This all falls apart when it comes to the fact that companies aren't controlled by the government, and they'll simply leave the country when taxed.

There's no way to offset costs enough through subsidization to support this, combined with the fact that many jobs require long hours for training purposes. A worker working 30 hours a week isn't going to be as well trained as one working 50 hours a week.

Also keep in mind that executives/high level business members will be automated as well. Who will you tax when the top 2% are replaced with machines?

What I always find fascinating is this idea that suddenly things are changing now more than ever, as if the industrial revolution wasn't a massive change, or that automation hasn't been continually increasing for much of the last 200 years since then. When it's just working/lower class jobs, nobody really minded, but now we see a computer painting a picture or winning at Go, suddenly there's panic, even though those tasks are in relatively low demand. It looks to me less like there's a change in automation quantity, more that it's started to affect the type of people who are able to write articles about it.

My perspective is: yes, the nature of work is changing, but mostly at the same rate as ever. Automation moves people out of jobs but it also reduces the cost of goods and services, which could mean that people don't all need to work the typical 35-40hr week. People might transition to more part-time work, on average. In the short to medium term there will be some roadbumps with this (e.g. how to encourage the system to employ 2 people to work 20hrs a week instead of 1 person working 40 and the other getting no hours) but I think this will be more easily adjusted with targeted help (e.g. funding for retraining) than with blanket measures that are very expensive and politically controversial such as universal basic income.

But the thing is, they arguably are changing more than ever. Yes, things have been changing since the dawn of time, shifting back and forth with things, and we saw massive changes in the last two centuries, but we are seeing job-replacement happen at a rate society has never seen before thanks to technology, and we are developing technology that is posed to rip apart industries like never before with automation.

Self driving trucks? Smart Cranes and forklifts? Automatic maintenance/servicing systems? Humanity is not that far away from materials going into a factory, being processed into finished packaged goods, shipped across the globe and arriving at a customer's door with zero human hands touching it. And this isn't "by year 2000 we'll all have flying cars, because science!" claims - This is real world existing technology demonstrations. We already have all the tech needed to do it, and it is now a matter of deploying it into finalized production designs and building support infrastructure.

"New jobs will replace what was lost. It has always been that way!" is a very short sighted and rather dangerous line of thinking. We are already seeing a workforce struggling to maintain itself, and falling behind what previous jobs offered. Consider how many North Americans were employed and buying homes in their 20's-30's back in the 70's. Now look at the same demographic today. Notice a difference? Sure, they're employed, but the jobs from the US industrial sector haven't been replaced by new positions. There are jobs, but many of them are not replacements of those that we have lost in that they do not provide the economic power or stability of the previous positions.

The argument of "New jobs will be created, they always have" is along the lines of saying "We don't need to worry about 'The Thing' taking away our computers, new computers will be provided in their place!", while ignoring that 'The Thing' is taking away the computer you have now and the only other computer you can get is 10 years older than your old one, if not more.

We are lacking access to capital and security of livelihood for far too many. Society either has to come together and address this issue, or risk creating two or more societies completely at odds with each other - And that typically does not "go smoothly" for anyone involved. And if anyone really insistent on saying "I'm okay with my life, why should I remotely consider any kind of change?!", then they might want to invest in a history textbook or two. The French Revolution might be a good place to start and then sit down to really think about deep seated beliefs in how society should function.

I'll start this by saying that I have a lot more experience with the UK economy than the US one, so what I say may not apply to you or what you experience.

we are seeing job-replacement happen at a rate society has never seen before thanks to technology

And yet employment in much of the Western world is at a very high level (with exceptions for some Mediterreanean countries). It looks like, for now, people are adapting to the changes and getting new jobs.

Self driving trucks? Smart Cranes and forklifts? Automatic maintenance/servicing systems?

Sounds bad for truck drivers, forklift drivers, etc. Doesn't sound worse than when cars replaced horses + carriages, when factories replaced ironmongers, when early computers replaced telephone exchange operators, or ATMs replaced bank tellers, or boring machines replaced miners and tunnel-diggers, etc. And there's not much work for wheelwrights, carters, cobblers, chandlers, or millers these days either.

I come from a place where early automation was such a threat to workers that they basically revolted against it, as it was such a massive change in how people lived and worked. I'm finding it hard to think that self-driving cars are an equally significant change, in a world that already has trains.

"New jobs will replace what was lost. It has always been that way!" is a very short sighted and rather dangerous line of thinking. We are already seeing a workforce struggling to maintain itself, and falling behind what previous jobs offered. Consider how many North Americans were employed and buying homes in their 20's-30's back in the 70's. Now look at the same demographic today. Notice a difference?

