What Jobs Will the Robots Take?
theatlantic.com
theatlantic.com
It's true that we're still below 50% unemployment, despite tremendous increases in productivity and automation over the preceding centuries. The standard explanation for this phenomenon is that technology eliminates certain jobs, but also creates new jobs that couldn't have existed before.
That pattern has indeed held true (in the long run) since the Industrial Revolution began. But crucially, we can't be sure the pattern will hold true forever. There is no proof that it constitutes an inviolable law of economics. It could very well be that technology can result in a major, permanent, net loss of employment, but just hasn't yet.
So what could change? What might break the pattern that has held true thus far?
Here's one hypothesis. There is an inflection point--a threshold where automation has become sufficiently advanced. Beyond this threshold, our capacity to automate processes is faster than the expansion of demand for those processes. Whereas, before that threshold, new demand could only be satisfied by human labor, owing to society's inability to immediately automate the requisite processes.
If this hypothesis is true, then once we cross the threshold, we should see a long-term decline in employment. That's not to say no new jobs will ever be created after reaching the threshold. No doubt they will. Even so, more jobs would be lost than created, resulting in a net decline.
If we take this idea to its logical extreme, we can envision a world in which every process and occupation can be trivially automated. If we assume such automation always costs less than subsistence wages[2], then presumably any given firm will choose to automate all of its processes rather than employ people. The result being 0% employment.
[1] http://en.wikipedia.org/wiki/Technological_unemployment
[2] This is a big assumption. There could be a dynamic equilibrium where market forces balance the cost of automation against the cost of living. Thus the TCO of a robot would in many cases be very close to the cost of a human employee. Processes would then be divided amongst humans and robots in accordance with the relative costs of automating each process.
When the animal's advantage over humans is sheer strength and endurance, machines usually work better. But when their advantages are senses and intelligence, they still outperform machines.
If the robots don't take over in the meantime.
The problem with those types of changes right now is that the world is still highly meritocratic. I know a lot of people would scoff at that idea, but they end up riding the bus to work in the morning thinking that the bus driver might be worth a comfortable living wage but definitely doesn't deserve their college-educated salary because they worked harder for it. And they think that they are better at their job than Joe CubicleMate, and they think they can run the company better than Director Jim, and they think that their friend's step son is a worthless piece of garbage that shouldn't be allowed to collect welfare unless he checks himself into rehab first.
And as such, the people who fight for reduced inequality may have a point about billionaires and such, but they don't want so much equality that they can't think they are better than their neighbor. If they want sweeping change, they have to change themselves. Until then, the only meaningful change can be targeted at the extremes.
I think we are just seeing the start of everyone enjoying robotic efficiency. 10-15 years ago a rapid prototype machine cost in $30-60k, for small objects. Last year the news was filled with 3D-printers in the range of $500+.
Alternatively, one can imagine an extremely hierarchical society where the fruits of automation are enjoyed only be a few, and the rest of the population is treated as a burden or a threat.
My hypothesis implies no value judgment about the future of human labor. Maybe 0% employment will be good, maybe it will be bad. I'm merely imagining how we could get to 0%, not what that will mean for us.
A notion of a minimum guaranteed income would be necessary in addition to all the robotic automation.
I meant Basic Income.
You probably have. There's certainly some confusion about, which is why I make a habit of noting the distinction - so we are at least clear which we're talking about.
The first stage of automation/industrialization used crude technology in the simplest way, for the brute force it could provide. At first you needed your factory to be close to a river so you could use a water wheel for power, and you needed a team of oxen to plow your field. In both of these cases all you needed was linear power greater than a human could provide. Once we figured out steam power and the ability to create simple linear motion in a piston, we didn't have to build our factories on rivers anymore, and thus rivers became "unemployed" in the sense that we no longer had to "employ" them to create things. Same with animals such as oxen that just provided simple brute force pulling/pushing something. Think about the relative ubiquity of the term "ox" in literature from the past and then think about the last time you actually saw an ox. Clearly animals have been dealing with rampant unemployment for quite a while.
The second stage of industrialization moved up the food chain and focused on automating things that were repetitive but required more complicated movements and dexterity. The sort of thing that we used to rely on assembly lines to do and such. At the previous level we replaced providers of brute labor with no dexterity. At this level we replaced providers of fine but repetitive labor, and still unemployment wasn't that bad because people could still move up the value chain and do finer labor that required decision making.
