A thought experiment: Imagine a pre-singularity world with near-intelligent anthropomorphic robots which can do most manual tasks - including resource mining, food production and building more of themselves. Imagine most menial white collar jobs have been replaced by clever software written by people on this site. There are still some jobs for humans to do - in game design, fashion, service, music, etc. But any way you look at it, there are not enough jobs for everyone to do. Lets call this the "high unemployment future."
Currently we exist in a "low unemployment present" but if you believe a "high unemployment future" is possible, you should reasonable believe we're smoothly approaching it. So if you think my thought experiment can come to pass - we need to be very concerned.
If you're going to bring out the tired argument about how the cotton gin didn't cause mass unemployment, instead I would challenge you to paint a picture of a far off future where most jobs haven't evaporated.
> So if you think my thought experiment can come to pass - we need to be very concerned.
We only need to be concerned if we make the assumption that societal expectations won't adjust as well. "Hard work" is seen as a virtue; someone who takes government assistance is pitied at best, considered "parasitic" at worst.
If we take those views into your world it will be a disaster. But instead the real question should be "why does everyone need to have a job?" If we provide for the needs of people displaced by the new machines, we'd free people up to pursue arts, philosophy, or even simple leisure. We'd be living in Bertrand Russell's paradise of 'idleness'.
Wealth is imaginary points. Standard of living is the true measure. If most folks didn't have any money at all, but food housing and entertainment were a civil right, then who's to complain or revolt?
And this is the real question (not the rhetorical "why does everyone need to have a job?"). Can people adjust culturally to the idea that guaranteed income/wealth/whatever could be a good thing, and lead to a pursuit of creativity and innovation, not merely hedonistic, fleeting pleasures?
I.e., can we as a society accept the possibility, and make it a reality, that a guaranteed income (in whatever form it takes), frees us from mindless grunt work and inspires us to pursue our dreams, or do we force those whose jobs have disappeared to suffer poverty because there is no 'need' for them?
Exactly. Consider that automation would be pointless if replacing N foo-craftsmen with foo-machines necessitated the hiring of N equally expensive foo-machine repairmen and assemblers.
Automation reduces jobs, wages, or both.
Nevertheless, we should embrace it. Increasingly strong social safely nets can be used to take up the slack.
I would challenge you to paint a picture of a far off future where most jobs haven't evaporated
200 years ago, nobody would have guessed that there would be millions of computer programmers employed in the world today, because the rise of computers couldn't be predicted. It's likely that in an era where basic needs are provided by robots, people are free to come up with new inventions that lead to the creation of new industries that don't exist today.
It's been well proven adding more developers to a single project has extremely diminished returns, so what you're saying is that there will be a free market demand for a huge amount of different software projects. While it's true that software is eating the world and thus we're seeing a net increase in projects started, we're also seeing a lot of consolidation in the more established software verticals.
Once we've digitized all the existing flows of information present in non-software industries, I would expect the number of developers employeed by a free market to begin to shrink.
so what you're saying is that there will be a free market demand for a huge amount of different software projects
That's not what I'm saying. I'm saying that there are currently millions of people of working in an industry that was unheard of a century ago because the modern day computer was invented. There are millions of people who have jobs today related to cars (manufacturing, sales, repair, driving, etc.) because the car was invented. There could be some new invention we haven't thought of today that spawns an industry that will employ millions.
The answer is going to be colonizing other planets/moons. As more manual and semi-skilled labor is outsourced to AI and robots in "modern society", the outskirts of humanity will degenerate into a frontier state because human labor will be cheaper than building and maintaining this automated future.
And I just summarized Firefly. Start learning Chinese.
Ray Kurzweil talks about this in "The Age of Spiritual Machines". It probably comes down to human morality (as with just about everything) with regards to whether this singularity is a good thing or a bad thing.
I purposely poised this pre-singularity to not derail the conversation, but I think the same issues around automation and the morality of wealth redistribution exist.
The simple answer to this, is that most liberal arts-oriented positions will morph into sales and marketing, and technical-oriented ones will morph into technical services (software engineering, maintenance). The world isn't going start starving, or somehow skyrocket into 99% unemployment rates. While there will be a "new norm" of higher unemployment rates for a while, that will quickly subside and people will find jobs, etc. We're just experiencing a cultural shift to a different type of business format. Call it the post-technology era.
