Will Robots Steal Your Job?
slate.com
slate.com
Small businesses will rescue the country you say? I'm not so sure. I own a reusable water bottle company. All of my warehousing and shipping needs are done by a fulfillment house (which does the same thing for other small businesses). Our shipping and warehousing needs are shared. And most of my marketing can be done through MailChimp,Twitter etc.
I don't need to hire any workers either.
Worse still, let's say my competition has 5-10 employees. All things being equal, I can undercut them ruthlessly if I wanted to because I have lower overhead.
I think an easy way to keep tabs on all of this is to simply look at unemployment. If it stays around 10% for another few years, it seems like we have a STRUCTURAL unemployment problem on our hands. And one of the causes is surely automation.
People have been saying that since the first mechanised loom machines appeared 200 years ago.
People may disagree about exactly when that will happen, but 20 years is probably a good inside guess computationally speaking, and anything much past 40 would mean that we really dropped the ball on AI software research (which is definitely possible, to be fair), so I'm not really so hot on the idea of writing off these concerns as overly premature or ill conceived. People were way off 200 years ago, but these days the literal end of human usefulness is visible on a very real horizon, possibly within our lifetimes.
Will robots buy the things that are made in the end?
Why will people build robot factories,cab drivers,teachers and doctors if there is no demand?
Automation is self-limiting in that the robots will only built if there are consumers who want to buy the end product.
Thats not a question a corporation is not equipped to care about.
Competition and shareholders ensure that a firm remembers its bottom line - to identify targets and to reduce costs - thats it.
Competition ensures that automation will stick around in the workforce - If 99% of all firms eschew automation, then the remaining 1% will beat the remainder. (caveat: Ideal perfect competitive situation used)
So no matter what - corporations will not care about that kind of thinking, that decision/processing is left up to the greater 'market'.
When the declaration of independence was signed, it took 19 out of every 20 people to feed America and provide and surplus for export. Today it takes 1 out of every 20 people to do the same thing. Did jobs disappear? Did we have unemployment at 10%? No. People adapted, and got jobs doing things that were useful and beneficial to the economy.
Robots will take revolutionize many industries, just like software is right now, but it won't mean the demise of the economy. In fact it will be a huge boon, and we will continue to advance as we have been for the past 500 years. By nearly any measure possible, we are better off today then 50 years ago. Technology is no small part of that.
It will get worse.
There is an aristocracy emerging of the intelligent/educated and the rest. It will get worse and worse, until you have 80 percent, then 85 percent, then 90 percent of the population economically useless. These predictions have been going on for decades (see "Player Piano" by Vonnegut). They are just now coming true.
Anyway there's data suggesting this transformation is already happening:
http://www.bing.com/search?q=median+wages+adjusted+for+infla...
I'm tentatively in favor of this change – I don't see how having efficient, intelligent, cheap means of production can hurt human civilization in the long run, so long as we can work out resource allocation challenges.
Not really, since this exact argument was previously made in the 1800's by textile workers. We can learn from that for sure. Humans are very complicated, but it is reasonable to say that we are fundamentally not much more complicated then we were in 1800.
Historical trends continue except when they don't. Housing prices are a great example.
I am tempted to say "personal servants for the rich" and I am only half joking. As you say, history repeats itself.
On a more personal note, I find this commonplace bigotry against the working class to be appalling and indicative of limited and closed minds. The idea that there is this huge labor pool of second class citizens, gammas or such-like, that is barely above the level of trained chimp, and their skills are so minor it's just a matter of time before the smarties of the world replace them with superior robotics, is disgusting. It is a fantasy. And worse it's a very derogatory and prejudicial fantasy.
I agree with you though that given the right incentives most people can learn new skills. I don't think they are inherently stupid at all, the mindless routine of their work just naturally limits them intellectually.
* Manna, by Marshall Brain, http://marshallbrain.com/manna1.htm
* Accelerando, by Charles Stross (goes way farther than Slate.com will ever go), http://en.wikipedia.org/wiki/Accelerando_%28novel%29 --- note that you can download the full text for free (licensed CC-BY-NC-ND).
Others have pointed out that foreign workers are cheaper than robots. Perhaps this will change when oil prices go through the roof, but for now biologicals are the preferred choice.
Manufacturing jobs are not fine, and they haven't been fine for a while. That's not just China -- the manufacturing that is done in this country (and Germany, and China) has become increasingly automated.
They're working on Agriculture, and as computer vision improves (as it's doing now) the agricultural jobs are going to fall through the floor. First, it will be robots that pick the easiest fruit, and humans finish the job. Then, the robots will get better and better, and eventually the humans will be too expensive to get the remnants.
The hidden piece to the law puzzle is that lawyers are becoming increasingly efficient.
Why do we need pharmacy techs anymore? Just design the pharmacy to auto-dispense the drugs, and have the pharmacist there to answer questions and give the directions.
As it is, technology is reducing the number of people who work in grocery stores. I'm sure that Walmart will figure out a way to automate stocking in the next ten years.
How close are we to machine learning and Radiology? Someone has to be working on a "See chart. See expert radiologist opinion. Give computer percentage and expert recommendation side by side. Put results afterward into computer to help train machine." workflow? How long, after that, is the machine learning better than the radiologist, who concentrates on a smaller and smaller number of hard cases?
How long after that, is a robot woodworker better than a human one? The robotic one can choose to leave human-like imperfections or not.
