Singularity flip side
September 24, 2013 11:03 AM   Subscribe

47% of US jobs under threat from computerization according to Oxford study. The study reveals a trend of computers taking over many cognitive tasks thanks to the availability of big data.

The future of employment: how susceptible are jobs to computerization? evaluated around 700 jobs, classifying them based on how likely they are to be computerized, from low risk occupations (recreational therapists, emergency management directors and healthcare social worker) to high risk ones (library technicians, data entry keyers and telemarketers).

The availability of big data was identified as a major trend that's given engineers huge amounts of complex data to work with, which has made it possible for computers to deal with problems that, until recently, only people could handle.

An appendix containing the full list of jobs considered can be found at the end of the study.
posted by stbalbach (120 comments total) 35 users marked this as a favorite
 
69. 0.015 Computer and Information Research Scientists
70. 0.015 Chief Executives


Heheheheheheh...
posted by Slap*Happy at 11:08 AM on September 24, 2013 [7 favorites]


Surely by automating more tasks, we can free our citizens from the drudgery of pointless makework and enable them to live fuller, happier lives!

Or, we can put them all out on the street I guess, whatever
posted by showbiz_liz at 11:10 AM on September 24, 2013 [75 favorites]


293. 0.48 Computer Programmers

So we got a roughly 50-50 chance they will learn to program themselves.

I predict that within the next 10 years most office computers will spend 4-6 hours a day looking at pictures of cats and playing minesweeper.
posted by Ad hominem at 11:10 AM on September 24, 2013 [25 favorites]


I'm sure that computerization will devastate the job market in the same way that the industrial revolution and the roboticisization have devistated the farming and manufacturing job markets in the past.

I came in here to snark, but this paper is actually quite interesting. Thanks for posting!
posted by rebent at 11:13 AM on September 24, 2013 [1 favorite]


Eventually, a computer will post your favorite comments and favorite them as well. This conversation seems very appropriate today as well. Would you like to read a bit. Boost the vocabulary and word power.
posted by MCMikeNamara at 11:23 AM on September 24, 2013



puts "Computer sez no"
posted by srboisvert at 11:23 AM on September 24, 2013 [1 favorite]


I don't see the problem here, other jobs will be created, like it always has when technology advances.
posted by MisantropicPainforest at 11:24 AM on September 24, 2013 [3 favorites]


People have been predicting the computerization of computer programmers since there have been computers.
posted by sonic meat machine at 11:25 AM on September 24, 2013 [1 favorite]


"254. 0.36 Telecommunications Equipment Installers and Repairers, Except Line Installers"


Future generations will be puzzled by "Logjammin".
posted by Hairy Lobster at 11:26 AM on September 24, 2013 [3 favorites]


702. 0.99 Telemarketers

Is this a good thing or a bad thing? I DON'T KNOW
posted by tittergrrl at 11:26 AM on September 24, 2013


"And so with the computerization of computer programming we saw the great Jorts Crash of 2019. Rivers of Bawlz ran through the streets and Doritos sat rotting in the stores as shopkeeps were unable to sell their now-useless items."
posted by Ghostride The Whip at 11:27 AM on September 24, 2013 [6 favorites]


I like how my job, which is to implement and optimize big data, has a less than 1% chance of being eliminated by my doing my job. It's nice to know that this article's algorithm thinks I'm not putting myself out of a job.

But I have second thoughts about whether or not the calculation is correct, seeing as I just saw half my team get laid off this summer.

Meanwhile, I look forward to the day when Watson calls me to offer me a deal. A telemarketer who will never be frustrated, always polite and attentive... you know, sales might go up.
posted by linux at 11:30 AM on September 24, 2013 [1 favorite]


I don't see the problem here, other jobs will be created, like it always has when technology advances.

One day a computer will hire me to play minesweeper against it and take pictures of cats.
posted by Ad hominem at 11:30 AM on September 24, 2013 [4 favorites]


702. 0.99 Telemarketers

Is this a good thing or a bad thing? I DON'T KNOW


I guess it's good in that humans will no longer be doing one of the worst, most soul-crushing jobs in the world. But don't fool yourself into thinking that the computerization of telemarketing would result in fewer telemarketing calls.
posted by baf at 11:34 AM on September 24, 2013 [1 favorite]


within the next 10 years most office computers will spend 4-6 hours a day looking at pictures of cats and playing minesweeper

Mine already does. Speaking as its manager, I'm trying extensive micromangement and 1-on-1 coaching but there's really only so much you can do.
posted by Kadin2048 at 11:34 AM on September 24, 2013 [6 favorites]


People have been predicting the computerization of computer programmers since there have been computers.

In a sense, it's absolutely true - programming languages have been getting higher and higher level since day one. So even though there is still a person at the helm, that person is doing the work of a thousand machine-code programmers. All while looking at cats and playing minesweeper.
posted by anonymisc at 11:35 AM on September 24, 2013 [11 favorites]


Within 20 years cat pictures will be used as currency.
posted by Ad hominem at 11:36 AM on September 24, 2013 [6 favorites]


So among the jobs listed at very high probability of being doable by a computer, they include bakers, salespeople, roofers, various types of repair work and manual labor, umpires, data entry, and computer programming itself. Oh, also models. Somehow.

Have these people even met a computer?
posted by ook at 11:37 AM on September 24, 2013 [18 favorites]


This is why I roll my eyes any time a politician promises jobs jobs jobs! We gotta drop the whole "you are allowed to exist because you provide 'value'" stance.

We work every day to make it possible for more work to be done by fewer people. This idea that people who don't work must suffer is going to hurt us.
posted by ignignokt at 11:37 AM on September 24, 2013 [36 favorites]


47% of US jobs under threat from computerization

I bet that a computer came up with that figure.
Possibly with the help of big data.
posted by sour cream at 11:38 AM on September 24, 2013 [4 favorites]


I guess it's good in that humans will no longer be doing one of the worst, most soul-crushing jobs in the world. But don't fool yourself into thinking that the computerization of telemarketing would result in fewer telemarketing calls.

Who in their right mind would buy something from a computerized telemarketer? I see it being efficient in terms of calls being placed, but in terms of sales? I was working as a telemarketer over a decade ago and even as a real person I had trouble making sales to people who weren't sleep deprived, drunk, or idiots.
posted by Bulgaroktonos at 11:38 AM on September 24, 2013 [2 favorites]


We work every day to make it possible for more work to be done by fewer people. This idea that people who don't work must suffer is going to hurt us.

I agree, but not with your use of future tense. :-/
posted by anonymisc at 11:39 AM on September 24, 2013 [5 favorites]


Computers taking over their own data entry would look like this I guess
posted by ook at 11:39 AM on September 24, 2013


When I graduated from university with a Fine Arts degree twenty years ago, there weren't a lot of jobs in my city. There was government, education and healthcare, plus service industries. And that was it. And tourism, if you like selling t-shirts and hot dogs.

Over the past twenty years, though, entirely new industries have developed here in just this city, based on software, engineering, remote sensing, systems integration, gaming, e-commerce platforms, visual effects, rail, composites...

None of these industries existed here 20 years ago, but the only thing that has changed is that workers have be more highly educated to do those jobs, which typically pay well.
posted by KokuRyu at 11:42 AM on September 24, 2013 [1 favorite]


This is why I roll my eyes any time a politician promises jobs jobs jobs! We gotta drop the whole "you are allowed to exist because you provide 'value'" stance.

We work every day to make it possible for more work to be done by fewer people. This idea that people who don't work must suffer is going to hurt us.