The third stage of industrialization focused on replacing fine labor that required decision making, where now you have a giant robotic arm that handles all of the spot welding, painting, and assembly of a car for example. If you don't think that this has resulted in a rise in unemployment then you haven't travelled through the American mid-west recently. Because at this third stage we are starting to replace people who are then unable to move up the value chain into labor based solely on analysis and reasoning, both because there are fewer of these jobs and because not everyone can do them.
We are now in the fourth stage of industrialization and it is here that we are really going to come to grips with massive automation induced unemployment, because automation has moved sufficiently up the value chain, from brute force, to fine but repetitive labor, to labor requiring limited decision making, and finally to labor requiring reasoning as its primary component. Look outside your window, look at all of the cabs driving down the street and realize that once a computer can drive a car, and it will be able to in a decade or less, all of those jobs will be obsolete. And if a computer can drive a car it can probably analyze an x-ray or mri, it can translate a document via ocr, it can cook your fast food, it can write a sports recap, and it can balance the books. Once this level of the value chain is displaced where do these people go? Into jobs requiring creative work? Jobs requiring design or analytical work? How many of those will there be, and is that a realistic alternative for everyone?
No, the Luddite Fallacy is just a convenient way to avoid thinking about the implications of the technology we build and what is happening to people as we work our way higher and higher up the value chain, crowding those at the bottom either up into the next level or out of the employment market. As as we start to reach levels which are practically the limit for larger and larger amounts of people we can expect to see greater and greater unemployment unless we figure out a way to grapple with these repercussions.
* Think about the relative ubiquity of the term "ox" in literature from the past and then think about the last time you actually saw an ox.*
I wonder what the global population of oxen (or any beast of burden) looks like now compared to when they were at their highest level of "employment."
I think that would give us a very interesting existential metric if, as economics does now, we view human labor and humans in general as tools towards some ends.
There was a time when 2% of the old U3 was full employment. Then there was 4% of the new U3 being full employment, but women were working so the total labor-force participation was higher. Now the USA (for which I'm aware of the numbers) has ~6.7% new new U3 unemployment with the lowest total labor-force participation in decades.
You also need to look at total work hours (ie: mean work hours multiplied by size of the active labor force), which have universally gone down with the advance of technology.
When you look at things this way, technology has already been "destroying jobs" for centuries. That's why we don't have women and children spending 60 hours/week in coal mines! It's why we now consider it normal for the elderly to retire at 65 instead of working until they die, and for children to complete 12 years of compulsory education (with the middle classes putting in another 4 years of economically compelled education at college/trade school/university).
Technology does destroy jobs, and that's a great thing. It is the means by which we have afforded humans the time and means to be human.
This is an important distinction. So far, technology has not made it impossible for most members of society to earn subsistence wages. But what if that changed? That, to me, is the fascinating question.
I believe it just doesn't happen because, although automation can increase productivity and lower costs long-term, it has a high initial investment compared to human labor for many non-critical tasks and the pressure everywhere is for profits now. Plus, human labor is cheap, and will keep being cheap as long as population is increasing, so this is a force that will push more and more against automation the more we automate. The critical threshold between human labor -> automation will happen only once (and if) human population growth turns negative, turning wages up.
As for call centers, they already are heavily automated, and have been for some time. Every touchtone menu you're forced to deal with is evidence of this fact. Nonetheless, there comes a point when touchtone menus can't solve the customer's problem. (I for one only call a company if I'm facing an unusual situation, so the touchtone menu never solves my problem.) Because artificial intelligence cannot yet listen to a person's problems and devise a solution, human operators are still needed.
Basically, any job that can be automated will be automated, as soon as it's feasible. This has been going on for millennia, ever since the first genius learned to herd goats rather than hunt them. People just have a very narrow definition of "automate". We don't plow our fields or harvest our crops by hand anymore, leading to massive reductions in farm employment over the past 100 years. But we don't think of tractors as "robots", so it gets ignored.
I have this saying that drives my whole career's thinking... "Have computers do what computers do well, have humans do what humans do well." Computers are wonderful about performing precision tasks perfectly, at incredible speeds. But they are terrible at making decisions in the face of incomplete data. They can be programmed to sort of simulate it, but it's not the same. Meanwhile, humans are incredible at making correct decisions using incomplete or incorrect data, but it's almost impossible to get them to do the same thing the same way twice in a row.
