Robots and the Future of Unemployment(business.theatlantic.com)
business.theatlantic.com
Robots and the Future of Unemployment
http://business.theatlantic.com/2009/08/robots_and_the_future_of_unemployment.php
2 comments
The problem, however, is that somewhere along the way to that utopian society, there's a point where our traditional economy collapses because too many people are structurally unemployed. How do we handle that? What replaces our traditional economy? That's the interesting question, IMHO.
I think the trick is to automate the production/maintenance of our basic needs (food, housing, transportation(?), energy(?)). Seems like we will have a lot of free time in our future (making it less scarce, perhaps but still finite).
Good question. One possible, partial answer is through making it easier to gain the new skills needed. The unskilled labor market can die relatively peacefully so long as there is a proper glid path to skilled labor for those people being displaced.
There is a limit to how many people can be trained to do skilled labor though. We may like to imagine that everyone could, given the opportunity, do "skilled labor", but reality will be that automation will continue to slide the bar up as to how talented a person needs to be to be better than their automated replacement.
Right now, maybe less than 1 percent of the population can't do something better than a machine could. Fast forward 50 or 100 years, and maybe half the population genuinely cannot do anything that a machine couldn't do better.
Right now, maybe less than 1 percent of the population can't do something better than a machine could. Fast forward 50 or 100 years, and maybe half the population genuinely cannot do anything that a machine couldn't do better.
I both hope and sincerely think you are wrong. People improve with practice. Give them enough training to get that practice and they will improve.
Yes, the bar of what the machines can do is high, and rising, but there are a lot of things they will not be able to do for the forseeable future such as program, write, do most repairs, or pick heads of lettuce. At least for the immediate future, anything that is not repetitive and fixed in task will be hard to automate, but relatively easy to train a normal person for.
Yes, the bar of what the machines can do is high, and rising, but there are a lot of things they will not be able to do for the forseeable future such as program, write, do most repairs, or pick heads of lettuce. At least for the immediate future, anything that is not repetitive and fixed in task will be hard to automate, but relatively easy to train a normal person for.
do most repairs, or pick heads of lettuce
These are purely rule-based, mechnical tasks. I would say those are among the first to go into robot hands.
At least for the immediate future, anything that is not repetitive and fixed in task will be hard to automate, but relatively easy to train a normal person for.
Well, immediate future is a relative term. Looking at what Asimo and other robotics projects can do today I very much expect the robots of 2050 to be capable of performing around 90% of all jobs better than a human.
There simply isn't much creativity or "humanity" involved in most jobs.
These are purely rule-based, mechnical tasks. I would say those are among the first to go into robot hands.
At least for the immediate future, anything that is not repetitive and fixed in task will be hard to automate, but relatively easy to train a normal person for.
Well, immediate future is a relative term. Looking at what Asimo and other robotics projects can do today I very much expect the robots of 2050 to be capable of performing around 90% of all jobs better than a human.
There simply isn't much creativity or "humanity" involved in most jobs.
Sometime the only way to beat the machine is to join them.
So that mean becoming a cyborg and improving our brain with mechanical brain cells and exploiting our natural parallel computing capacity.
We just need a programming language and interface/translation machinery and firmware to take advantage of these natural CPUs.
So that mean becoming a cyborg and improving our brain with mechanical brain cells and exploiting our natural parallel computing capacity.
We just need a programming language and interface/translation machinery and firmware to take advantage of these natural CPUs.
In order to beat machines at being employed, the special cybernetic ability you would need is "working for free".
An interesting thought, and we may eventually move that way, but I suspect we are a long way off from that.
Repairs generally are not mechanical or rule based, depending on the complexity of the system involved of course. Picking heads of lettuce is, but there is no robot on the horizon capable of doing it without destroying them.
There is always scarcity of time and aesthetics.
Ironically, how much of the scarcity of time is caused by the need to work?
Neither of which can be provided by unskilled workers.
Huh? What is a maid but an unskilled laborer who saves me time?
And do not underestimate a good maid. She may not have a PhD but she is more skilled, faster or efficient than you imagine.