Well, are we wanting a slow draw down of population naturally over the course of decades or a radical shift in the way the US public views the lower classes?
posted by Slackermagee at 11:44 AM on September 24, 2013


I wanted to comment on telemarketers but apparently my job (title examiner) is the very next thing down on the list. To which, my first reaction was "I'm not sure these people know anything about what I actually do." But then, lately the trend in this industry has been to replace high-quality/low-output professionals with high-output/low-quality wageslaves, and then to just not care about the error-prone results. So yeah, actually, the step from there to error-prone machines seems pretty likely.

In other words, it's not that replacing me with a machine is impossible; it's that replacing me with a machine and not getting some kind of Brazil-esque nightmare of a real estate industry as a result is impossible, and most likely The Powers That Be (big banks, mainly) will actually be all too happy to go that way in exchange for a quick buck.

If anybody wants me I'll be over here, coming up with a new career plan. And buying a boat.
posted by mstokes650 at 11:46 AM on September 24, 2013 [8 favorites]


Mathematician, my job, is at 0.047, tied with floral designer. I therefore compute that if I learn how to make flower arrangements on top of what I already do, there's only a 1 in 453 chance I'll become obsolete!
posted by escabeche at 11:46 AM on September 24, 2013 [8 favorites]


If anybody wants me I'll be over here, coming up with a new career plan. And buying a boat.

Good luck proving you own it.
posted by escabeche at 11:47 AM on September 24, 2013 [7 favorites]


Oh, also models. Somehow.

Given how much of a model's appearance is a product of Photoshop — because apparently even people in the top 0.1%ile of beauty, wearing outstandingly-crafted make-up and photographed by a professional, are still not beautiful enough for mainstream magazine covers — I think it's not that unlikely we'll just go 100% full-on Idoru in the next 5-10.
posted by en forme de poire at 11:47 AM on September 24, 2013 [11 favorites]


None of these industries existed here 20 years ago, but the only thing that has changed is that workers have be more highly educated to do those jobs, which typically pay well.

It's great that there are new industries but the total number of full time jobs has declined since twenty years ago. By an awful lot. So your point is ...weird.
posted by srboisvert at 11:48 AM on September 24, 2013 [4 favorites]


> Well, are we wanting a slow draw down of population naturally over the course of decades

um... "naturally"? What's that supposed to mean? "Naturally", populations tend to increase. The way that populations "naturally" decrease if there's a great deal of excess mortality - i.e. war, famine or plague. So which are you talking about?


> I'm sure that computerization will devastate the job market in the same way that the industrial revolution and the roboticisization have devistated [..] farming

I... think you're being snarky there, but surely you must be aware that the industrial revolution did, in fact devastate the farming industry. In 1910, more than half of Americans were rural - now it's down to 16%.

I've been predicting this for a long time, even on the blue, but it's not at all satisfying to see it work out in practice.
posted by lupus_yonderboy at 11:49 AM on September 24, 2013 [1 favorite]


And as you go forth today remember always your duty is clear: To build and maintain those robots.
posted by borkencode at 11:50 AM on September 24, 2013 [3 favorites]


Heh, so I used to work at an intellectual job. And much of my busy work could have been greatly simplified by clever computation. Instead, we had team meetings to come up with process documents that explained to you how to do your job in the most painful iterative process imaginable (because nothing says "Turing test -here I come!" quite like an eleven page document) and then, when you found things that didn't work (because the system used the simplest acceptable equation rather than actually using equations discovered by famous (and long dead) scientists, we had a series of emergency hand wringing session since we wouldn't be following our official processes.

I for one welcome our can find their ass I/O ports with both hands two registers overlords.
posted by Kid Charlemagne at 11:50 AM on September 24, 2013 [1 favorite]


My take on these articles is like chess computers:

computers can beat grandmasters, but (grandmaster + computer) dominates all, and makes for better chess.
posted by The Ted at 11:52 AM on September 24, 2013 [1 favorite]


That's my take on masturbation.
posted by cjorgensen at 11:55 AM on September 24, 2013 [1 favorite]


Good luck proving you own it.

I haven't heard of many wrongful foreclosures on peoples' boats. Though I suppose once they replace the Coast Guard with automated drones it'll be anybody's game.
posted by mstokes650 at 11:57 AM on September 24, 2013


Eh, while it's sort of amusing that they basically used big data techniques to try to estimate the probability that jobs would be made obsolete by big data (and other things), I consequently don't trust these results any more than I would trust a computer to actually replace a nuclear power reactor operator (p=0.95). Probability that Working Paper Author will be replaced by a computer? (Well, actually, "Survey Researcher" is at p=0.23, maybe that's similar.)
posted by advil at 11:57 AM on September 24, 2013


Oh, also models. Somehow.

Given how much of a model's appearance is a product of Photoshop — because apparently even people in the top 0.1%ile of beauty, wearing outstandingly-crafted make-up and photographed by a professional, are still not beautiful enough for mainstream magazine covers — I think it's not that unlikely we'll just go 100% full-on Idoru in the next 5-10.


OK, I'll accept that -- but that would require a whole team of creative human workers: character designer, animators, modelers, mocap, voice acting, not to mention probably dozens of coders and QA and etcetera.

Unless the phrase "after we invent hard AI, and the Singularity happens" appears somewhere in this report, consider me just a teeny bit skeptical.
posted by ook at 12:01 PM on September 24, 2013 [1 favorite]


Mathematician, my job, is at 0.047, tied with floral designer.

Except 'mathematician' isn't a job, it's more like a hobby or a lifestyle. If you are at a university, you get paid to apply for federal research grants and teach certify future "symbolic analysts." Neither of which is looking too good as a career right now.

But I guess there's always being a letter steamer at the NSA to tide you over...
posted by ennui.bz at 12:02 PM on September 24, 2013 [1 favorite]


47% SAME AS IN TOWN HA HA HA HA THANK YOU I WILL BE HERE ALL WEEK FOREVER FIXED THAT FOR ME HA HA HA HA THANK YOU
posted by not_on_display at 12:04 PM on September 24, 2013 [31 favorites]


The way that populations "naturally" decrease if there's a great deal of excess mortality - i.e. war, famine or plague. So which are you talking about?

In modern urban societies, population growth tends to fall below the replacement rate. This is the experience in Japan, most of Western Europe, and even the United States, which would be below replacement rate today were it not for immigration (fertility rate of 1.88 per woman).

I... think you're being snarky there, but surely you must be aware that the industrial revolution did, in fact devastate the farming industry. In 1910, more than half of Americans were rural - now it's down to 16%.

The farming industry is doing great, though. It's producing a lot more than it did in 1910; output has increased by probably a factor of five. Per-worker productivity has skyrocketed. And subsistence farming as a way of life in the United States has practically disappeared, which is a good thing.
posted by mr_roboto at 12:06 PM on September 24, 2013


The outliers in this are fun to poke at (watch repairer 99% likely... does anyone still do this?) but I can see some wisdom in the patterns. If I was a young and enterprising person (I'm neither any longer) I would look carefully over this list when plotting my future career. If I was a brilliant young person I would latch onto the careers likely to be computerized and pursue a cross disciplinary study of that career and programming to lead the charge and make out like a bandit.
posted by dgran at 12:12 PM on September 24, 2013 [1 favorite]


Republican members of congress? Simply object to everything that's logical or compassionate.
posted by jimmythefish at 12:12 PM on September 24, 2013 [1 favorite]


They categorized my job as "non computerizable". Good luck to all y'all mathematicians/floral designers.
posted by Peccable at 12:14 PM on September 24, 2013


There are people today who have jobs or even entire careers in fields that were unknown or unimaginable twenty or thirty years ago. I don't see this trend changing.

