What jobs will employ the masses?

[billvon 2,400]

>I think you are vastly underestimating the number of employees that will be replaced by robots.

Well, heck, go back to the 1950's and we were all supposed to be replaced by now. Didn't happen. In fact, new industries spawned by automation (say, the Internet) actually created _more_ jobs, and more wealth.

[quade 3]

billvon

Well, heck, go back to the 1950's and we were all supposed to be replaced by now. Didn't happen.

Have you visited a car or parts manufacturing plant lately?

When was the last time you visited the steno-pool at a major corporation?

Have you visited a check-in counter at an airline?

Been to an ATM machine lately?

Self check out at the grocery store?

By and large the fears of the 50s did, in fact, come true. Thousands and thousands of jobs no longer exist due to automation and it's going to increase as time goes on.

Pretty much any job that can be replaced by robotics, will be replaced by robotics and the robots are rapidly winning.

quade -
The World's Most Boring Skydiver

[billvon 2,400]

Quote

Have you visited a car or parts manufacturing plant lately?

Yes, I work at one.

Quote

When was the last time you visited the steno-pool at a major corporation?

Never.

Back in the 1980's, how many people did you know who were web designers? Mobile app programmers? Cellphone accessory manufacturers? Wireless service retailers? Solar installers? Server maintenance guys?

Quote

By and large the fears of the 50s did, in fact, come true.

OK. Then we've all been replaced and there's nothing more to worry about.

[wmw999 2,123]

Quote

Ultimately we're going to get to the point that most of our population can't work for private industry except as human conspicuous consumption (uniformed human maids, gardeners, cooks, and butlers) for the very well off and there won't be enough of them to employ the rest of us.

Once past that point those people will be receiving government stipends and voting for politicians who ensure the payments are not stingy.

This. We're partway there already. How many ditch diggers do you see any more (or any other jobs that just require brawn and an ability to follow basic instructions)

Wendy P.

There is nothing more dangerous than breaking a basic safety rule and getting away with it. It removes fear of the consequences and builds false confidence. (tbrown)

[quade 3]

billvon

Back in the 1980's, how many people did you know who were web designers? Mobile app programmers?

^^^ Fair, but not a significant amount to take up the slack.

Quote

Cellphone accessory manufacturers? Wireless service retailers? Solar installers? Server maintenance guys?

^^^ Will be replaced by robots.

quade -
The World's Most Boring Skydiver

[Bignugget 0]

ryoder

Someone needs to install & service the robots.

Yea, other robots!

[ryoder 1,384]

quade

***Someone needs to install & service the robots.

Nope. We're well past spoiler period in for the movie Elysium, but that was one of my big issues with it. The robots were sophisticated enough to do all sorts of precision work. Hell, even today FoxxCon is replacing it's human workforce with robots to make products like iPhones. If robots can make something as precise and elegant as an iPhone, they sure as hell can make other robots. So why were humans such as the Matt Damon character ever given jobs to build them? Made zero sense.

In the future, robots will repair robots. I don't think that's even speculation. Simply a fact that hasn't happened yet.

We have had robots in auto factories for many years now, but I have yet to see one in an auto repair shop.

"There are only three things of value: younger women, faster airplanes, and bigger crocodiles" - Arthur Jones.

[DrewEckhardt 0]

ryoder

We have had robots in auto factories for many years now, but I have yet to see one in an auto repair shop.

Computer vision wasn't up for the task, processing power was insufficient, and if that wasn't the case capital costs of sophisticated robots would have precluded human replacement in that situation.

Cars also weren't yet designed for easy robotic service although that's changing too - the Tesla Model S is built for robotic battery changes and in the future we could do the same thing with other field replaceable units.

[jgoose71 0]

Maybe if we can get the robots to unionize and demand Obamacare, we can price humans back into the market.

"There is an art, it says, or, rather, a knack to flying. The knack lies in learning how to throw yourself at the ground and miss."
Life, the Universe, and Everything

[davjohns 1]

masterblaster72

***

What jobs will employ the masses?

Fracking jobs, because the energy companies said so in their websites and advertisements.

Since watching the remake of Battlestar Galactica, I can't see the word 'fracking' without a completely different meaning popping in my head. Changes the meaning of your retort tremendously.

I know it just wouldnt be right to kill all the stupid people that we meet..

But do you think it would be appropriate to just remove all of the warning labels and let nature take its course.

[kelpdiver 2]

billvon

>I think you are vastly underestimating the number of employees that will be replaced by robots.

Well, heck, go back to the 1950's and we were all supposed to be replaced by now. Didn't happen. In fact, new industries spawned by automation (say, the Internet) actually created _more_ jobs, and more wealth.