Certainly. You get good at doing something if you do it all day. But it isn't a requirement.
I don't expect this to be a problem within my lifetime. As we get more efficient we can easily reverse the trend toward 2-income families, and then start shortening the work week. Who could say no to a 32 hour week, or even shorter?
At some point we may have to switch to an alternative economic system not based on labor (perhaps Cory Doctorow's "Whuffie", and I recommend "Down and Out in the Magic Kingdom" for anyone interested), but that will be a long time from now and I suspect by then the choices will be more obvious and incremental.
At some point we may have to switch to an alternative economic system not based on labor (perhaps Cory Doctorow's "Whuffie", and I recommend "Down and Out in the Magic Kingdom" for anyone interested), but that will be a long time from now and I suspect by then the choices will be more obvious and incremental.
The problem is that today's unskilled workers aren't going to be the robot owners of the future. Increases in productivity will allow us to produce more with less work, but access to those gains won't be equal. You can't just start making everyone work less and have the numbers magically work out to support everyone. You would have to force those who own the means of production to share their wealth with those who don't.
How old are you? If you're less than 40, I wouldn't be so sure about this not being a problem in your lifetime.
How old are you? If you're less than 40, I wouldn't be so sure about this not being a problem in your lifetime.
"today's unskilled workers aren't going to be the robot owners of the future"
Careful with your crystal ball there; you might end up sounding like the people who predicted similar things about computers. Robots will become very cheap in the future; they won't be exclusive to the "bourgeoisie".
"You can't just start making everyone work less"
Who said anything about making? That's far too authoritarian for my liking. People will choose to work less when they can achieve the lifestyle they want with less work. It will take a while for this to become socially acceptable, but I think it will happen.
I am less than 40, and I would be happy indeed if this became a problem during my lifetime, but I'm still skeptical.
Careful with your crystal ball there; you might end up sounding like the people who predicted similar things about computers. Robots will become very cheap in the future; they won't be exclusive to the "bourgeoisie".
"You can't just start making everyone work less"
Who said anything about making? That's far too authoritarian for my liking. People will choose to work less when they can achieve the lifestyle they want with less work. It will take a while for this to become socially acceptable, but I think it will happen.
I am less than 40, and I would be happy indeed if this became a problem during my lifetime, but I'm still skeptical.
The 'price' of robots as well as computers could be greatly affected by increased scarcity in materials. A lot of the base materials in today's electronics aren't exactly abundant.
I think this is one of the problems with a lot of technological 'forward-thinking'. Issues like this tend to be swept under the rug as if supplies of crucial building materials were not a concern at all. I would say it's a very real possibility that we could run out of a particular material with no replacement/alternative, especially as the march of progress causes us to consume more and more and more of these source materials.
I think this is one of the problems with a lot of technological 'forward-thinking'. Issues like this tend to be swept under the rug as if supplies of crucial building materials were not a concern at all. I would say it's a very real possibility that we could run out of a particular material with no replacement/alternative, especially as the march of progress causes us to consume more and more and more of these source materials.
As nanotech advances we will gain the ability to replace more and more rare elements with common ones. In the far future there's no reason why our technology couldn't be based entirely on nanoscale structures made of only the cheapest, most abundant elements. After all, nature does it just fine.
msluyter said it better than I did above. The problem is the path from today to ubiquitous cheap automation of everything. In between, we have automation that is expensive, but cheaper than humans. What will happen during that gap, and how will society make it past that point to where everyone can enjoy such advances?
The endgame may be as benign as you're suggesting, but I don't think there's any way we'll get there without complications, if we get there at all.
The endgame may be as benign as you're suggesting, but I don't think there's any way we'll get there without complications, if we get there at all.
"we [will] have automation that is expensive, but cheaper than humans."
I don't think that's true. The expensive part of robots is the hardware. I think we will have hardware able to replace humans long before we have the software. ASIMO's body is probably capable enough to do many tasks today. The hardware will become cheap long before the software becomes capable enough to replace humans in most jobs, and the software will improve gradually. People will have time to adapt.