Also, just because a computer can do it, doesn't mean a computer can do it better.
posted by mrbill at 12:20 PM on September 24, 2013


It's great that there are new industries but the total number of full time jobs has declined since twenty years ago. By an awful lot. So your point is ...weird.

It's "weird" you had to add that little dig in there just because I stated an opinion that differed with yours.

" the total number of full time jobs has declined since twenty years ago" is empirically untrue, at least where I live.

Anyway, I think you'll find if you do the research that the crash of the housing sector in the States in 2008/09 wiped out a lot of jobs. For example, areas of the States that are currently experiencing high levels of unemployment, such as Imperial County, relied on the housing boom of the 00's to create jobs.

The solution, as always, is more education.
posted by KokuRyu at 12:22 PM on September 24, 2013


I guess it's good in that humans will no longer be doing one of the worst, most soul-crushing jobs in the world. But don't fool yourself into thinking that the computerization of telemarketing would result in fewer telemarketing calls.

But that's okay, because we'll have computers to answer the telemarketing calls for us.


The farming industry is doing great, though.

The farming industry is basically Cargill, Dow, and sharecroppers-on-their-own-land.
posted by Foosnark at 12:22 PM on September 24, 2013 [3 favorites]


I think it's not that unlikely we'll just go 100% full-on Idoru in the next 5-10.

ook: K, I'll accept that -- but that would require a whole team of creative human workers: character designer, animators, modelers, mocap, voice acting, not to mention probably dozens of coders and QA and etcetera.

Don't you think that once the designers/coders etc get done there will be a bunch of algorithms that grab new ideas & fashions of the web and start subtly altering Idoru to match the current trend? I'd be willing to bet that people are already working on categorizing a whole level of subconscious attributes we collectively think of as "beauty", or whatever one needs to be popular.

Also, it might not be very difficult to fool most of the people some of the time. It will be interesting to see what happens when there are multiple digital avatars representing competing commercial interests. I dread the day that happens, but it will be interesting.
posted by sneebler at 12:24 PM on September 24, 2013


It's great that there are new industries but the total number of full time jobs has declined since twenty years ago. By an awful lot. So your point is ...weird.

I was just going to make the same point. It's not necessarily a net good that there are new, exciting kinds of jobs, if those new kinds of jobs means there are fewer jobs overall. Since pretty much the entire reason people are looking to tech is to improve operational efficiency (which is more or less synonymous in business-speak with requiring fewer people to do the same amount of work), those new industries might not be quite the positive sign you're taking them to be.

You're ignoring that the whole argument for these new kinds of jobs is that they will ultimately yield improvements in overall business efficiency. If these new industries succeed, almost by definition, there will be fewer jobs overall (or else eventually these industries will die because who's going to keep paying for efficiency that never materializes).
posted by saulgoodman at 12:25 PM on September 24, 2013


Urban and Regional Planners, 0.13. Kindergarten teacher 0.15.

It's good to know that 5-year-olds will be taught by robots before I'm replaced.
posted by jimmythefish at 12:26 PM on September 24, 2013 [3 favorites]


Who in their right mind would buy something from a computerized telemarketer?

Not my new AI answering machine:

Good evening, [click] Mister [click] Homeowner of [click] Dutch Elm St., would you like to buy some [click] aluminum siding?
    I'm sorry I didn't understand that.
I'm sorry I didn't understand that.
    Well, I'm sorry I didn't understand that.
Well, I'm sorry but I still didn't understand that.
    I didn't understand that first.
Look, I still don't understand that.
    But you're not even trying to understand that.
You're mumbling.
    ~♫~'You're mumbling.'~♫~
You started it.
    You started it.
Real person.
    Real person!
REAL PERSON!!!
    REAL PERSON!!!
. . .
 
posted by Herodios at 12:28 PM on September 24, 2013 [6 favorites]


I actually automated my entire design department out of existence. I don't know how I feel about that, but I do know that at another job I have been trying my damnedest to make people's lives easier with automation and everybody is so scared of it messing up (even though the sum total of watching automation is less than doing the actual work) that most attempts have fallen flat.

People are very resistant to automation and it scares me. However my first sentence probably scares them, so I guess we're even.
posted by Brainy at 12:31 PM on September 24, 2013


0.099 Travel Agents: I would have thought this would be higher. I haven't used a travel agent for personal travel in quite some time - are there that many people working alongside William Shatner?
posted by skyscraper at 12:34 PM on September 24, 2013


If only humans could actually mine for bitcoins...
posted by Theta States at 12:36 PM on September 24, 2013 [2 favorites]


that would require a whole team of creative human workers: character designer, animators, modelers, mocap, voice acting, not to mention probably dozens of coders and QA and etcetera.

Sure, but once you have all that set up, they could probably crank out a whole bunch of simulated models. That's how they'd justify their existence and sell their product: if it took more engineers, making more money in total, to create synthetic models than it took to actually hire a human and then airbrush the hell out of him/her, then nobody would pay for the synthetic one. So if we assume that synthetic models will displace human ones, it necessarily means that there will be increased unemployment as a result.

And that's the lie behind the "well, everyone will just be robot mechanics / computer programmers / creatives!" narrative. If everyone who gets displaced by a robot gets to become a robot mechanic (or programmer, or "creative"), then there's no reason to ever bring in the robots in the first place.

In constant-demand conditions, there has to be either net headcount decrease, or decreasing skills / pay (high-wage artisan job becomes unskilled low-wage assembly-line job), or some combination of the two, in order for automation to have a positive ROI.

You don't replace a factory staffed by 1000 people with robots and then keep all 1000 people around as robot maintainers. If you did that, you could have just saved all that capital that you invested in robots. Nobody brings in robots (which are really just an extreme case of efficiency-improving machines) just for the fun of it.

During much of the 19th and 20th century this rather obvious truth was compensated by overall expansion of the economy (e.g., you can increase productivity and keep everyone employed if demand is constantly increasing). As long as you grow demand faster than productivity, everyone can move up that ladder and become machine operators and robot technicians or whatever. But you clearly can't keep doing that forever.
posted by Kadin2048 at 12:37 PM on September 24, 2013 [9 favorites]


659. 0.97 39-3021 Motion Picture Projectionists

Damn you, modern world.
posted by Annika Cicada at 12:38 PM on September 24, 2013 [1 favorite]


Surely by automating more tasks, we can free our citizens from the drudgery of pointless makework and enable them to live fuller, happier lives!

Or, we can put them all out on the street I guess, whatever


don't you understand, everyone has to work even if there's nothing to do, otherwise Contentment
posted by threeants at 12:39 PM on September 24, 2013 [1 favorite]


I like that the article (when I see it, anyway) is prefaced with a google ads sidebar:

Payroll Training Program - www._______________.ca
Accounting and Payroll Programs. Get your Diploma Fast. Free Info!

Second Career Ontario - www._______________________.com
A government program to help pay for your education for a new career

Attract Money - www.___________________________.com
Learn to Jump Into Another Universe Where You Have Money

A Self-Publishing Success - www._____________.com
100% Canadian owned. Learn How to Self-Publish your book.


I wonder if these correlate at all with the jobs likely to vanish in a decade. That said, this quantum universe shifting deal is something my guidance counselor in school never once mentioned to me, the bastard.
posted by ricochet biscuit at 12:43 PM on September 24, 2013 [2 favorites]


People are very resistant to automation and it scares me. However my first sentence probably scares them, so I guess we're even.

I don't really think whatever your fear about their resistance to having their ability to feed and house themselves removed is really is equivalent to the fear that engenders that resistance at all.
posted by Steely-eyed Missile Man at 12:45 PM on September 24, 2013 [2 favorites]


I predict that within the next 10 years most office computers will spend 4-6 hours a day looking at pictures of cats and playing minesweeper.

