained this well.

I see you in your office.
Sending and recieving checks.
But i know the quest for money.
Comes from a quest for sex.
So i cut to the chase.
And forgot about my job.
And decided to stay home wit yo mama on my knob.

As a fan of history, can we make a parallel between a society where most of the low wage workers are replaced by robots, such as ours might become, and the Roman Empire, where low wage work was replaced by slaves? Will we create a large underclass with few real prospects, living on the dole, obsessed with entertainments and sports to pass the time? Have we already?

But police, security guards and soldiers will also largely be replaced by robots controlled by a smaller number of smarter, better-trained people, so drone riots won't be nearly so scary for the powers that be in the future, either.

Robot Juvenal is still going to ask who watches the watchmen though.

The people in charge of the society, by definition.

Why is anybody worrying now about elimination of jobs by automation, when there is not the slightest bit of evidence automation is occurring at an above-normal rate or soon will accelerate. Jared Bernstein has a nice chart here: http://jaredbernsteinblog.com/wheres-the-automation-in-the-productivity-accounts/ It shows that ever since WWII, productivity has increased at about 2% annually with three notable exceptions: The trend got above 3% in the dot com era and it sank to about 1% in the late 1970s and again since 2010. In other words, at the present time automation is eliminating fewer jobs than at any time in over 30 years.

History teaches us that machines don't kill jobs.

Actually, employment in manufacturing has declined enormously in most economies with increasing automation (probably also a lot due to an increasingly international division of labour).

The reason for continued job growth was the explosion of the services sector.

What machines do NOT do is depress wages, because one hour of your time can be so much more profitably paired with machines (assuming you have learned how to use them).

Just noticed that Bernstein's chart does not go back to 1948. Mine does, but it's not online. Between 1948 and 1970, the story was the same--productivity increased about 2% per year most of the time but with a surge to more than 3% in the 1960s.

One job where people have a comparative advantage over robots is prostitution. That ties into argument # 1, # 6, and maybe also # 7.

For most of history many men were ambitious because a good job and money conferred status. Now that we live in a more transactional and online society, where women view men as more a source of sex than a source of financial support, good Facebook photos and status updates are worth more than a good job in many respects.

So a robot might take your job but it's not going to fuck your wife.

That would be my job.

See, that additional automation that made everyone richer *did* create more service sector jobs. Comparative advantage FTW.

Actually, you missed the reference. Gigolo Joe was a robot from the movie AI.

Automation has a big externality that the "free market" using comparative advantage doesn't account for. A corporation might reduce costs by replacing a worker with a robot, but if that robot responds by going on the dole, then the taxpayer is paying for that. A corporation might save 200$ a month by replacing the worker with the robot but the taxpayer is charged an extra 800$ a month.

That's not an aspect of the "free market". A Welfare system puts a floor on the price of labor and reduces the incentive to seek it out.

Cyclical no doubt, as it increases the incentive for firms to reduce the burden of the various welfare state mandated worker compensation by replacing labor with automation or outsourcing.

A well designed tax system would ensure that $200 of corporate savings wouldn't put people onto $800 doles.

I read the entire comment section and no one commented on the weakest point that Cowen made. Namely "Physical strength will matter less, conscientiousness and teamwork will matter more, and much of the burden of these adjustments will fall on lesser educated men." I would like to see this assertion justified. I imagine (thought I don't like to put words in other people's mouths) that Cowen would largely use the past to support this assertion. And it is very true that the history of automation up until the present time has largely followed that pattern. But the past is a very poor guide to the future. I submit that anyone who thinks that the past pattern will hold true for future jobs has no understanding of what jobs still require physical strength. Most of the easy ways of taking physical strength out of the equation have already been done. Those that remain will be very hard to replace with robots and I think most other service jobs will be replaced with robots long before they are.

You don't need much in physical strength to work on most assembly lines anymore. But you do need physical strength to be a roofer, Iron worker, tree man (very rarely a woman for good reasons) linemen, and so on and so forth. A lot of these jobs are a lot easier then they were thanks to machines, but they are still very physically demanding and this is because there are key parts of the job that are almost impossible for machines. I personally think robots will be able to replace a store clerk long before the can have the skills to prune a tree, put a roof on, or replace down power lines. I think I am reasonably up on current robotics research and most of the advances I see coming down the pipe are automatons of what machines can already do (i.e. taking humans out of the equation on jobs where the work is already done by machine). To put it another way, I see bulldozers that can be told to do what needs to be done with out a human operator in the near future. But a robot that can jump down in a trench and have the judgment and skills to make a necessary splice is a ways off.