So the jobs that will last are the ones that require decision-making in the face of incomplete/incorrect data. The jobs that will be automated are the precision, repeatable tasks.
Humans have developed a large skillset through evolution, and it is a skillset that includes highly sophisticated heuristics for problem solving, communication, and physical interaction. Even 15 years ago, the top Ops Research and Artifical Intelligence researchers and their computer models couldn't outperform an ordinary truck driver on the Traveling Salesman problem on average. Now, there isn't a human in the world that can outperform their models. 15 years ago, there wasn't a single robot in the world that could compete with an ordinary human in the task of walking. Now we have Boston Dynamics.
The idea that humans are special at making decisions with incomplete data is short sighted. It assumes that computers can't improve at the task. They are certainly a far ways away from making general purpose decisions with incomplete data, but there are several current systems out there that make very sophisticated (even unparalleled) decisions in specialized domains, and they do it with incomplete data.
That isn't to say that your saying is wrong. It is just to say that it isn't a given that something that humans do comparatively well today may not be that way in the future.
All of these are problems with complete information.
Chess is a solved game, however you can't analyze the whole problem quickly enough to cover it all.
Driving a car, a computer can have access to all relevant data--far more than a human. Sure, new data can be introduced, but the computer can pick it up and handle it as quickly as a human.
Poker is a better example of a problem with incomplete information--and computers do NOT excel at it. They do play excellent heads-up limit, which is more-or-less a solved game. They play much-less-excellent heads up no-limit, and they certainly don't approach new situations in a novel manner. The same goes for full table games.
I would expect computers to be very good at all the tasks you mentioned, and mediocre (but trainable) at those like poker. I expect that this, too, will change in the future, when we manage to invent better ways to handle incomplete information, especially if there's a rapid feedback loop.
Air Traffic Control employs certain models of behavior of planes and air flows, which are not exact. Understanding of situation could be different among participants, so are actions. Driving Cars doesn't have information about all surroundings (say, behind the bush) - and definitely not about plans of other agents in the area. Logistics have to deal with uncertainties of transportation - you can't be sure a particular truck going through Oregon won't be delayed by a storm. Chess isn't a solved problem (checkers is).
Pick a harder problem - the game of Go. Computers have only recently been able to beat a reasonably talented child or inexperienced adult at the game. When the data set becomes too large to brute-force (Go has many more possibilities than chess), computers can't hold a candle to human intuition.
Sure, it's possible that at some point, computers will be able to produce better results than humans at some surprising real-world tasks. But that doesn't mean computers will do them well. Doing it well versus doing it very fast are two entirely different things. If you can't understand the distinction, you can't understand where humans will continue to outperform computers.
Heck, I'm fairly certain that every master game ever recorded was shoved into Deep Blue and other high-end chess programs.
The crazy thing about human heuristics is that they can be learned. Deep Blue was created more than 15 years ago, but in the time since, Houdini can do even better than Deep Blue, and it can do it on a dual core laptop. How? By learning heuristics from Grand Masters.
If there's a problem that fits the description it's medical diagnosis , and it's been known for quite some time that machines are better at this.
There is a crazy amount of wall square footage in offices/commercial/malls/hallways etc where a robot traveling on a mobile linear beam can totally do the work.
As somebody that spend some amount of time every week hiring, one of the biggest issues is not automation but the workforce itself.
Finding people that are willing to show up and do jobs is just a major problem.
Companies want to hire but the candidates that show up at the door can often be barely employable.
All the millenials want work from home jobs with executive pay.
The idea of starting in the back sweeping the floors (no matter your education but to show work ethic and flexibility) is gone.
The working world will get harder over the years. The "raised like a delicate flower" to "love their job" are in serious trouble in the future.
Putting food on the table takes time and effort. It's called "work" not "play" for a reason.
That company has no reason to promote people beyond the cleaning floor job as that is their main business & curtailing HR costs is probably their biggest driver to profit. So in the best case they keep workers who can't find better jobs at low cost at the same position for years, and in the worst case they actually "sub-contract" those positions out to companies that exploit illegal immigrants, who have a near limitless supply of under payed workers to cycle through.
So forgive me, if instead of reading your comment as a jab against overly entitled "millenials" and instead read it for what it is, companies reaping what they sow. If companies wanted employees who will start on the ground floor and work up, they should foster an environment where that can actually happen rather than treating their employees with a cavalier mercenary attitude, and getting huffy when their employee pool reacts in kind.