The current issue of Fortune has an article on IBM and Watson in particular. It's not a long distant future where it and other natural language processors will have a highly disruptive effect on the number of people it takes to process information for a call center, or a pharmacy, or who name it.

[ibx 2]

Quote

This. We're partway there already. How many ditch diggers do you see any more (or any other jobs that just require brawn and an ability to follow basic instructions)

Wendy P.

Actually there are still tons of jobs that you describe(fruit picking, construction, yard work), just most of them are done out of sight by the so called illegals.

[Fast 0]

I'm excited. An appropriately programed computer won't fuck up my order at the drive through when my self driving (seriously coming soon to a car near you) car takes me for fast food at 2am.

In all seriousness though - the issue is quite multifaceted. I mean, I have been working on a very basic project with an arduino for a halloween costume. The things I could do with some money behind it would be amazing. Just having a vision for that kind of thing - the kind of thing we couldn't have even dreamed about being reality 20 years ago.

As a kid I played with capsela toys and you could do very low level basic robotics with it. Lego, etc... The toys of today are designed around those things and training the minds of our youth. Too bad there are going to be too many of them (over population).

This post couldn't possibly be long enough to get a handle on my train of thought on this subject. It's very frustrating.

~D
Where troubles melt like lemon drops Away above the chimney tops That's where you'll find me.
Swooping is taking one last poke at the bear before escaping it's cave - davelepka

[AdamLanes 1]

If only we could get robots to do all the work for us, then we humans could just kick back and RELAX!

[lurch 0]

"In the future, robots will repair robots. I don't think that's even speculation. Simply a fact that hasn't happened yet."

(Laughter)
Quade, what is your experience with industrial robotics? Do you have any idea what the state of the art in industrial robotics actually IS? I do.

Have you ever seen one? Operated one? diagnosed or fixed one?


This IS my line of work and it is the reason I can even afford to skydive. I've worked on hundreds of them in dozens of industries. I AM one of the guys that fixes the robots. I used to design and build 'em, too, but that job got outsourced to Tijuana. These days I get paid the same for FAR less work at a much smaller factory with really small cruddy robots. To everyone else in the plant, they're scary mystery boxes with arms. To me they're simple tinkertoys. 6-axis basic programmable jobs without even any serious machine vision or REAL high technology... junk I could have thrown together back in high school. Heathkit stuff.

Say what you will about the state of the art in technology itself and how nobody can predict it, because of how fast it advances, but 99% of that advancing is in information tech, chips, surface mount technology and so on. Manipulation of electrons, not the machines' ability to manipulate physical reality. Physical technology develops much, much slower, which is why you can buy a smartphone lightyears more advanced than the last one every 2-3 years most of which can be serviced by plugging it into a computer and having the computer talk to it but things like plumbing and cars are still serviced the old fashioned way by a guy with a wrench.

I'm very familiar with most forms of industrial automation, enough so that when I come across a new type I haven't seen before I already automatically know mostly what it is, what it does, and how to fix it just by looking at it. Sometimes to diagnose the deeper subtler failures I have to cycle it a few times to get the idea, but to this day after a nearly 20 year career in industrial machine technology I have yet to encounter a piece of gear I couldn't understand within minutes.

Once you've worked on enough of these things, despite their fantastic variety of form and function, fundamentally they're all built the same. As if they were built of Legos. Ten thousand times as much variety in possible parts than Legos have, but still, its like complex legos for grownups. It all goes together the same way. For every possible need or function there are dozens of manufacturers making components to fill the need, various levels of interchangeability. Same basic parts. Servos. Stepper motors, their drivers, and the computers that tell those drivers what to do. Conveyors. Air cylinders that push and pull things. Solenoids controlling those cylinders. Sensors... optic, fiberoptic, magnetic, proximity, capacitive. I/O boards to take info from sensors, PLCs (programmable logic controllers) or dedicated computers to decide what to do, and more I/O boards to take the resulting instructions, trigger solenoids or stepper motors and act on those decisions.

Now, cars have grown amazingly reliable... and yet they still fail and wear in ways the computer in the car does not and cannot understand. To actually teach a car to monitor and report its own health... say, to detect and report a worn ball joint, universal joint, or pittman arm, would require the vehicle to be 200X as complicated as it is already. Vast areas of sensors and computing apps would be required, all of which would need to be refined to nearly 100% reliability before it was deployed as a model you could buy. 99% of a car is still stone-age stupid with zero awareness of itself. The ONLY part of a car with serious awareness is the engine, and at that, everything the engine computer says has to be interpreted and taken with a grain of salt because the computer cannot tell one failure from another.