I think you make a pretty good point. I suspect a lot of physical jobs will become easier due to automation, but it's quite likely that they won't be replaced at a higher rate than less physical jobs will be.

Very good comment.

Where do ZMP workers come into the picture here? Does technology expand the fraction of people who are ZMP workers somehow? That would lead to an expanding pool of unemployable people.

I think that's a question that no one has shown a good answer to. Basically we don't know and there's a lot of debate between two different positions:
a) automation will follow the historical pattern of replacing certain jobs, but not decreasing the total number of jobs [work force participation remains roughly the same]
b) automation (in conjunction with the existing welfare state) makes the lower half (or more) of the IQ bell curve largely unemployable

Does anyone track labor force participation by IQ? What has been the historical pattern?

Most people of high IQ don't preoccupy themselves with such silly questions as whether some subjective measure of "intelligence" is particularly related to employability.

Most often, unless you're literally retarded, the ability to communicate, empathize, and in various ways be human, is far more relevant than IQ. Most of us can learn to do most jobs, given half a chance.

My fear is that one day we will become too smart for our own good and destroy everything. Maybe there's a good reason that so many guys love dumb blondes!

One question is whether we will make robots precisely to disemploy the ZMPs first.

If only. I am sure I will be working for dumbass robot.

This is much more interesting that Engineering's treatment of Robotics. There it's all about numerical analysis, high-order and non-linear ODE's, translation and rotation matrices, dynamic stability, feedback and logic controllers, structural analysis, etc, etc. And that's just the intro class. It gets hard at the graduate level.

My guess: The ZMP universe will expand. Rapidly.

I don't know. There are vast amounts of the economy that require skill and expertise where automation is far too costly. There are enormous numbers of maintenance and repair jobs available, and as things become automated and more tightly controlled to meet energy standards or simply the operating parameters have narrowed due to cost cutting in manufacturing, the more people will be needed to keep the stuff running. We think of automobiles or cell phones when we think of manufactured goods, but what in your home would you be happy with if it needed replacement every two years, or every five years for autos? I work on stuff that has a minimum life expectancy of 20 years, and many businesses run on equipment 40 years old. Replacing them with new might cut maintenance costs by 1/3, and the numbers just don't make sense.

The challenge isn't lack of work. Just ask anyone who has a skill and the ability to perform. The problem is skill and ability to perform. The first can be fixed with experience and training. The second is a social issue; someone with a recreation chemical habit, someone brought up without a father, someone who is dealing with marital problems or the complications of mixed up families will have difficulty meeting the demands of the work.

What is gone are jobs that you can show up stoned or drunk and get by enough to get paid.

'What is gone are jobs that you can show up stoned or drunk and get by enough to get paid.'

In an average is over world, that is completely untrue - those with sufficient wealth can get as drunk or as stoned as they wish, and have no concerns about their 'job'. Mel Gibson comes to mind as an example. Mainly because his job these days is in movie production - perhaps you have heard of 12 Years a Slave? - http://en.wikipedia.org/wiki/Icon_Productions

Damn you're dumb.

Is there evidence that, summed over the planet, the number of low-end jobs has decreased, rather than moved due to globalization? Not being snarky, I'm actually curious but lack the time and expertise to figure it out ("it" = "are we blaming technology for the effects of a massive labor supply shock due to globalization").

I think low wage Americans are aware that international competition for low skilled jobs has done more to reduce their opportunities than automation.

Look at shares of populations working in agriculture. Lower end manufacturing and services jobs have exploded worldwide, while people continue to exit agriculture in droves, in part afforded by enormous improvements in productivity.

I think you could argue that advances in communications and transportation technology have increased the mobility of labor sourcing, that our increased connectivity also makes it easier to outsource jobs to countries with inexpensive labor. Not exactly robots making the replacement, but technology still paves the way for efficiency.