More likely they'll alternate between looking at pictures of Minesweeper and saying, "What a cute little program!" and prodding cats and other mammals to run around doing pointless things for their amusement.
posted by straight at 12:45 PM on September 24, 2013


So let's see,computerised everything.That's doable. Problem is, how will the masses earn a living, if their participation is not only no longer needed, but is also an obstacle or is perceived as such?
posted by elpapacito at 12:52 PM on September 24, 2013


and prodding cats and other mammals to run around doing pointless things for their amusement.

What, cats won't be replaced? Surely cats will be left to sit around and...uh...do...oh wait. Yeah. Nevermind.
posted by jimmythefish at 12:53 PM on September 24, 2013


Breakthrough!: Program all machines to treat humans as cats. Leave a steak and salad out three times a day, a bowl of beer, a patch of sun. Done.
posted by jimmythefish at 12:54 PM on September 24, 2013 [5 favorites]


More likely they'll alternate between looking at pictures of Minesweeper and saying, "What a cute little program!" and prodding cats and other mammals to run around doing pointless things for their amusement.

And there is some evidence that this has already happened.
posted by straight at 12:54 PM on September 24, 2013 [1 favorite]


So even though there is still a person at the helm, that person is doing the work of a thousand machine-code programmers.

Meh, not really. The machine-code jobs still exist and actual human systems programmers still get hired to write them. The difference between then and now is that a lot of the low-level code that was designed decades by humans is still good enough to serve as an architecture to build higher-level code on top of. The same way that highways in a country allow long haul truckers to do their jobs, without negating the need for people to be hired to build new highways or maintain and improve the ones that already exist.
posted by burnmp3s at 12:55 PM on September 24, 2013


Breakthrough!: Program all machines to treat humans as cats. Leave a steak and salad out three times a day, a bowl of beer, a patch of sun. Done.

If they're smart, they'll also have us de-clawed and spayed.
posted by straight at 12:56 PM on September 24, 2013


373. 0.67 33-9092 Lifeguards, Ski Patrol, and Other Recreational Protective Service Workers
379. 0.68 29-2021 Dental Hygienists


I am a luddite, I know, but I am not sure that everyone will be entirely eager to have their visits to the dentist include a multi-armed stainless steel machine scaling their teeth. And perhaps it is a failure of imagination on my part, but I am having difficulty conceiving of what degree of computer intelligence would be required to even recognize someone drowning at a beach among hundreds of people splashing around, let alone what remedies it could provide to the situation.
posted by ricochet biscuit at 12:57 PM on September 24, 2013


In a sense, it's absolutely true - programming languages have been getting higher and higher level since day one. So even though there is still a person at the helm, that person is doing the work of a thousand machine-code programmers. All while looking at cats and playing minesweeper.

Eh, lisp has been around for a long time, and it's pretty high level. The more accurate statement is that hardware has reached the point where we can actually run high-level languages. The issue is that running high level languages doesn't actually make complex processes simple; it just makes the bit juggling that you have to do in order to model those complex processes less obnoxious.

Most of the time I spend is on grotesque business logic issues that boil down to really long chains of boolean logic.
posted by sonic meat machine at 12:58 PM on September 24, 2013


I am having difficulty conceiving of what degree of computer intelligence would be required to even recognize someone drowning at a beach among hundreds of people splashing around, let alone what remedies it could provide to the situation.

Remember when Neo gets pulled out of the Matrix after unplugging? That thing.
posted by jimmythefish at 12:58 PM on September 24, 2013 [1 favorite]


I was all prepared to scoff at this paper but it is actually a moderately interesting technique. They picked 70 occupations that "everybody knows" can or can't be automated (tax preparers yes, dentists no). Then they used a classifier to figure out the "goodness of fit" for every one of the other jobs.

Of course the actual value of this information to an economist is minimal. For one thing, they didn't include any information about the number of people participating in a job at all -- I mean, I'm sure there will be at least one tax preparer in the country in 100 years, and the number of emergency management directors today has to be small enough to have zero effect on the labor market. IT has the capacity to absorb an enormous fraction of the labor force -- all those new automated systems have to be developed, marketed, purchased, maintained, etc, and that only accounts for a small proportion of the job categories in the list.

So I think the message here is more about an interesting maybe-useful way to predict things that are going to happen, and less about the specific results of that prediction method.
posted by miyabo at 1:01 PM on September 24, 2013


32. .0065 Computer Systems Analysts
293 .48   Computer Programmers
Does anybody know what that is supposed to mean? Does somebody still think that there are "analysts" who can write unambiguous, correct program specs, and that there are "programmers" whose job it is to convert these specs into code?
posted by mr vino at 1:07 PM on September 24, 2013 [2 favorites]


> Hello, I am Eliza.
* Hello Eliza, I am not_on_display.
> Did you come to me because you are not_on_display?
* No, I came to you because 47% of US jobs are under threat from computerization.
> I see. 
* Doesn't this concern you?
> You're not really talking about me, are you? 
* I am totally talking about you!
> Do you believe it is normal to be totally talking about me? 
* Totally.
> OK... "I are totally talking about you". Tell me more. 
* 47% of US jobs are under threat from computerization!
> That is quite interesting.
posted by not_on_display at 1:08 PM on September 24, 2013 [13 favorites]


jimmythefish: " let alone what remedies it could provide to the situation."

I'm thinking life guard turrets mounted along the beach equipped with retractable laser guided harpoons.
posted by Hairy Lobster at 1:10 PM on September 24, 2013 [2 favorites]


People are very resistant to automation and it scares me.

This is not my experience. I've found people to be stupidly, hilariously receptive to automation, when it's framed as "oh hey this makes my life so much easier!" ... without thinking too hard that anything that makes your job "easier" almost certainly makes you, as an employee, that much easier to replace. You do not ever want your job to become suddenly easier.

Also, very few jobs are suddenly eliminated wholesale by machines. When it happens, it's usually notable. Much more often, automation is a slow process by which one person slowly ends up doing more than what would have been one person's worth of work. Not in whole-person multiples, but little by little: a 100-person department gets a new computer system, and slowly becomes a 90-person department. If desired, you can do that over time without actually even firing anyone: you just decrease headcount through attrition and don't hire anyone new, spreading the workload around on the remaining employees (if you can force them to actually do more work just on their own initiative, bonus). The only people who really get hurt are those who would have gotten hired and now don't.

I am having difficulty conceiving of what degree of computer intelligence would be required to even recognize someone drowning at a beach among hundreds of people splashing around

It's probably not realistic for machines to completely replace humans 100% of the time in many jobs. But that doesn't mean that the numbers of people working in that job couldn't be decimated.

Just off the top of my head, in terms of lifeguards: rather than having one highschool kid sitting on a wood chair for every 200 yards of beach (or whatever they can cover just with their eyeballs), instead you could have a network of camera buoys, or just cameras on towers, monitored by one person (perhaps with the help of software to keep a count of people in the view of each one), with another person out on a jet-ski and a radio, waiting to get dispatched by the controller. Two people could probably cover a lot of beach that way. If they could cover, say, 5x as much as a single person, then you have 3 FTEs of headcount decrease with which to justify the cost of the system, less whatever pay the controller and jet-ski guy demand for their additional skills (maybe not much, since there will be 3 now-unemployed lifeguards gunning for their jobs!).