Another report: As more jobs are getting automated, it opens up "new" job opportunities in new fields that are coming up. So net effect: not much. Point being that for centuries, we have been automating things that were done by humans at some point in time. Not much different even today.
Take one example of cellphones. It probably eliminated many "jobs" for people working in landline industries in the last decade or so BUT it probably added a lot more jobs in the telecom industry in general.
EDIT: based on some replies, I am not dismissing the fact completely that automation can have impacts on jobs. They definitely do but the point is that the number of jobs themselves do not dimish. It is the "type" of job that is changed which could surely be a problem for the displaced workers because they might not have the new skillset/experience ready to be employed.
That is (very likely) true in the long run, but it would be foolish to overlook the short-term effects that can be devastating for many individuals.
The weavers faced real problems, had a real existential crisis and struggled with real poverty and hunger for many decades. Yes, the industrial revolution brought a tremendous increase in wealth and living standards to everyone, but only eventually to everyone.
It is tremendously important to take care of the people who are cast by the wayside by those changes.
I agree with this. Not sure how viable, but it'd be nice to see some kind of federal program which allowed for a 5-year transition program for people where their industries collapse precipitously (by some definition) and are guaranteed 100% their wages but must study/prepare for a new career.
Not sure how to avoid the occasional dilly-dallier who thought they'd make a good chef, but maybe we'd just have to put up with that fact.
To address the dilly dally issue you would more than likely have to only give the transitional aid to someone involved in Science, Engineering, and Math fields, much like current STEM¹ scholarships.
How is it true in the long run when we have robots that can walk talk and think as well as our bottom 40% but take one tenth the cost in food / shelter?
To be honest, I’m probably not as sure about the long term effects as I might have sounded. Looking into the past I see almost universally massively positive long-term effects. A few centuries ago almost everyone worked in farming, now almost nobody does, all those jobs have been automated away – but it didn’t matter in the long run because more people could work first in industry, then in services.
Maybe future transformations will work out similarly. However, even if they don’t many solutions for the short-term problems can easily be made to work as solutions for possible long-term problems.
Automation eliminates jobs - that's the whole point. One robot replaces 20 workers and you only need 1 person per every 20 robots to manage them.
Sure, you've got jobs in robot building and such but when a robot shop with 20 employees turns out 200 robots in one year and displaces 4000 workers...
You need to go to the next level. People automate things not to eliminate jobs (that's not the whole point), but to save money. So now the product is cheaper.
So the buyers of the product have extra money to spend - which opens whole new fields of products and services for them, which will (at first) be run by humans and have jobs.
It's a cycle that repeated many many times in history. It's just happening faster now. So if you want to always be employable (and you are not particularly skilled) be flexible - don't get stuck in a single field, but be willing to constantly learn new ones. (This is hard for some people, but some extra funding for adult vocational school should help.)
The customers as a whole don't have more money, because half of them just got automated out of their jobs. Maybe enough new jobs come around that it isn't a long term problem this time, but there will come a time when essentially every job could be done by a computer. What then?
If that point in time comes, and the computers still "obey" humans, yet still have enough control so that there are no wars between different computer groups, then nobody will have a job and we'll just ask robots to make us stuff and they'll mine other planets and make stuff for us, and we'll live in outerspace in spaceships and relax a lot.
That's not really how things end up working, though; if you're living hand to mouth you end up spending all you make. If you're wealthy, you save much of your wealth.
Yes, people have more money to spend, but no, trickle down economics doesn't really work out for the jobless.
The problem with this logic is that it assumes a zero-sum game, and a constant demand.
Example: If we have 10 guys making hydrogen cells, and all the sudden we can replace them with one robot and one guy, and that takes an unviable industry to a viable one, well then they are going to need a lot more robots and guys, a lot more than the 10 they previously had.
Except that never ended up happening. I live in a rust-belt city which was booming during most of the 1900's and ended up dying off when the 60's and 70's came around.
Today I was at lunch in what's left of our downtown district, and there were old photographs on the wall of business that used to be in the area (mostly manufacturing). I was amazed that some of the photographs showed huge rooms of women sitting at work tables assembling parts, folding boxes, and packing items.