Automation keeps manufacturing in the US. Lower-skill jobs disappeared mainly because its was simpler to offshore than to automate.

If automation is a threat to the US, it's through the rising productivity of EM manufacturing raising the bids for natural resources.

Eventually robots will be able to meet our every basic need in the economy. Sure, there will be some very high paying jobs, and much opportunity in industry which have not yet even been dreamt of. But when automated production produces well beyond our needs for food, housing, transportation, etc., then distribution will be the real question, since in a capiatlist system people seem to assume the right to capitalize on the entire history of technological development as reflected in the latest innovation.

Probably there will be many ZMP workers and individuals from the strict economic sense.

We should institute a guaranteed minimum income to free people up for innovative and/or entrepreneurial endeavour long before that time comes.

I suggest a minimum number of accredited volunteer hours per week for guaranteed minimum income earners. No freebies unless you're a gimp or well beyond grey.

Note that de-industrialization is often viewed as a sign of progress towards a modern economy.

Automation will be a part of moving onto ever better things. Managing the social impacts of the transition will be the challenge.

Underlying all of this is the assumption that strong AI is not possible.

Even if it isn't, every job reducible to a routine set of instructions or quantitative processing of some kind is going to be automated. Period.


But the question is, what will that do to employment?

Perhaps all that added wealth will make more jobs possible--freed from the need to dig ditches by hand or farm by hitching up a plow to a mule, we had people who could go operate machines in factories.

Perhaps all that added wealth will make it possible for the winners from this change to just not need the losers very much. You can see this phenomenon now in US ghettos--the productive, wealthier part of the society mostly has little to do with them or their inhabitants. Places with a persistently high unemployment rate are places where, right now, nobody is all that excited about offering the locals jobs. What if that's what the future looks like?

Tyler's model seems to be something between these two extremes. The returns on low-skill, low-IQ work go down, because less of it is needed now that automation can perform some of it. The increasing wealth of the society also probably means more alternatives to work to survive--public assistance, sponging off your girlfriend or mom, begging, small-time crime, whatever. Overall, you see more people remain unemployed.

The real critical question comes down to whether the returns on low-skill, low-IQ work must go down. Perhaps as existing low-skill jobs get eaten by automation, it will make sense to move low-skill, low-IQ people into new jobs--think telemarketing, or professional line-standers, or something. Or perhaps it won't, in which case we end up living in a rather darker future.

Why the excessive interest in IQ?

I've never heard of an employer with particular interest in IQ. Training, education, experience, interpersonal skills ... but not IQ.

If it were as relevant as you seem to think, many employers would give use the test for screening. They don't.

>>Overall, you see more people remain unemployed.<<

That's exactly what I foresee. What we're seeing now is just the thin end of the wedge.

I think the techno-optimists are full of it. They refuse to see that, if strong AI is possible, this industrial revolution is inherently different. Once machines can do EVERYTHING--physical, cognitive, even "high-touch" emotional--that people can do and do it better, faster, and cheaper, our current economic model unravels completely. How does that model function when the labor of the overwhelming majority of the population, regardless of how well-educated they are, is worth nothing?

The answer is it doesn't.

I'm mystified by the assumption or argument, in this thread and widespread elsewhere, that automation creates jobs. How many does it create, exactly the same number as it displaces, more, or fewer? How does anybody know that? What's the model? How does the model fit with post-WWII data? If automation creates more jobs in the short term, as perhaps it did in the dot com era, the jobs eliminated are eliminated permanently, no? If so, which time frame is most important?

I also think this would be a good place to introduce the discipline of micro-foundations. If it doesn't make sense for any one firm to automate when no jobs are eliminated, why does it happen across a whole economy?

Sure it *creates* jobs, just not as many as it destroys. As you point out, common sense dictates that the only reason to automate is to save labor costs (per unit of production), otherwise why do it?

Take the very near-term real world example of fast food automation:

The burger machine and the cashier kiosk (linked below). Both of these machines are designed to replace several workers at each location. Jobs are created by the manufacturers of said equipment and via the maintenance/repair of them. But realistically, what are we talking about? maybe 50-75 workers at each manufacturing plant and one guy servicing 10-30 machines in a given geographic area? Each pair of machines takes the place of 5-7 shift workers (10-14 part-time workers assuming 2 shifts/day). At the point of 100 pairs of machines in operation - 1000-1400 fast food jobs have been 'destroyed' and 175 manufacturing/servicing jobs take their place. The next hundred machines 2800 fast food jobs are kaput with maybe another 25 service jobs in place....and on and on.