I would argue that is a rather different problem, to be addressed by different means. For example, if you double the number of people in a country but the land mass stays broadly the same, and you do nothing to compel that land to be used more effectively, you can imagine housing prices to increase significantly.

But I do expect the cost of living to decline as a result of increased automation, which will help somewhat.

Society either has to come together and address this issue, or risk creating two or more societies completely at odds with each other

No argument there, but I think this is far more about globalisation and attitudes to income and wealth than anything to do with automation.

It seems to me we should try to define the problem in a somewhat more "formal" basis, so we know at least roughly on what the 2 "camps" here agree and disagree on.

-Things we all seem to agree on:

1) Jobs are simply tasks we need performed. Those tasks usually require a combination of strength, speed, and cognitive abilities/intelligence.
2) Machines have been for very long vastly superior to humans when it comes to any task that requires *just* strength and/or speed.
3) Machines are right now still inferior when it comes to most "higher" cognitive abilities - they do number crunching very well though.

4) Each new generation of machines "destroys" some amount of existing jobs - they are either automated or become obsolete.
5) Each new generation of machines creates some amount of new jobs - so far, that amount is greater than the amount of jobs destroyed.

6) It is possible that the amount of new jobs created by each new generation of machines will be *always* greater than the amount of jobs destroyed.
7) The nature of those new jobs we usually can't imagine - nobody 2 centuries ago would be able to understand what most of us in this board do for a living.
8) Out of the "pool" of those "newly created" jobs(say Jc), some can be also automated by the same generation of machines that created them(say Ja); some can not be automated by the current generation, or at least it's still not cost-effective to be automated, and so are still performed by humans(say Jh).
9) This is an old debate; each generation that saw a huge change in automation worried about it(Luddites, etc).
10) The worries, thus far, have been misplaced; at least when it comes to the general situation : the ratio of Jh/Ja seems to always be >=1(or at least >=C, where C is whatever constant is needed for the system to not "break down").

-What we disagree on:

Basically, on the question whether the trend we defined on (10) is going to continue indefinitely.

Camp A believes the "function" is a line that always "goes upwards". It has done so in the past, does so in the present, and will continue to do so in the future. The trend will continue as always; no matter how many jobs will be automated by a new generation of machines, enough jobs will be created that are more cost-effective to be performed by humans than the current generation of machines.

Camp B believes the "function" is a curve, that so far has been going "upwards". On that, like everything else, we agree with Camp A. Our difference is that we believe that, eventually, we will reach a tipping point, when most of the jobs created by the new generation of machines can (and is cost-effective to) be performed by the very same generation of machines that created them in the first place. The curve will "slow down", and eventually "go downwards". And, at that point, a new situation, a new "quality" will arise that humanity hasn't dealt with before.

Do I have this right? :)

Camp A believes the "function" is a line that always "goes upwards". It has done so in the past, does so in the present, and will continue to do so in the future. The trend will continue as always; no matter how many jobs will be automated by a new generation of machines, enough jobs will be created that are more cost-effective to be performed by humans than the current generation of machines.

Camp B believes the "function" is a curve, that so far has been going "upwards". On that, like everything else, we agree with Camp A. Our difference is that we believe that, eventually, we will reach a tipping point, when most of the jobs created by the new generation of machines can (and is cost-effective to) be performed by the very same generation of machines that created them in the first place. The curve will "slow down", and eventually "go downwards". And, at that point, a new situation, a new "quality" will arise that humanity hasn't dealt with before.

I think that this is a question of recursion. Recursion definitely exists, so the "Camp A" type of prediction cannot necessarily hold forever.

I'm more interested in having a contingency plan in place in case that new "quality" you mention arises.

I'll start this by saying that I have a lot more experience with the UK economy than the US one, so what I say may not apply to you or what you experience.

we are seeing job-replacement happen at a rate society has never seen before thanks to technology

And yet employment in much of the Western world is at a very high level (with exceptions for some Mediterreanean countries). It looks like, for now, people are adapting to the changes and getting new jobs.

Yea people are employed, but just look at the recent election campaigns, at least in the US. There was one common theme: loss of manufacturing jobs. Jobs issues in general. Perhaps many of these people are underemployed? Manufacturing and low skill jobs in general are disappearing. Just look at Detroit, or any number of manufacturing towns. I come from Pennsylvania, and if you know your history, Pennsylvania was once the center of steel manufacturing. At one point in time, this state was booming. That's changed now. While I live in an urban area, all I have to do is drive about an hour or so on the interstate and go to any number of steel mill towns. There's nothing left in those places. They are ghost towns of what they used to be.