That stuff is all gone. I could imagine a handful of machines replacing that group of people. When you go from a city having 50,000 factory workers to the 4,000 or so that we have now, it's a huge deal.
The thing about economics though is it's about the average case, not the individual case. There's always going to be winners and losers, but as long as the winners beat out the losers things work okay. I think a future dedicated towards decentralized/distributed micro-economies relying on coordination and communication would be ideal, because it could prevent exactly what your town experienced. Automation in conjunction with networking could make your town a viable manufacturing micro-hub, for example, instead of those sorts of things being shipped off to Shenzhen. I'd argue The reason china is such a power house is precisely because we don't have the automation necessary to compete with them.
The problem is that society's support system is nearly non-existent. If people have to do less work to create the same product, they should be awarded with more time away from work to enjoy their life. But without some ownership over the products of one's own labor, this is not an option. Rather than having a system where technology improves the lifestyle of the middle class, our status quo is a system where technology battles against the worker and puts her at risk of poverty.
"As more jobs are getting automated, it opens up "new" job opportunities in new fields that are coming up."
That's what I've been taught in school, but I'm not sure it's that simple. For one thing, I would say the jobs getting automated are less qualified, manual jobs. And new jobs opening up require more skills. But can we expect everyone to perform a highly skilled or intellectual job?
It seems to me that either we have to accept that a proportion of the population doesn't work, or we have subsidy manual jobs to compete with developing countries with cheap labor.
Either way, the ones benefiting from automation have to pay for the unskilled to survive.
> That's what I've been taught in school, but I'm not sure it's that simple. For one thing, I would say the jobs getting automated are less qualified, manual jobs. And new jobs opening up require more skills. But can we expect everyone to perform a highly skilled or intellectual job?
More importantly, if we were replacing less-skilled jobs that are being automated with more-skilled, higher-paying jobs at a 1:1 ratio, automation would be driving up production costs rather than controlling them, and there'd be no motivation to do it.
I think you are confused in the same way that people are confused who think that economic pie is fixed size. The point of automation is it would expand the overall economic pie so that those more skilled higher paying jobs would be necessary to do something that would not happen otherwise - and that something, whatever it is, would be useful and profitable
> The point of automation is it would expand the overall economic pie so that those more skilled higher paying jobs would be necessary to do something that would not happen otherwise
No, the point -- i.e., why firms adopt automation -- is to cut the production costs of the firms adopting automation.
The hoped for social benefit -- hoped for largely by people who aren't the ones adopting automation -- is of an expanded overall economy that not only provides aggregate growth but also distributes that growth in a nice manner so that not only does the mean income improve, but the total share of population productively employed stays the same or improves, and everyone does better.
There's no real reason to expect that is either likely as the near-term result of automation, or necessary as its long-term result as a result of pure market forces. And it is certainly not the point of automation from the point of view of the firms doing the automation.
the point is that the number of jobs themselves do not dimish
I agree there. Think of the jobs that have been lost over the years as stores have closed due to competition from the internet. During my lifetime, I've seen independent bookstores merge and disappear, and even the mega-chains go away (e.g. Borders). Music stores - gone (Tower Records, Wherehouse, Sam Goody). Video rental - gone (Blockbuser). Big box electronics - going (Circuit City, Good Guys, Tweeter).
Despite all of this, unemployment stayed relatively constant, aside from the spike during the recent economic downturn (and these jobs were not lost as a result of automation).
Those are all retail jobs, so they're in the same industry. We probably have some new stores for products that didn't exist. We also have more warehouse jobs for the online merchants that replaced them.
Just thoughts, I don't know the numbers or anything.
This has been discussed on HN many many times, and I could probably just dredge up old posts and cut and paste them, but rather that re-iterate those, let me focus on the following: how would one falsify your thesis? What would it take before we could confidently agree that automation was not creating an equivalent number of jobs as it was creating?
I guess it feels like there were other manual labor jobs to move to back in that day and age.
But if nearly every manual labor and semi-skilled labor job is going to be automated, I don't know what those people will do. The people who dig ditches or drive trucks today are not going to be happy or excel in office jobs.
Is there a difference between machines designed to perform a narrow set of very specialized tasks and machines designed to perform a broad set of very generalized tasks?
If it's not different this time, it will eventually be. That's the goal.