Again, it's not like you need to go from "done by a person" to "done by a machine" to be disruptive. You only need incremental, continuous efficiency/productivity improvements, and if they exceed growth in demand than you are going to have a labor surplus.
posted by Kadin2048 at 1:10 PM on September 24, 2013 [10 favorites]


jimmythefish: "Breakthrough!: Program all machines to treat humans as cats. Leave a steak and salad out three times a day, a bowl of beer, a patch of sun. Done"

Add TV and you've got yourself a deal!
posted by Hairy Lobster at 1:12 PM on September 24, 2013


I believe the folks who fool themselves into thinking that increasing automation/computerization in the workplace will just magically produce an equal number of other jobs either believe in endless growth like the free market fetishists do, or they're just going through the first Kubler-Ross stage. Once we as a society move past Denial, we may reach Anger. When the revolution fails we'll Bargain with our capital-owning masters for whatever subsistence-level wages they'll give us for our back-breaking labor, then move through Depression to Acceptance of our new status as cogs in the machine.

Population growth does seem to be plateauing in industrialized nations, but I don't think the slight negative rate will be able to keep up with the exponential rate of increasing, machine-driven efficiency.
posted by Thoughtcrime at 1:12 PM on September 24, 2013 [3 favorites]


Except 'mathematician' isn't a job, it's more like a hobby or a lifestyle. If you are at a university, you get paid to apply for federal research grants and teach certify future "symbolic analysts." Neither of which is looking too good as a career right now.

AND THAT IS WHERE MY CUNNING FLOWER ARRANGEMENT PLAN COMES INTO PLAY
posted by escabeche at 1:17 PM on September 24, 2013 [2 favorites]


increasing automation/computerization in the workplace will just magically produce an equal number of other jobs

Hmm. I do believe this, because there will always be someone willing to hire someone else if the price is low enough. The problem is that we may have 5% of the population designing the robots, writing the software, and building the factories, and 95% of the population working for minimum wage at the fewer and fewer jobs that are not easily automated. That is why increased efficiency has to go hand in hand with better redistribution of the proceeds of that efficiency.
posted by miyabo at 1:18 PM on September 24, 2013 [1 favorite]


Oh, also models. Somehow.

Hi.
posted by ymgve at 1:20 PM on September 24, 2013 [2 favorites]


ook: So among the jobs listed at very high probability of being doable by a computer, they include bakers, salespeople, roofers, various types of repair work and manual labor, umpires, data entry, and computer programming itself. Oh, also models. Somehow.

Have these people even met a computer?


Bakers: Robotic bakeries. Many products could be made fresh, eliminating a lot of lossage due to stuff being made, not sold, and discarded.

Salespeople: Greater emphasis on internet sales, which are depersonalized. When people are completely used to buying things from websites and not from dealing with salespeople at all, business sales may start to occur the same way.

Roofers: Better manufacturing techniques and increased labor costs resulting in more pre-fabbed roof designs that require less manpower to install and less maintenance.

Repair work: Many things becoming un-repairable and disposable, like Apple devices are now.

Umpires: Visual recognition systems and teleconferencing. You'd probably still need a few people to oversee the process, but not nearly as many.

Data entry: Better scanning and OCR, better and automated import and export of data between systems. Data entry being stuck on the consumer or customer.

Computer programming: Better programming tools that allow fewer programmers to do more work, and eventually, tools that allow designers to make programs directly.

Models: CGI, obviously.
posted by Mitrovarr at 1:26 PM on September 24, 2013 [1 favorite]


Also,

OK, I'll accept that -- but that would require a whole team of creative human workers: character designer, animators, modelers, mocap, voice acting, not to mention probably dozens of coders and QA and etcetera.

You have to remember that after the first virtual model has been designed and built, the framework is in place and people are familiar enough with the tools that they could probably crank out a new model each month, maybe each week.

In addition, once you have a virtual model, you don't have to schedule photo shoots and compete for time. One virtual model could do the work of hundreds of real models.
posted by ymgve at 1:27 PM on September 24, 2013 [1 favorite]


In my opinion the one thing a human can do that a machine can't is suffer. When I want to feel powerful I make someone suffer. I can't do that to my Dell Optiplex, lord knows I try.
posted by Teakettle at 1:28 PM on September 24, 2013 [6 favorites]


I fear for the day when the management caste is staffed with computer-savvy young people, instead of completely snowed Baby Boomers who believe you when you say it'll take you 3 hours to do that thing they asked you to do.
posted by overeducated_alligator at 1:29 PM on September 24, 2013 [7 favorites]


there will always be someone willing to hire someone else if the price is low enough

If this were true, and it might arguably be so in some sort of optimal, frictionless economists' universe, then you'd never have any actual unemployed people. Since we do have unemployed people, and historically there have been periods when there have been lots -- to the point of social disruption, famine, etc. -- of unemployed people, I suspect that it's not really the case.
posted by Kadin2048 at 1:31 PM on September 24, 2013 [1 favorite]


After a quick scan I didn't see copywriters anywhere on the list, so I'll assume we are at minimal risk of automation along with "Authors" and "Graphic Designers". Aww yeah, whose English degree is worthless now, huh?!
posted by jess at 1:44 PM on September 24, 2013 [2 favorites]


I am somewhat comforted by observing that this site -- telling about how computers will soon eliminate many careers that involve cognitive tasks -- is itself unable to keep track of what kind of JOBS it's talking about when it's time to deliver Related Articles.
 
posted by Herodios at 1:45 PM on September 24, 2013


Eh, lisp has been around for a long time, and it's pretty high level.

I would assume that even Lisp is orders of magnitude "higher" now than 20 years ago, if only because of things like growing libraries of libraries. Using a library can make each keystroke do the work of a thousand keystrokes.
posted by anonymisc at 1:54 PM on September 24, 2013


By the very nature of the work, automation freezes processes into a technological stasis. For a lot of tasks, this will never be a problem and is actually desirable, but there are also a number of areas where there is simultaneously an extremely strong business case to be made in favor of automation but also the unavoidable reality that once the investment (often massive) in automation is made, changing directions becomes a tremendous risk-the-business level gamble.

Some companies will figure out how to roll with this, but others won't and they will stagnate. This is one of the new kinds of post-automation opportunities: doing it better than the last guy. I'm no economist, but it's hard to imagine that this is sufficient to replace lost jobs.
posted by feloniousmonk at 1:58 PM on September 24, 2013 [3 favorites]


After a quick scan I didn't see copywriters anywhere on the list, so I'll assume we are at minimal risk of automation along with "Authors" and "Graphic Designers". Aww yeah, whose English degree is worthless now, huh?!

Question then becomes, how many of these can the market support? Main threat to authors (and possibly to graphic designers) are the huge numbers of people who are willing to give it away for free, and the huge number of people who feel that they need not pay for digitized creative work.

So the future is either working at Trader Joe's or selling handmade stuff on Etsi. So, at least, it would appear from a sampling of people I know whom automation and downsizing has hit hard.

Prepare for deflation.
posted by IndigoJones at 2:12 PM on September 24, 2013 [2 favorites]


Computer programming: Better programming tools that allow fewer programmers to do more work, and eventually, tools that allow designers to make programs directly.

As mentioned earlier in the thread, this has been the holy grail of programming since programming was invented, and we're not any closer now than we were decades ago. For example, web programming has been around for a while and people have tried to make WYSIWYG editors that allow someone with no technical knowledge to make websites, but if anything it's harder now than it was in the early days of the web for novices to design their own site without just using a pre-made template.

Unless there is a major breakthrough in AI just around the corner, the only way that there are going to be less programming jobs is if less code needs to be written (the templates mentioned above are an example of this, less people write their own website because they just use Facebook or Tumblr and don't need to design their own special code).

I would assume that even Lisp is orders of magnitude "higher" now than 20 years ago, if only because of things like growing libraries of libraries. Using a library can make each keystroke do the work of a thousand keystrokes.