Self driving trucks? Smart Cranes and forklifts? Automatic maintenance/servicing systems?

Sounds bad for truck drivers, forklift drivers, etc. Doesn't sound worse than when cars replaced horses + carriages, when factories replaced ironmongers, when early computers replaced telephone exchange operators, or ATMs replaced bank tellers, or boring machines replaced miners and tunnel-diggers, etc. And there's not much work for wheelwrights, carters, cobblers, chandlers, or millers these days either.

I come from a place where early automation was such a threat to workers that they basically revolted against it, as it was such a massive change in how people lived and worked. I'm finding it hard to think that self-driving cars are an equally significant change, in a world that already has trains.

Look at it this way. Say you have a system which can get the raw material, put into a factory, manufacture it, then spit out a finished product and deliver it to a customer all without a human in the loop (which is by no means unrealistic). What happens then? What jobs will the unskilled drivers, etc. do? The only possible jobs that could be generated are potentially people to monitor those systems, and of course the programmers and designers for these systems. Now not everyone can be a programmer, designer, software dev, etc. so those jobs are out for the unskilled guys who just lost their job. That leaves maybe the monitor guys? How many monitors do you possibly need for this? And how long will it be before even those are replaced by better systems or by redundant automation itself? What I fail to see is how self sufficient automated systems will result in more jobs. The whole point of automation is so that humans have to work less.

"New jobs will replace what was lost. It has always been that way!" is a very short sighted and rather dangerous line of thinking. We are already seeing a workforce struggling to maintain itself, and falling behind what previous jobs offered. Consider how many North Americans were employed and buying homes in their 20's-30's back in the 70's. Now look at the same demographic today. Notice a difference?

I would argue that is a rather different problem, to be addressed by different means. For example, if you double the number of people in a country but the land mass stays broadly the same, and you do nothing to compel that land to be used more effectively, you can imagine housing prices to increase significantly.

But I do expect the cost of living to decline as a result of increased automation, which will help somewhat.

Society either has to come together and address this issue, or risk creating two or more societies completely at odds with each other

No argument there, but I think this is far more about globalisation and attitudes to income and wealth than anything to do with automation.

Ok I will grant you this that globalization has lead to some of these problems as much as automation, and that I think is probably deserving of its own thread but I'll say just this: it is simply impossible for a worker in a country in the West to compete with people who are practically willing to work for a penny.

It seems to me we should try to define the problem in a somewhat more "formal" basis, so we know at least roughly on what the 2 "camps" here agree and disagree on.

-Things we all seem to agree on:

1) Jobs are simply tasks we need performed. Those tasks usually require a combination of strength, speed, and cognitive abilities/intelligence.
2) Machines have been for very long vastly superior to humans when it comes to any task that requires *just* strength and/or speed.
3) Machines are right now still inferior when it comes to most "higher" cognitive abilities - they do number crunching very well though.

4) Each new generation of machines "destroys" some amount of existing jobs - they are either automated or become obsolete.
5) Each new generation of machines creates some amount of new jobs - so far, that amount is greater than the amount of jobs destroyed.

6) It is possible that the amount of new jobs created by each new generation of machines will be *always* greater than the amount of jobs destroyed.
7) The nature of those new jobs we usually can't imagine - nobody 2 centuries ago would be able to understand what most of us in this board do for a living.
8) Out of the "pool" of those "newly created" jobs(say Jc), some can be also automated by the same generation of machines that created them(say Ja); some can not be automated by the current generation, or at least it's still not cost-effective to be automated, and so are still performed by humans(say Jh).
9) This is an old debate; each generation that saw a huge change in automation worried about it(Luddites, etc).
10) The worries, thus far, have been misplaced; at least when it comes to the general situation : the ratio of Jh/Ja seems to always be >=1(or at least >=C, where C is whatever constant is needed for the system to not "break down").

-What we disagree on:

Basically, on the question whether the trend we defined on (10) is going to continue indefinitely.

Camp A believes the "function" is a line that always "goes upwards". It has done so in the past, does so in the present, and will continue to do so in the future. The trend will continue as always; no matter how many jobs will be automated by a new generation of machines, enough jobs will be created that are more cost-effective to be performed by humans than the current generation of machines.

Camp B believes the "function" is a curve, that so far has been going "upwards". On that, like everything else, we agree with Camp A. Our difference is that we believe that, eventually, we will reach a tipping point, when most of the jobs created by the new generation of machines can (and is cost-effective to) be performed by the very same generation of machines that created them in the first place. The curve will "slow down", and eventually "go downwards". And, at that point, a new situation, a new "quality" will arise that humanity hasn't dealt with before.