In the long run, I don't think so. You could argue change is happening much faster today, and people need to adapt to new jobs and skills faster, but on the other hand, there are also many more opportunities to choose from these days.
I think a difference is perhaps that the time investment required to adapt to the old automation was a lot less than that required to adapt to this one, while human lifespan hasn't categorically changed (the most dramatic shifts in life expectancy in the last century and a half have come in large part from infant mortality improvements). Working in a factory in the Industrial Revolution didn't require much training/education at all. Post-WWII, the shift in job-skills to essentially requiring literacy was dealth with by a society that tried to ensure that its citizens were educated enough, with great results.
This time, starting around the 70s and 80s, we entirely failed to educate our populace to the point where the majority can be fully-productive members of the new job marketplace. It's becoming conventional wisdom that having only a high school degree (as a crude heuristic for skill level) is not enough to be productive economically, and this is becoming increasingly true of having a college degree where you don't learn anything useful. In every cycle of improved tools/higher skill bar, there are casualties that weren't able to improve their productivity fast enough; this is one of the uncommon cases where (in the US, at least), "people not keeping up" is a systemic issue instead of a transient one (I don't think the overall education system today is all that much better than 20 years ago).
It's a "this-time-is-different" situation in a limited sense; that is, it's not inherent to the nature of the automation-cycle itself, but rather to the nature of our response to it.
Note that this isn't some engineering-centric notion of needing immediately applicable skills from college. Studying something more academic in college has just as much value, as long as it's something that's useful (a good rule of thumb is "as long as you use the things you learned in college in the future"). I'm referring to the fact that it's long been the practice for many people to get a college degree in literally anything with the knowledge that the existence of the degree itself would be a good enough signal for intelligence/skills that you could get an okay job afterwards. This is increasingly untrue.
It depends on how fast we can find new material needs. We don't seem to be doing that very quickly though (in lots of areas we are regressing, becoming more efficient with our inputs).
With strong AI, wouldn't 100% of U.S. jobs be at risk? On the bright side, as software engineers, I would think our jobs will be the last to go.
My guess is as a species we'll just find something else to do, just like we always do. Who knows, maybe inefficiency will become fashionable at some point.
Yes, but realistically, we are no where near strong AI. We're getting pretty good at making AIs to handle single specialized tasks, even ones that we consider very difficult, but making a generalized AI that can handle any task a human can is probably not going to happen in the near future.
I'm glad I didn't heed the advice of my grade 12 computer science teacher in 1991. He was convinced that we wouldn't need programmers in the future because we'd have AI which could write software for us.
The train example is a pretty bad one, I'm the first in four generations not to work for the local railroad and five union dudes on a train crew pretty much went away in the 1970s. You do typically have two guys and you really need two for safety as sometimes you need human eyes looking out both windows at once (freight trains have poor visibility compared to trucks).
A freight train with one dude strikes me as unlikely. The few single person trains out there have excellent visibility for the one driver. So you'd have an interesting capital expense if you tried that with freight trains. I do know some small switcher locos run single person but thats in a barb-wired switchyard well away from the general public. So you'd "have to" eliminate all grade crossings or something, another big expense.
The problem is that the relative price of human time decreases, while the relative price of land and resources increases. So wealth moves to people who own resources.
Another problem is that the standard deviation of the price of human time is increasing. The typical example is someone of average intelligence who was an accountant but her job got automated. In general, jobs requiring an average intellect are disappearing and people who did these jobs are more often moving down (manual labor) than up (programmer, designer, etc.).
So the real problem of automation is not unemployment – you can always find a slave-like job for a few bucks per hour – but wealth inequality.
Not even an estimated time schedule. Not like it's gonna suddenly drop 47% employment next year. It would be a slow process of replacing laborious jobs to intelligent jobs. Adapt and survive, isn't that how it suppose to be?
I say AMEN! Imagine a society where 25 hour work weeks is the norm, everyone gets a living wage, some of it possibly subsidized by the government, -- each household earns a guaranteed 2500 per month to grow the economy. Every worker gets 4 weeks paid leave. Companies move from being self-serving to realizing if they make their employees super happy -- they will go out and buy their products, or other companies products - but support the economy in general and grow all companies.