I don't use Lisp specifically but I've been writing applications for almost 20 years now and things have really not changed that much. Writing a new-fangled iPhone or Android app involves using Objective C or Java in ways that are pretty much exactly the same that an Objective C or Java developer wrote apps in the 80s or 90s. The libraries and whatnot are different than they used to be and include different types of technology that didn't exist back then, but they are not orders of magnitude higher in level, just slightly nicer and more powerful. You still have define all the same sorts of fiddly crap and write the same kinds of routines to do the same kinds of stuff. If there were really huge advancements in the core aspects of programming, then programming languages and core libraries from decades ago probably wouldn't still be used all over the place for new projects. As it is, programming is still pretty much programming, and nobody has found a way to improve it enough to be worth giving up the old tools.
posted by burnmp3s at 2:15 PM on September 24, 2013 [1 favorite]


Bakers: Robotic bakeries. Many products could be made fresh, eliminating a lot of lossage due to stuff being made, not sold, and discarded.

But that is a perfect counterexample. We already have robotic bakeries -- well, factories that use machines to turn out unbelievable quantities of perfect baked goods at staggeringly low cost. It turns out that people who have some disposable income -- white collar workers like myself -- would prefer to spend $5 on a handmade loaf of bread even when a machine could make one just as good for $1.
posted by miyabo at 2:19 PM on September 24, 2013 [1 favorite]


watch repairer 99% likely... does anyone still do this?

Most of them, along with similar jobs like shoe repair, are actually fronts for professionals who provide new identities for criminals in the lam. Or so the manager at my local Cinnebun told me recently.
posted by Bunny Ultramod at 2:24 PM on September 24, 2013 [1 favorite]


Consider an example of human bias: Danziger, et al. (2011) demonstrate that experienced Israeli judges are substantially more generous in their rulings following a lunch break. It can thus be argued that many roles involving decision-making will benefit from impartial algorithmic solutions.


Just machines to make big decisions
Programmed by fellows with compassion and vision
Well be clean when that work is done
We'll be eternally free, yes, and eternally young
What a beautiful world this will be
What a glorious time to be free . . .
 
posted by Herodios at 2:30 PM on September 24, 2013 [1 favorite]


miyabo: But that is a perfect counterexample. We already have robotic bakeries -- well, factories that use machines to turn out unbelievable quantities of perfect baked goods at staggeringly low cost. It turns out that people who have some disposable income -- white collar workers like myself -- would prefer to spend $5 on a handmade loaf of bread even when a machine could make one just as good for $1.

Well, you have to admit there are fewer bakers now than there once were. I merely postulate this trend continuing; I think you could build a good small-scale auto-bakery that could make baked goods rapidly on demand, which will eventually be cheaper than a person doing it. That would compete effectively with human bakers, and reduce their numbers further. Combine that with the sharp reduction to the middle and upper middle classes that everyone can see is coming, and I don't think there will be a ton of bakers in the future. And the few that remain won't be there because society needs bakers, it'll be because they amuse the rich. Which is the way a lot of the 'artisan' fields will go.
posted by Mitrovarr at 2:42 PM on September 24, 2013


I can see the direct-to-garment inkjet printers in my rearview mirror, but I hope to be retired before they get as fast as a screen printing press. That, and I know how to run one at least, so my $10.00 per hour future is secured.
posted by Devils Rancher at 2:49 PM on September 24, 2013 [1 favorite]


Writing a new-fangled iPhone or Android app involves using Objective C or Java in ways that are pretty much exactly the same that an Objective C or Java developer wrote apps in the 80s or 90s. The libraries and whatnot are different than they used to be and include different types of technology that didn't exist back then, but they are not orders of magnitude higher in level, just slightly nicer and more powerful.

I suspect you have acclimatised - those phone apps are doing a whole lot more under the hood. If we tried to create a modern app in all its functions, using what was available to us 20 years ago, with only the time and budget it would need today, I think you'd be in for a rude awakening. I'm guessing we wouldn't even get past stuff like building basic building-block crap after discovering that the network drivers lack all the features that today are taken for granted like the air we breath.
In my little sphere at least, it's orders of magnitude.
posted by anonymisc at 2:54 PM on September 24, 2013


As mentioned earlier in the thread, this has been the holy grail of programming since programming was invented, and we're not any closer now than we were decades ago.

The funny thing is, we're probably even further than we were before. 15 years ago, I think VB6 and Access were about halfway there in terms of a programming environment that could be used by less sophisticated users. Even earlier than that, there was stuff like Hypercard. Those tools haven't really been replaced and they've become obsolete.

I think the reason is that because the experiments weren't 100% successful at solving the problem, the general consensus became that the problem was not solvable. I don't think that's the correct conclusion.
posted by zixyer at 3:25 PM on September 24, 2013 [1 favorite]


The funny thing is, we're probably even further than we were before.

Yeah, in 8th grade I was taking Spanish and I made a quick Hypercard stack for Spanish verbs. You'd enter a root word, and it would automatically create cards for all the regular conjugations and let you override the irregular ones. Then it printed out double-sided flash cards. Today I'm a professional programmer and, while I could do something like that, it would take me probably a week of full-time work.
posted by miyabo at 3:35 PM on September 24, 2013 [3 favorites]


I don't see the problem here, other jobs will be created, like it always has when technology advances.

I hope not, MisantropicPainforest! What jobs have we created recently?

We've vastly expanded administration and management positions, eroding funding for folks who actually do important work, like university lecturers and professors. Afaik, these new administration positions are inherently vulnerable to automation partially because they're so trivial and useless.

We vastly expanded law enforcement and the surveillance state. In fact, all discretionary spending categories have declined relative to GDP since the early 70s, except law enforcement which contributed substantially to the ballooning federal budge, and uniformly negatively impacted all our lives.

There are new jobs that actually contribute to society, obviously the technical jobs, but also cultural jobs like VJs, yoga teachers, etc., but they're relatively few. We'd accomplish more by giving everyone an extra day or so off per week so that they can engage in constructive hobbies instead of playing on facebook.

Automate administrative and managerial positions as quickly as possible. Also, demonstrate the expansion in law enforcement hurts the economy, ditto many governmental administrative positions. Address unemployment by shortening the legal work week, just like we did last century.
posted by jeffburdges at 3:37 PM on September 24, 2013 [2 favorites]


15 years ago, I think VB6 and Access were about halfway there in terms of a programming environment that could be used by less sophisticated users. ... Those tools haven't really been replaced and they've become obsolete.
Tell that to all the businesses that still use Access daily.

Please.
posted by aw_yiss at 3:46 PM on September 24, 2013 [1 favorite]


Most of the time I spend is on grotesque business logic issues that boil down to really long chains of boolean logic.

94% of the time I spend is strategically putting more bandaids on a giant pile of poorly planned, poorly written legacy code.
posted by Foosnark at 3:48 PM on September 24, 2013


I'd expect fashion designers should score up around musicians' 0.074, not the 0.021 they got. Aren't machines already simplifying many fashion designer tasks? There are major subcultures rejecting the whole "authority" model fashion designers thrive upon. These two trends should collide with interchangeable inexpensive ready-to-wear fashion elements that replace designers, well beyond musicians replaceability now. Anyone with a designers eye could maintain blog readership though I guess.
posted by jeffburdges at 4:15 PM on September 24, 2013


Don't you think that once the designers/coders etc get done there will be a bunch of algorithms that grab new ideas & fashions of the web and start subtly altering Idoru to match the current trend?