Some random service doesn;t take any amount of care of interest in the job.
You are far better off to pay somebody to do it and be invested in the task with a sense of pride then to employ a random cleaning company and whatever worker they happen to send that week.
What I do find is that employee's think hiring is like being on a talent show.
Clearly you have never done any hiring. Once you get out of tech you'll be lucky if 50% of workers show up for work the day after they have been hired.
Fully 10% of workers will be hired, you will arrange a day to start work and they won;t even show up. You'll never hear from them again.
Companies can spend time training people over and over. Millenials expect to be promoted in 6 months and be constantly feted for their work.
Hopefully one day you can own a company and spend your time training workers to have them leave when they become competent.
I've never been responsible for hiring outside of the technology sector, that said, I've been responsible for hiring, and have hired across a variety of age groups and have noticed no trend among any of them for who works harder/more efficiently/etc. Employee retention is most certainly a hard job, but it isn't because of the age of my employees, it's because I strive to hire really good ones.
For what it's worth, I put this question to my wife who hires in a non-tech small business setting, and she has also never experienced specific problems with millennials that don't exist in other age groups.
Perhaps it is your hiring practices that are at issue not who you are hiring?
I can get driven to work by a nice guy in a clean car for $11 instead of a surly driver in a grotty $25 taxi (who probably has a great work ethic), because the UberX driver is trying to support his unjob lifestyle instead of being good at being an employee.
Do you really not perceive the logical contradiction here?
How to fiddling through the coming automation without unrest the society is the only question, before our next generation all live like the 1% of today's first world someday in the future.
Automation, not war, should be the only force to propel civilization and living standard. Instead of minimum wage, minimum welfare would be a feasible answer to the future.
What if the super rich are the ones who get the most advantage out of the automation and use it to keep the rest of the poor under control, I'm talking Ghetto's with automated riot control.
I'd think such things where unthinkable but if you look at the history of a company like Nestle nothing would surprise me.
Fully automated enterprises will have huge incentives to avoid the taxing power of large states. They don't need a workforce, remember. A robotized manufacturer would set up in some small, possibly autocratic state.
How are the unemployed masses going to get their subsidies when the taxable productivity is on some small island in the pacific?
Around 22 minutes in, he talks about whether labor and capital (machines) are complements or substitutes.
I'll summarize his argument for the time-constrained: There are many tasks that need to be done in an economy. Machines are more efficient at some of these tasks, humans at others. Over time, machines have been able to do a greater portion of these tasks efficiently.
Today, machines are mostly complements. They help humans do tasks more efficiently. But as machines get better, they're able to do a larger portion of tasks. At some point, they'll start substituting for people. This has happened for some basic jobs in the past, but not to the extent we'll soon see.
On the bright side, the limiting factor in most of our economy today is people. Machines can be mass-produced, but raising and training people is hard. Machines can augment people, but run into diminishing returns. Having two monitors won't make you twice as productive, and two keyboards won't let you type twice as fast. If we can do all the tasks in our economy without people, we can get insane growth rates in GDP.
Unfortunately, with no people in the economy, everyone would be out of work. Only the machine owners (or investors) would benefit. Everyone else would be dependent on social programs or charity.
In Robin's example, AI is the machine that can do all tasks. But we don't need to build such amazing machines before putting a lot of people out of work. Most jobs are not AI-complete.
In SU, everyone works for government, and salaries aren't too different. So, with greater and greater employment of robotics, salaries can gradually rise, and working hours gradually decrease - and the whole society gets benefits more or less at the same time. More money supports demand for products - to some extent - and more time allow to improve the quality of life - people devote more time to kids, to arts, to each other, to interesting ideas etc.
I didn't watch the video yet, but one thing often missed in the whole automation debate is that for a given task, you don't need to replace 100% of the people to cause great, painful economic shifts. Just complementary-automating things enough so that 5 people can do the work that 10 people used to do is incredibly impactful.
Thanks, Halting Problem!
As most things, I really doubt "this time" is any different. I have no numbers / research to back it up but I'd bet pre-steam loom UK, every agricultural country, and many others faced similar job elimination with advent of mechanization. Not as fast as 20 years though.
That's awesome. That means I can then automate answering the phone and determining if it's a telemarketer on the other end (in the case of an unknown number).