As it is, almost nobody appreciates or understands just HOW smart our engine monitoring systems are these days... my Jeep can largely troubleshoot itself... but thats because its got a single-purpose, highly evolved system, designed for an extremely high-demand application whose development was driven by a virtually unlimited consumer market demand. Everybody needs one, and one way or another, 4, 6, 8 cylinders, twin cam single cam whatever, within a certain relatively narrow range, all cars are essentially the same.

This is not true of robots. They're -built- the same, (PLCs, air lines, belt drives, etc ad infinitum) but they do NOT take roughly the same form and perform roughly the same function.
Not everybody needs a gantry robot that can take 32 parts out of a mold, detect their presence or absence, decide to either dump them down a reject chute or an accept chute, verify the chute is clear, take intervention action if it is not, detect, stop and set off an alarm if its intervention fails, move to the next slot, and repeat.

Every last fraction of a step in even that little process needs all kinds of custom, one-off application specific physical tech to do that. A basic blind robot is simple. Giving that robot the ability to just detect a stuck chute gets MUCH more complicated. Giving the robot the physical ability to ACT on that one, static laboratory controlled-condition state multiplies that complexity X10, giving it more than one option for action and the ability to decide what to do multiplies that another 10 times over.

And thats just for one simple small subfunction. Now what happens if the air cylinder it uses, wears out its seals, and responds slowly or not at all? Its easy to make the robot detect that somethings wrong, and where, but not WHAT.

Same as engines. When I had issues with my jeep's engine, the computer knew about it and could tell me it was misfiring sporadically on cyl.3 but not what the failure was or why. It COULD detect things like a burned ignition coil or failed injector, IF the failure was electrical. If the problem is physical, and the engine has no means to see it, the computer is utterly helpless to know or do anything about it. The engine computer knew something was wrong with that cylinder by analyzing crank rotation timing and detecting the faint lag in 1/6th of a power cycle, but since electrically it all checked out, all it could do was tell me "something wrong in cyl3."

The problem turned out to be a worn lifter and cam lobe... and we are still 20-50 years away from anything like engine technology that could have understood that for itself, and even further from anything that could FIX itself. And this, mind you, is the pinnacle of a technology with all the world's automakers and uncounted billions driving it. A relatively restricted application. Just cars. (trucks, etc. 4 wheeled engine driven stuff.)

And that's just the engine. Try implementing a system that could detect a worn out balljoint. Or tie rod. Or control arm. Each separate possible failure would need a subsystem developed for it and dedicated to it, able to detect most or all possible states, and tell you about it.

Giving the car the ability to DO something about it introduces orders of magnitude more complexity, cost, etc.

And again, thats for a common, universal, restricted application with virtually infinite money driving its development.
Doing the same for something as fluid as manufacturing robotics, the complexity required goes up exponentially... and thats -physical- complexity not software, so no shortcuts via software upgrade. No, you have to design and build the sensors, brackets, actors...

A very limited form of self-repair ability CAN be done, mostly by modular, yank-and-replace modular construction, but again, the robot to DO that would be 200X as complex as the robot its designed to fix... and as soon as it encounters a failure it has not been taught to deal with or has no means to see, understand, and act on, it just sits there blinking.

And 99.99999% of industrial automation is NOT amenable to being designed for such automated self-diagnosis and repair, because of how custom every app is. Its hard enough to design a smart enough robot to simply detect and unjam itself, having one that understands the other well enough to tell what is wrong, and act to repair it... its not impossible but the ROI on the investment to build one IS.

Physical technology always develops the slowest because it is physical. Making physical technology smart, is even slower. Software tech is easy... fundamentally its a lot of typing and telling software to write other software. New things can exist just by declaring it to be so. A mere few billion instructions, gates on a chip changing states. To do the same things physically, every last pixel of it must be MADE... in a machine shop... milling. Adding threads to a hole so you can screw in a push-connect fitting in turn allowing you to plug in an air line.

Bottom line is, we are far, far, FAR away from robots that can fix other robots. Not years, try many, many decades. Right now we could spend a billion dollars, on a simple pick-and-place bot, giving it self awareness, 99% of its complexity would be in that awareness and action ability not in its actual function, and the moment something happens where the ability to detect or act on it was not engineered into it, it will simply stop, helpless. One blown gasket in a solenoid, one loose hex cap screw causing a loose and floppy bracket in turn causing erratic readings from the sensor mounted on that bracket... (how do you teach the bot to tell the difference between a failed sensor, a blocked sensor, and a sensor on a loose bracket?) one slightly leaky push-connect fitting, one dry bearing causing resistance, slow response, servo miscounts... the robot might be able to tell that its happening, but not what or why... why is my conveyor not responding properly...? Stuff stuck in it? Motor drive blown/not responding? Power surge tripped a breaker on the motor drive power supply subsystem? Belt broke? Dry bearing? Something blocking a sensor making me think there's parts still on it or not clearing the gate keeping me from giving the OK to start the conveyor in the first place?