No. No, I do not. "Grabbing ideas" and "matching trends," like many of the other things various people are suggesting in this thread, are exactly the sort of thing that computers can't do at all.
posted by ook at 5:45 PM on September 24, 2013


I'd expect there to be 3 categories of work in future: artists, scientists and decision makers.

And anyone working as any of these 3 would be really really good at what they do. They would be highly educated and very skilled. and there would be very few of these jobs.

Say 100 people know a particular skill. If we rank them in decreasing order of expertise, as automation and robotics become more and more prevalent, fewer and fewer percentage of those 100 will be employed.

We are already working on robots who can hold/manipulate things and software/hardware which can see things. As these are perfected and pass a certain level of competence while being cost effective, jobs will increasingly be switched. Its a one way switch, once something can be done cost effectively by a machine and if you can replace 10 people with 1 machine and 2 humans, it will be done. Unless those machines are destroyed, that job is not coming back to 10 humans. Sure we will create new skills, but machines would be able to catch up much faster with humans and it will be obsolete pretty quickly.

Given this, it would mean that most of the population would be unemployed and would be living on dole/support doing whatever they want to do, pursue hobbies or just waste time. In case a society is not willing to provide the dole and assistance, well, poor unemployed ppl will die.

On the other hand, the skilled and employed would become pretty expensive to employ and hence rich. The inequality would increase. With this inequality, most of us would not be able to buy things unless the prices were decreased a lot.

Elysium is not so far-fetched actually.
posted by TheLittlePrince at 6:32 PM on September 24, 2013 [1 favorite]


Of course the actual value of this information to an economist is minimal.

I get the impression that most information is of little value to economists. ([Heads over to the Popular Scienccee thread...]

aw_yiss: 15 years ago, I think VB6 and Access were about halfway there in terms of a programming environment that could be used by less sophisticated users. ... Those tools haven't really been replaced and they've become obsolete.

Tell that to all the businesses that still use Access daily.


That doesn't have much to do with Access as a tool though, does it? It's just that people like to play with databases, and then they think, "Hey, I could solve this problem." A couple of years later your business is littered with half-baked, unsupported Access "solutions", some of which have acquired an importance far beyond their original intent. Apparently one of our departments found 120-odd undocumented Access apps being used after their creator retired. Which is why I had to sigh a declaration in BLOOD before I was allowed to have it installed on my machine, that I wouldn't create any applications.

I used to be one of those people, and now I'm much more in line with "No, we need to follow the corporate IT rules & process, because even if it's no fun in the short term it's going to be way wore effective over several years to the lifetime of our existing environment."

Nevertheless, I see things like Access as Job Creators.
posted by sneebler at 6:48 PM on September 24, 2013 [2 favorites]


The funny thing is, we're probably even further than we were before. 15 years ago, I think VB6 and Access were about halfway there in terms of a programming environment that could be used by less sophisticated users. Even earlier than that, there was stuff like Hypercard. Those tools haven't really been replaced and they've become obsolete.

I think the reason is that because the experiments weren't 100% successful at solving the problem, the general consensus became that the problem was not solvable. I don't think that's the correct conclusion.


Programming is easy. Asking the right question is hard.
posted by ryoshu at 9:12 PM on September 24, 2013 [1 favorite]


What's going to stop all these millions of unemployed, bored, possibly impoverished people from revolting?

Paying half of them to keep a boot on the other half?

What the fuck kind of life is that?
posted by This, of course, alludes to you at 10:48 PM on September 24, 2013 [2 favorites]


To take this in a somewhat different direction, what ryoshu is saying is particularly true for machine learning. Machine learning methods are insanely powerful, but if misapplied, will happily give you a perfectly correct solution for the wrong problem. How exactly you train and evaluate the model hugely affects what kind of results you get (particularly for harder problems), so you need some domain-specific expertise to make sure you're getting more than the equivalent of word salad out the other end.
posted by en forme de poire at 10:52 PM on September 24, 2013


What's going to stop all these millions of unemployed, bored, possibly impoverished people from revolting?

Initially a shorter work week, but later a basic income.

Paying half of them to keep a boot on the other half?

I hope not! At the moment, we're pursuing some middle ground where all increases in federal spending go to either the voting elderly or jackboot lickers, ala law enforcement, the surveillance state, etc., while work time is increasingly replaced by monitored internet time. We need to break both the social prejudice against doing anything that benefits everyone and well as the spending on jackboots.
posted by jeffburdges at 1:19 AM on September 25, 2013 [2 favorites]


I disagree with your artists, scientists and decision makers list, TheLittlePrince.

Art, music, etc. aren't necessarily going to remain paid employment indefinitely because technology should make hobbyist art good enough. Already happening in music.

We'll hopefully replace much arbitrary decision making with more evidence based tools, but currently entrenched corrupt interests still retains a tight hold, ala war on drugs, global warming, etc. We're still developing those evidence based policy tools though, so your decision maker class might change radically down the line. In most countries, local corruption creates an enormous drain on the economy, so one could imagine replacing local politics with demarchy, meaning legislation by jury trial with evidence based advocates. I donno if this eliminates public policy jobs, well someone must collect the evidence and make arguments, but you'd hardly call them "decision makers" anymore. Also, we've created host of "petty laws and useless officials" via bloated administrations that might not qualify as "decision makers" now but do wield some power. We might eliminate these jobs by automating the modicum of real work they actually do. In any case, we need political progress to eliminate decision maker jobs, but automation facilitates that.

Is there any limit on the amount we can spend on engineering and science besides maybe the educational system. Won't they just invent and discover stuff faster the more money you give them? In these fields, automation just increases the amount that gets done without engineers or scientists even noticing if they budgets allocate more for machines and less for personnel. I'm therefore unsure it even makes sense to talk about automation taking jobs here. In the long run, if the work week drops to one day or less, then science might become simply a hobby, albeit with government funding for equipment, which arguably means all science jobs disappear, but that's still a political response to automation elsewhere, not automation itself.
posted by jeffburdges at 3:51 AM on September 25, 2013


457. 0.83 47-2221 Structural Iron and Steel Worker

I suppose that for new construction we'll see more pre-fab buildings going up but a lot of the maintenance work that I've done in the trade up to this point would be incredibly hard to automate. A welding robot is pretty huge right now and has to be setup in a certain place and programmed to do certain welds. A lot of the work I do isn't in the same place or regular in any way whatsoever. Layout work is getting a lot easier with lasers and gps.

I've worked with machines that weld rails together and that is a lot faster and better than doing it manually. I imagine as robots get better we will be working with them more and more and the job will migrate toward moving the robots and setting them up in the right places. Maybe one day instead of actually welding myself, I'll just draw a line for the robot to lay down a perfect weld faster than I could. I think it'll be a long time before there are robots wizzing around the coke overs searching for and repairing all the cracked welds like some sort of welding roomba.

Long ago before they used bolts to connect things, there were 3 or 4 guys that would work together to rivet every single connection in a building. Hi-tensile bolts and elevated work platforms really tanked the number of people needed to assemble structural steel but I can't see automation making humans redundant for a long time.
posted by glip at 6:54 AM on September 25, 2013


I expect art to remain a paid profession as its basically dependent on human whimsies. There would be software that could prepare a work of art, but given our tendencies in past, just the fact that it has been prepared by a software would affect how we enjoy it and what all we see in it. Its like a bakery as mentioned up thread. Even though machines can be used in chain bakeries to produce bread as good and hygienic as anything, people are still able to charge higher for hand made products.

Regarding scientists, i expect it to be a paid profession as i don't think machines would be as intelligent as humans in near 50 years. Though they would use different methods to resolve problems. A machine might just try out all 100 Mn combinations to identify the right answer while a human uses more pattern recognition and intuitive methods. I think it will be more cost effective to use humans for original research for quite some time.