Do I have this right? :)

That is more or less sums up the problem here. I'd argue that automation on a massive scale is much closer than people think.

There are still some very big issues to get over though, with the original solution proposed.

1): Ultimately we have to at least initially rely on taxation of the people who would be working and managing automated systems. Cultural perceptions of this can help, but there is the fact that some will feel like they are working for everyone and not getting much out of it. That can lead to income disparities, which may potentially be offset by the fact that automation has made production incredibly cheap. The other thing is that it may be offset by the fact that doing something is better than being bored out of one's mind doing nothing all day.

2): Will people who have nothing to do simply be content with that? That's the real question.

This all falls apart when it comes to the fact that companies aren't controlled by the government, and they'll simply leave the country when taxed.

What if you put a tax on companies that leave the country?

There's no way to offset costs enough through subsidization to support this, combined with the fact that many jobs require long hours for training purposes.

That's a fairly definite claim. Do you have numbers to back it up?

A worker working 30 hours a week isn't going to be as well trained as one working 50 hours a week.

But a worker working 30 hours a week may be better rested and have a higher quality of life than the one working 50 hours a week, resulting in the work done during those 30 hours being better quality than the work done in the 50 hours, resulting in better work being done for less time. Would you claim that hours worked does not impact quality of life or that quality of life does not affect quality of work?

@samoth: You keep indignantly yelling "you socialists get your hands out of my pocket!", yet you never yell "you capitalists start paying decent wages to the people that bring/make my coffee/tea/fruits/clothes!".

Is it presumably because the latter is providing you(all of us in the 1st world, that is) with cheap products, and you(we) very much like those products keep being cheap?

Ah, one of the classic arguments. But like all the classic socialist arguments, it's a lie. Why is it a lie? Because the "low wage" that you pay to those poor people is not low at all in their country.

But even if the argument was true, it wouldn't change the fact that you socialists are stealing from me. They always say "those fucking rich don't pay taxes anyway, even though the top 1% owns 90% of everything", but this is yet another lie.

The truth is, the top 1% owns maybe 20% of everything, if that, but there exist a handful of people (less than a dozen per country, which is more like 0.0001%, not 1%) who own 50-70% of everything, depending on which country you look at (more extreme in e.g. the USA, less extreme in EU).

Those very few, singular people who are super-ultra-stinking rich usually indeed pay hardly any taxes, nor do they (usually) do something useful, I'll admit that to my "socialist comrades". However, and that's the important detail for staying with the truth, all other "rich bastards" worked their ass off, and pay an outrageous amount of tax, leaving not much more than average in the end -- to feed people who are too lazy to work. Which is nothing but theft. Legalized theft.

I'm not saying the rich shouldn't pay taxes. But so should everybody, not just 1% of the population (or 3% or 5%, choose your number).

Titus petronius had a phrase for that socialist behavior two thousand years ago already: stratum caedit qui asinum non potest.

The problem is that traditionally new technologies created new jobs.

The change from horses to internal combustion powered vehicles made farriers obsolete, but created massive industries in terms of manufacture, sales and maintenance of those cars, and people needed to develop new skills to handle all this. Yay progress!

The problem is with AI and automation... we are not the people; we are the horses.

When you can automate an AI/robot to do a job cheaper than a human (remember, it doesn't have to be better or faster, just cheaper), then you have zero incentive to employ a human.

"Ah, but someone will need to maintain the automatons, thus creating new jobs for humans!"

Yeaaaahhhhhh..... no.

1) The maintenance of said automatons will be carried out by, yep, other automatons.

2) As the cost of these automatons comes down (and it inevitably will), any serious issues will just result in the automaton being replaced.

Most people seem to think that automation of jobs will kill off menial work, and people will just have to do jobs that AI can't do. The problem is that the number of jobs AI can't do is shrinking exponentially.

Doctoring? Watson is getting there

Lawyering? Software is already doing discovery.

And for something more relevant to this site.... programming? Oh look, machines are getting better at that too.

Even if you work in a "creative field", you're not safe.

Music? AI can do that

Painting? AI can do that too

Journalism? Yep

And that's just what's available now.

This is not a luddite position. I'm not opposed to this. It's going to happen and we can't stop it.

And nor should we, because it's ultimately the way we free ourselves from having to work and transition to wanting to work.

But that's not going to happen for a generation or two because short-sighted people will whine about taxes and productivity and lazy dole bludgers and waaah wah wah.

So we are in for a bumpy ride over the next century (assuming climate change doesn't get us first).

Disclaimer: A lot of the examples were pulled from CGP Greys excellent Humans Need Not Apply video. It really is mandatory watching for any discussion on this topic.