A robot able to diagnose any and all possible failures of just that one conveyor, would be more complex than the entire factory it lives in... more than ten such factories combined...

Now try building a robot capable of dismantling the other one, and changing out the bearing.
Now try making a robot that can figure out which sensor is failed/obstructed. Not too hard... but now try making a robot that can DO something about it. A billion dollars worth of the finest automation man can create can and will be defeated by dust on the business end of a single obscure fiber optic sensor.

Or say you build a bot that can do all that... did you know fiber optics can fail? Vibration causes microcracking... affecting light transmission... the yes/no starts degrading and giving "maybe" answers the bot can't deal with or understand... the sensor behind the fiber starts showing erratic signal returns. Gets decalibrated. Gives a yes when reality is a no. Robots only see and know what their sensors tell them. It takes a human to understand all this. Giving a robot that ability... every possible failure needs awareness and action capability engineered into it. To deal with every possible state the robot can experience. The more complex the robot the more failure states it can experience and the more complex the robot to detect and fix it needs to be. The complexity rapidly ramps up to infinity. Just creating a robot that could detect, and change out a dry bearing, (massive complex disassembly procedure) would cost thousands of times what the basic robot it is fixing would cost.

I'm not gonna say its never gonna happen, but it sure as hell isn't going to happen in my lifetime, or my grandkids' lifetimes.

The distance from here to robots fixing robots is about the same as the distance between early steam engines and airliners with autopilots and ILS.

Its strictly science fiction and its why I don't worry about the machines rising up a'la The Matrix. If they could design, build and fix themselves, it'd be a problem. If there were a robot, even on the 80-year distant horizon, that could spot a leaky pipe take apart the plumbing and fix it, I'd be nervous. But now and for the foreseeable future, all robots in existence need constant help from humans, and the moment we stop dicking with em they grind to a halt within minutes...hours... days at the most. The industry to design and build the robot-fixing-robot isn't even nascent, it doesn't exist. Its like worrying about self piloting airliners when we can't yet build a steam engine to run for 5 minutes yet. You'll see rosie the housecleaning robot, (a "simple" task, comparatively) decades before you'll see a robot capable of coping with industrial troubleshooting and repair, and Rosie is still a century away or more... think Roomba and how helpless THAT silly toy really is.

If you know machines, you're not gonna be out of a job anytime soon. If you don't, well, good luck in the bread line.
-B

Live and learn... or die, and teach by example.

[billvon 2,400]

>Fair, but not a significant amount to take up the slack.

Given that employment hasn't changed much (on average) from the 1950's - they _have_ taken up the slack.

>Will be replaced by robots.

I think you might have an overly optimistic picture of what robots can do.

[champu 1]

quade

***Back in the 1980's, how many people did you know who were web designers? Mobile app programmers?

^^^ Fair, but not a significant amount to take up the slack.

This sounds very much like an "everything's already been invented" kind of argument.

One of the few types of social programs I think we need to focus on is a more robust infrastructure of flexible trade schools. Like sixth month-ish certificate programs that can be retooled as time goes on and certain crafts become more and less relevent. The moral of the story when you talk about robot economic takeovers is that high school doesn't cut it anymore at making people employable and not everyone wants to, can afford to, or would do well to get a college degree.

[masterblaster72 0]

AdamLanes

If only we could get robots to do all the work for us, then we humans could just kick back and RELAX!

Yup, that's exactly what the book People's Capitalism is about. However it makes the very naive assumption that the executives of companies replacing a labor force with robots would choose to distribute the wealth the machines create, rather than keep the money for themselves.

Be humble, ask questions, listen, learn, follow the golden rule, talk when necessary, and know when to shut the fuck up.

[kelpdiver 2]

champu

******Back in the 1980's, how many people did you know who were web designers? Mobile app programmers?

^^^ Fair, but not a significant amount to take up the slack.

This sounds very much like an "everything's already been invented" kind of argument.


It's more that these progressions cut down the manpower required substantially. The natural language developments mean that we'll need far less librarians or human encyclopedias for a given subject matter (like in call centers). If the information can be retrieved in seconds at > 95% accuracy, you can have far fewer question takers. And a few hotshit software developers. So yes, new jobs are created, but far fewer, and very specialized.