And decision makers would still be humans (whether figureheads or not) because we humans like that illusion and the idea of being ruled by a machine is still hard to accept. I agree a lot of decisions would be mechanical and through rote but we would need a change in societal thought process before machines become acceptable as decision makers.
posted by TheLittlePrince at 7:02 AM on September 25, 2013


I suspect you have acclimatised - those phone apps are doing a whole lot more under the hood. If we tried to create a modern app in all its functions, using what was available to us 20 years ago, with only the time and budget it would need today, I think you'd be in for a rude awakening.

"All its functions" is kind of cheating because a lot of Internet-centric functions didn't really exist 20 years ago. We have libraries now that we had no need for back then because there are brand-new types of problems that need to be solved. My point is less that there aren't any new libraries for new technologies, it's that programming with the same technologies that existed before isn't any easier because of great new libraries that made those kinds of tasks trivial. For example, an Android app I use that would have been made pretty much the same way 20 years ago is this Kitchen Timer. I could take VB3 from 20 years ago and I would do most of the same tasks (working with the system clock and events, laying out widgets on the screen, dealing with and saving user settings, etc.) that I would do writing a modern app on any platform.

I'm guessing we wouldn't even get past stuff like building basic building-block crap after discovering that the network drivers lack all the features that today are taken for granted like the air we breath.
In my little sphere at least, it's orders of magnitude.


I'm not a networking expert but again, it's not magnitudes easier to do the kinds of networking tasks I have done. Back in the 90s, if I had to invent or implement some kind of protocol I would use something socket level like WinSock. It would involve tasks like setting up code to open a socket, sending or receiving data, parsing the data according to how the protocol worked, keeping track of state, etc. Back then I wrote an FTP client from scratch on the socket level because there was no built-in FTP support in the language I was using and all of the existing FTP libraries cost money that I didn't have as a hobby programmer. These days there are a lot of free libraries for all sorts of protocols, including FTP, SMTP, web services, etc. But recently I had to implement a client for a proprietary protocol that no libraries exist for, and it's almost exactly the same sort of work that I did back in the 90s. The only way you save on writing new code is if the exact problem you want to solve has already been solved by someone else and is available to you. Which is the same way it was 20 years ago. Nobody has made the core part of programming, coming up with novel solutions to novel problems, any easier.
posted by burnmp3s at 7:08 AM on September 25, 2013


Correspondingly, I find the notion that SDN will make network engineers obsolete to be giggleworthy. SDN means we're just about able to keep up with the new tech being rolled out (lateral networks, iSCSI, BYOD, UTM) by abstracting some of the scut work.

Aww, who am I kidding? It means we'll also need to be able to go over solution stack source code to figure out why and how the devs managed to put three thousand VLANs on an eight port switch.
posted by Slap*Happy at 7:45 AM on September 25, 2013


I expect art to remain a paid profession as its basically dependent on human whimsies. There would be software that could prepare a work of art, but given our tendencies in past, just the fact that it has been prepared by a software would affect how we enjoy it and what all we see in it.

Not to be harsh, but you're not a musician, are you? You may not see the machines nipping at your heels yet, but musicians are, as a class, a type of artist which has been largely made redundant by the computer. Not that there aren't still some musicians making a living wage, but compared to the 70's and 80's, it's a wasteland out there. If the public is perfectly happy with computer-generated music, don't kid yourself about computer-generated art. In fact, all commercial art is now created on a computer, which has cut the hours per piece, and the attendant costs and profits, down to fractions of their former selves. Some folks are already perfectly happy to hang prints of fractal visualizations, and consider them art. No one is really immune in the arts at this point.
posted by Devils Rancher at 11:57 AM on September 25, 2013


all commercial art is now created on a computer

A lot of people here seem to be confusing "using the computer as a tool" with "being replaced by the computer".
posted by ook at 12:40 PM on September 25, 2013 [2 favorites]


It turns out that people who have some disposable income -- white collar workers like myself -- would prefer to spend $5 on a handmade loaf of bread even when a machine could make one just as good for $1.

Well, you have to admit there are fewer bakers now than there once were.


Also, there is gigantic difference between a nice sourdough or something and a loaf of Wonderbread.
posted by Steely-eyed Missile Man at 2:05 PM on September 25, 2013


There are few machines that completely remove all humans form the equation, ook and TheLittlePrince. We've mostly discussed situations when one worker with a new tool replaces numerous workers with old tools.

In music, automation is rapidly replacing studios, meaning more musicians produce professional-ish product. There are no limit on the amount of music people might enjoy. Yet, consumer spending limits shall prevent many from earning their whole income as musicians. It's arguably more "fair" that thousands of hobbyists become minor rock stars, but technically this eliminates musician jobs, even if the only jobs directly replaced were studio jobs. In fact, said studio jobs were merely simplified enough for ordinary musicians, not completely eliminated.

Actually, I said science jobs were mostly determined by the political will to fund them, TheLittlePrince, with scientists never even noticing that automation replaces anyone. I'd only expect science jobs to disappear if we drastically reduced the work week, instituted a basic income, etc. Yet, automation still plays a role in facilitating that.
posted by jeffburdges at 4:01 PM on September 25, 2013


fwiw, eric schmidt asked larry summers this and he gave a pretty good answer!
Imagine A Better World - Lawrence H. Summers - Zeitgeist Americas 2013

also btw Google Working to Solve Big Data Problem :P
Mr. Kurzweil said he is working on a way for computers to understand natural language rather than approximate the meaning of individual words by finding their synonyms. The problem that Mr. Kurzweil is employed at solving at Google will be applied to the 10 billion documents on the Web and the 10 billion pages of books on the Web, but the solution he develops could equally be applied to data repositories inside corporations, he said. This will help companies extract more relevant meaning and make more effective decisions based on Big Data repositories.

Mr. Kurzweil says Google can already reliably establish what individual pages, books and even video are about by extracting words contained in those documents and associating them with concepts in its database structure that indicate what they mean. But he says Google is now developing a data model that will allow its systems to understand the underlying meaning of language. “We haven’t done this yet — this is what I’m working on at Google — is to understand the meaning of documents… We don’t do as good a job on book pages as on Web pages because you don’t have links, for the most part, in books. So to make up for that, we have to understand what the heck the book is saying rather than just do these tricks with words… We really want to model the meaning of language.

“We have developed a model of the meaning of language so we can actually take a document and put it in this data structure that reflects its meaning, so even if you articulate things in different ways… it will have a consistent form in this meaning representation and actually model the meaning of all the 10 billion documents on the Web and 10 billion pages of books and then really enhance your ability to find the right information, summarize things to you, based on concepts,” he said.
oh and...
Monetizing Immortality: How Silicon Valley's Tech Titans Are Trying to Disrupt Death (I'd like you to eventually die! ;)
posted by kliuless at 5:19 PM on September 25, 2013


A lot of people here seem to be confusing "using the computer as a tool" with "being replaced by the computer".

No, I understand the difference, but the original post is about job displacement, and to a very large degree, things that were a lengthy, expensive process, like color separating, camera work, etc have all been replaced by computers since the DTP revolution in the early 90's. The computers aren't making the designs yet, but they've taken away about 80% of the process of commercial art creation. How many camera operators or scanner operators do you know? When was the last time you even saw a french curve or a piece of rubylith?
posted by Devils Rancher at 5:56 PM on September 25, 2013 [1 favorite]


« Older Baa baa black sheep / have you any wool / yes sir...   |   They’d look at me like I’m saying "I’d like to... Newer »


This thread has been archived and is closed to new comments