It all sounds a bit Luddite to be negative on this, but I think the speed at which this will happen will catch most off-guard. At some point the utter unemployability of these millions(?) of displaced workers will need to be addressed. I know its an article of faith to many economists that the creative destruction outlined above always pencils out (ultimately) but can we realistically forsee a new set of large-scale low-to-medium-skill human labor intensive industries being created in the next 10-20 years to balance everything out? I doubt it. So what then?


Machines could save and augment labor. I can see it happening and I don't see why it's a bad thing. Sure, it'll cut out some unskilled labor and jobs, but it would save more money and time that could also be put back into the economy. It makes things easier, time more productive, and time is a precious resource that can be put into other things. The opportunity cost of a worker doing something in several hours with a machine doing it in two minutes is too high if we let the worker do it. So why not?

Doesn't it depend on the deltas. If automation replaces workers faster that they can find new work there will be involuntary unemployment.

Many factory workers in the US today already work alongside robots and fully automated manufacturing equipment. So what do they do that the robots can't do? Setup the robots, make adjustments to the machines, perform quality control, fix problems. Mostly tasks that require perception, judgement, or mental or physical flexibility and adaptiveness that machines have trouble replicating.

Also, don't forget about economies of scale. Even when a robot could do a job, it isn't cost effective to buy a $500k robot to do that job unless the robot will do that job over and over again millions of times. Some factory tasks only get repeated a few times per week or per month - it makes more sense to use a flexible adaptable human to do lots of these tasks rather than multiple less flexible robots to perform each of them and then stand idle waiting for the next opportunity to do the task again.

I agree that we will lose our jobs to machine to some extent, but only the jobs that most people do not enjoy doing anyways. If you look at the facts it is already happening. We now have machines performing duties that is productive and does not require unskilled labor. We have machine factories mass producing our food, clothing, correcting our work, and even transcribing the words we speak. Automation is designed to better our lifestyles and make it more efficient in a business sense. Duties that use to require unskilled human labor should be enforced with robots because from a business standpoint it simply makes sense. Robots that can perform work related duties would be a great investment due to its efficiency and reliability. I am sure most employers would appreciate automation at their businesses because robots don't show up late to work, don't take too long of a break, for "call in sick". On the other hand, they do break down, but the calculated down time is probably less than the hours required through human labor. Overall I think robots are great for business and efficiency. They should be implemented whenever possible and logical. That being said, jobs that do require cognitive thinking and making moral decisions should only be fulfill by human beings who are skilled and sensible. I feel that those jobs are safe from automation and will not be affected by drones or robots. Machines are made by men, and men is far from being perfect. Therefore, our jobs are still secured enough, where we should not feel threaten by machinery.

"As machines become more and more efficient and perfect, so it will become clear that imperfection is the greatness of man." —Ernst Fischer

Photography is one from the most important things of one's wedding day (besides your dress plus
your groom obviously). These bouquets could be "tipped" through the florist to your color more suitable for the theme.
They feel that many healthy dogs just like to graze being a cow, but sometimes
a puppy will eat grass and after that vomit it right back

If you aren't getting any flowers from a Peace Lily, try moving it to your location with
somewhat more light. Andrei and Natasha, owners of In Waters, happen to be guiding customers to select the top flowers more than 10 years.

I drop a comment when I appreciate a article on a website or I have something to add to the conversation.
Usually it is triggered by the passion displayed in the post I looked at.
And on this article Will you lose your job to a robot?.
I was actually excited enough to drop a thought ;-) I
actually do have a couple of questions for you if you don't mind.
Is it just me or does it look like like a
few of the responses look as if they are coming from brain dead
individuals? :-P And, if you are writing at additional social sites, I would
like to keep up with everything fresh you have to post.

Would you make a list every one of all your social pages like your linkedin profile, Facebook
page or twitter feed?

Comments for this post are closed