It used to be that you'd have one system administrator for every 30 or so servers. This was a rather normal ratio through the 90s. But as grid computing became more common, that sort of scaling broke down and system automation has advanced dramatically in the past decade - it has to. My current company has 16000 servers, and could easily be at 100k in a few years as we think about switching towards cheaper commodity compute units. Even if you could afford to hire people at that 30 or 100 ratio, you couldn't effectively manage them and have them acting consistently. So the end result will be a few more low level datacenter staffers, and a small core of high end system engineers. The computers will be doing much of the work that the humans used to do. (yes - it's skynet)

[quade 3]

AdamLanes

If only we could get robots to do all the work for us, then we humans could just kick back and RELAX!

You understand the human nature and desire of the situation.
You do not understand the reality.

If all you're doing is kicking back and relaxing, how are you paying your bills?

quade -
The World's Most Boring Skydiver

[billvon 2,400]

>If all you're doing is kicking back and relaxing, how are you paying your bills?

You are working LESS overall but making the same or more money. (See real world example above from Lurch.)

[quade 3]

Quote

The distance from here to robots fixing robots is about the same as the distance between early steam engines and airliners with autopilots and ILS.

You do realize you just said it's less than 100 years away; right?

quade -
The World's Most Boring Skydiver

[quade 3]

billvon

>If all you're doing is kicking back and relaxing, how are you paying your bills?

You are working LESS overall but making the same or more money. (See real world example above from Lurch.)

You continue to repeat that as if it's some sort of given. It's not. Further, it might apply to a few jobs, but it almost certainly won't apply to the vast majority of them.

Go back to the article I posted to start this conversation. Robot baristas. The Starbucks I visit on a regular basis has four people working at any given moment. Those four people are minimally educated. They're never going to be software engineers. They're never going to be robot technicians.

What are they going to do for a living when all of the jobs like that are done by robots?

Don't say they'll be paid the same and just work fewer hours. They won't. They'll be completely replaced altogether.

quade -
The World's Most Boring Skydiver

[billvon 2,400]

>You continue to repeat that as if it's some sort of given. It's not.

It's a pretty basic principle of economics.

If you reduce overall hours worked per week (i.e. reduce labor across the board) AND you improve productivity per hour so the reduced number of employees produce the same amount - then the total paid to labor does not need to change to have no impact on a business.

>Go back to the article I posted to start this conversation. Robot baristas. The
>Starbucks I visit on a regular basis has four people working at any given moment.
>Those four people are minimally educated. They're never going to be software
>engineers. They're never going to be robot technicians.

Good example. So you have four people working at any given time, presumably with the work spread out amongst a pool of 10 people or so. Say they all work 40 hours a week. Starbucks makes X from that store a week.

Now add a robot that does something (doesn't matter what as long as it improves productivity.) Now you don't need 400 hours of labor a week to run that place - say you only need 200. Now you have 10 people each working 20 hours a week. Given the assumptions above (i.e. that Starbucks still makes X a week) then you can pay those people exactly the same amount total (i.e. twice their hourly wage) and still make the same profit.

And if every coffee shop/Wal-Mart/Burger King etc on the planet added that automation at the same time you would see such a reduction in work hours without a corresponding decrease in pay. Since that won't happen all at the same time, you will see elasticity in the market - some guy wants to make twice what he used to make, so he still wants 40 hours, and they let the laziest guy go. But again that's because of not _enough_ automation and/or uneven rollout of it.

This is going to lead to something of a sea change because right now the societal norm is that you work 40 hours a week. In places with very high productivity (like Norway and France) this has already led to the reduction of hours worked, since people simply don't need to work as hard to generate the same output, and thus can make more money while working fewer hours. Here in the US our culture is tied to the 40 hour work week, and that will take a while to change. But it will as wages continue to increase; people will simply be able to afford to work less.

[quade 3]

billvon

Now add a robot that does something (doesn't matter what as long as it improves productivity.) Now you don't need 400 hours of labor a week to run that place - say you only need 200. Now you have 10 people each working 20 hours a week. Given the assumptions above (i.e. that Starbucks still makes X a week) then you can pay those people exactly the same amount total (i.e. twice their hourly wage) and still make the same profit.

You've been to the US; right?

Sure, there are exceptions, but by and large the only reason a lot of employees are paid as much as they are is because it would be illegal to pay them less.

So, yes, in a sense you're absolutely right, the companies could pay their employees more, but seriously, who the is going to do that? What is the incentive when you can replace them with robots?

quade -
The World's Most Boring Skydiver