Furthermore, this argument seems based on the premise that there's a huge underclass of people who can't possibly be trained in a skill once robots take the manual labor away, and that's an elitist view I simply don't agree with. America is already a country full of people who shun manual labor; only immigration fills the lowest level of jobs. Americans are not genetically superior to the immigrants who do our manual labor; we just have better opportunities. As robots make us more productive those opportunities will be offered to more and more people, and the number of workers willing to do manual labor will naturally dwindle.
I don't think that's true. The expensive part of robots is the hardware. I think we will have hardware able to replace humans long before we have the software. ASIMO's body is probably capable enough to do many tasks today. The hardware will become cheap long before the software becomes capable enough to replace humans in most jobs, and the software will improve gradually. People will have time to adapt.
Furthermore, this argument seems based on the premise that there's a huge underclass of people who can't possibly be trained in a skill once robots take the manual labor away, and that's an elitist view I simply don't agree with. America is already a country full of people who shun manual labor; only immigration fills the lowest level of jobs. Americans are not genetically superior to the immigrants who do our manual labor; we just have better opportunities. As robots make us more productive those opportunities will be offered to more and more people, and the number of workers willing to do manual labor will naturally dwindle.
"this argument seems based on the premise that there's a huge underclass of people who can't possibly be trained in a skill once robots take the manual labor away"
Who is going to pay for that training? I don't see any way for people who are currently unskilled laborers to make it through the gap when their current work is no longer available without any assistance. Moreover, plenty of skilled labor will be automated in the future. Nursing, for instance, requires a significant amount of knowledge to do competently, but most of the tasks nurses (as well as family practitioners) do will probably be automated in the next 40 years. Accounting? Skilled profession, but it could probably be automated. Anything that doesn't require creativity seems capable of being automated to me, even without drastic, unforeseen, internet-like advancements in technology. How are all of these people going to be retrained? What will they be retrained in?
"The hardware will become cheap long before the software becomes capable enough to replace humans in most jobs, and the software will improve gradually. People will have time to adapt."
I don't think so. Most of the problems that need to be solved to replace various forms of human labor are fairly similar. When the necessary technology is developed, several industries will start to automate simultaneously. Tens of thousands of jobs will be permanently lost annually. Even at a more gradual rate, I doubt there will be enough work to absorb all of those people.
To recap:
1) What jobs will people switch to?
2) How will there be enough of them?
3) How will people afford retraining when plenty of people barely have enough money to save today?
4) How will they retrain fast enough? Most of the jobs that seem difficult to automate to me require a college education, which many people aren't currently prepared enough for.
I think things will end up being okay, but it will take massive government intervention.
Who is going to pay for that training? I don't see any way for people who are currently unskilled laborers to make it through the gap when their current work is no longer available without any assistance. Moreover, plenty of skilled labor will be automated in the future. Nursing, for instance, requires a significant amount of knowledge to do competently, but most of the tasks nurses (as well as family practitioners) do will probably be automated in the next 40 years. Accounting? Skilled profession, but it could probably be automated. Anything that doesn't require creativity seems capable of being automated to me, even without drastic, unforeseen, internet-like advancements in technology. How are all of these people going to be retrained? What will they be retrained in?
"The hardware will become cheap long before the software becomes capable enough to replace humans in most jobs, and the software will improve gradually. People will have time to adapt."
I don't think so. Most of the problems that need to be solved to replace various forms of human labor are fairly similar. When the necessary technology is developed, several industries will start to automate simultaneously. Tens of thousands of jobs will be permanently lost annually. Even at a more gradual rate, I doubt there will be enough work to absorb all of those people.
To recap:
1) What jobs will people switch to?
2) How will there be enough of them?
3) How will people afford retraining when plenty of people barely have enough money to save today?
4) How will they retrain fast enough? Most of the jobs that seem difficult to automate to me require a college education, which many people aren't currently prepared enough for.
I think things will end up being okay, but it will take massive government intervention.
Your questions are about resource allocation, and the answer as usual is free markets. An argument about the efficacy of free markets vs. central planning would be offtopic, so let's just agree to disagree on this one.
If there are no scarcity, then there are no need for the existence of jobs and we all live happily in a utopian society.