We are work-a-holics, and then we spend our weakest part of life sitting on our arses waiting to die- if we had more quality of life time to travel, see the world, spend time with our families, while in the prime of our life--time to pursue entrepreneurial ideas, philosophical/artistic endeavors...
Not only would poverty be put to shame, but quality of life would rise universally. Sure, sounds like a Utopia, maybe it is, maybe it's what we're headed towards..it's not like there's not enough money floating around to pay for it...
They appear to be using the BLS categorizations[1][2]. Essentially, the difference is that Computer Programmers do none of the planning or design that Software Developers do; they only do routine coding. Since Computer Programmers only do routine coding and none of the more intellectually challenging parts of software development, they can more easily be replaced by using higher-level languages or better tools. It's worth noting that Software Developers are already 3x more common than Computer Programmers and the number of Software Developers is increasing 10x as fast, so even many of the people who call themselves programmers are probably classified as Software Developers.
I'd like for my work to be automated as long as I can keep my salary. But it's not going to work because I won't be the owner of the "machine" replacing me.
It's hard to see this is anything but the symptom of sick global economy, and a sick global society. Automation should be making people's lives better in the short term, not just in the hypothetical long term. We exist within a system that doesn't respect human life.
The probability for software engineers is 4.2% in the paper.
All jobs will be automated in 100 years in my opinion (if technological development continues). In fact, work itself will be automated away. (But some will continue to 'work' for other reasons than gatheting resources).
Yeah, it's a prediction which has always been wrong. However, my prediction depends on the development of an artificial general intelligence which will make human work worthless.
Thanks, hadn't re-read this PG essay in a while. It makes some good and important points, but if the theme is don't hate on the wealthy.. I don't think you can ignore the fact that many people don't believe the 100x productive/add value idea.. Although I think many of us technologists and/or capitalists know first hand that it is true, that is, it is possible..
But, all the wealth accumulation is not the result of a pure capitalism/reward system. 100x wealth does not necessarily equal a 100x contribution of value to society.
There does exist undeserved rewards. That perception, sometimes true, sometimes not, is at the core of the wealth haters and quite a bit of outrage.
Much wealth is controlled by entities that have quasi-governmental status, special protection and privileges (ie.. the entire banking system). Interestingly enough a more pure capitalist system may be more effective at leveling the field instead of a socialist system.
Based on this, and other essays, I am beginning to suspect Paul Graham is the technological equivalent of David Brooks (the NYT columnist). I'm a well educated middle middle class white American and after reading this essay, I immediately groked how the non-white Other groups (gays, black, etc) in society get offended by the ignorant and simplistic assumptions or stereotypical thinking of the In groups (the politically correct brigade being outrageously sensitive to even mere nuances or surface details, of course, in a needless fashion). I am beginning to understand the various criticisms of the HN community... then again, my life has been much more varied than the average middle class American and I'd like to think I have a more comprehensive understanding of life (yeah, arrogant). Or I just think and read too much.
Or maybe it is more a matter of values, and the values of rich white guys in Silicon Valley are hardly those of most of humanity. One's values are factors in the blinders you were while interpreting reality, along with whatever unconscious crap you're not dealing with (or maybe those feed into the values) and from that arises your attitudes, etc.
Anyway, my immediate emotional reaction on reading this was WTF? I wasn't impressed.
All that is solid melts into air, all that is holy is profaned, and man is at last compelled to face with sober senses his real conditions of life, and his relations with his kind!
Faster! Get rid of all the jobs before people can get used to permanent unemployment-ridden depression! Push them into revolting!
I wish I could hide luddite articles like this permanently.
Technology improves efficiency, which eliminates the need for workers. Economics is not a zero-sum game, however. More money saved means more money invested, either by the individual or the bank in the case of pure savings.
Easier lending means more businesses in more industries. Everyone didn't starve to death during the industrial revolution. Dock workers didn't just commit suicide en masse when most of them were obviated by the invention of the shipping container.
Currently we exist in a "low unemployment present" but if you believe a "high unemployment future" is possible, you should reasonable believe we're smoothly approaching it. So if you think my thought experiment can come to pass - we need to be very concerned.
If you're going to bring out the tired argument about how the cotton gin didn't cause mass unemployment, instead I would challenge you to paint a picture of a far off future where most jobs haven't evaporated.