Now, which of these seem unlikely, given what we know about machine learning and computer vision algorithms? It's not that the problem is hard as much as the computing power/$ isn't there.
Yet.
Machines need markets to sustain their existence or else their owners abandon them or turn them into scrap parts.
The way I understand it, the automation (or technological improvement) of an industry lowers prices, and therefore helps another industry. Consumers to pay less for products, so extra funds directly (purchases) or indirectly (savings, investment) flow into new industries.
In the short term, workers displaced by automation are unemployed, but in the long term, they have growing employment opportunities elsewhere (or even in the same industry, later).
What's more, I feel like once automation takes hold in an industry, efficiency gains must accelerate relative to human laborers-- helping everyone. Sure, greater education and methodological improvements help humans to produce more for cheaper, but Moore's Law is a hell of an ally.
IANAE(conomist), but I couldn't explain why the world didn't disintegrate after the Industrial Revolution(s) any other way.
The remainder of humanity will subsist on some flavor of socialism. Most likely they will have a higher standard of living than the middle class today, but inequality will be dramatically higher.
Libertarian as I am, I wouldn't object to a welfare state which supports people at college student levels. It would be so cheap that I just wouldn't care.
Because, after all, the robot job holocaust isn't in the future; it's in the past. As someone else pointed out, in the past, even the relatively recent past, the vast bulk of humanity was directly involved in growing food. Now those jobs are gone. So what happened?
Other jobs, including types of jobs never imagined before.
Capital producers, such as programmers, those still involved in mining, etc. aren't going to want to just sit there and hoard their programs, iron, food, etc., and they will find some way to spend it on other goods and services that the robots can't do, or that humans simply won't accept robots doing. The future has a lot more artists in it, I think.
The unemployed will make money when you decide that there is something in your life that you want, be it art, services, or even just robot wrangling itself, and you want that good or service enough that somebody is willing to provide it for you for a price you'll take. Not you in the abstract, you. If you get desperate enough you may well be willing to pay them to learn how to do this thing you want, or at least subsidize their way through it. Things may change a lot, but the fundamental forces of economics don't break down until this set dwindles to zero (presumably because you can just instantiate an AI to do what you want), at which point all bets are off anyhow, and the poor will probably still be mind-blowingly rich by our standards at that point.
Now, that's not to say everything's hunky dory. There is some legitimate concern about the rate of job elimination exceeding the rate at which people can retrain into other skills. I'm pointing this out though to make the observation that we have over a hundred years of historical answer to your question, a continuous cycle of destruction of industry after industry, yet barring a fundamentally finance-and-stupid-policy crisis that we happen to be in now, we're still looking at single-digit unemployment with people working ~40 hours a week, numbers that are very stable across decades (at least in comparison to the horrible pictures people paint of this scenario). It's interesting to discuss what's different about this innovation surge, but "Who will buy the Model Ts?" ceased to be an interesting question decades ago.
(And personally, while the topic may be in the air, it hasn't happened yet. I don't think this unemployment spike has anything to do with robotics. The business cycle suffices to explain it. I don't expect the next wave of roboticization to really start impacting the job market for a good 5 or 10 years minimum. There's still a lot of pieces to put together.)
In the long term, things look really good. In the short term, it's looking like a lot of pain.
The complication is the resource crunch, particularly in rare metals and water.
I have to say that am so conflicted by this.
On one hand, I enjoy farming and run my own small farming operation. On the other hand, I love the technology. Lately, I have even been writing computer vision software in my spare time. Seeing my interests come together like this is really amazing, but also concerning towards my ability to continue farming in the future.
Maybe I will eventually find myself in the best of both worlds; developing software for agricultural equipment.
It's just that so much programming work is required that we don't notice it in terms of less programming jobs. Of course, there's no saying what will happen in 25,50,100 years...
One in my city - in 2011 it handled almost 1000000 cases already (it employs 102 "judging" persons - 4 judges and 98 "junior judges"). So it is 1000000/(270*102)=36 cases per day per person (not counting free days). Human input is very limited in such systems, and throughput is huge - freeing humans to do more creative work. I can only imagine it getting more automatic every year and handling greater varety of cases.
Article about it (in Polish - sorry) - http://wiadomosci.gazeta.pl/Wiadomosci/1,80273,10326080,Jedy...
I believe AI will be stealing our jobs long before robots will be stealing jobs of cleaners or builders. That's because we are much better at scaling software than hardware. Write program once and it can handle any amount of information given enough processor time. And processor time is commodity now.
And special purpose AI is getting better all the time.
Just what in the world is a robotic lawyer good for?! It's hard to imagine a profession in which a weird corporeal presence might actually be less useful. Yes, databases and expert systems are taking up a lot of the work that fresh law school graduates were doing a few decades ago. So the poor humans these days need to spend more time with badly paid and unexciting activities, like confronting hostile witnesses and persuading juries...
There won't be a C3PO walking awkwardly into court room, but some day most of the cases will be handled in the software, leaving controversial matters and appelations to humans.
In math: The Boyer-Moore theorem solver. This completely replaces a mathematicians job of proving theorems, right? Wrong. It replaces the menial task of connecting the steps, and making sure the proof is correct. Mathematicians can now focus on the bigger picture, more complicated proofs, instead of wasting time verifying the correctness of their proof.
In law: I'm not familiar with the exact system name, but there are systems that make retrieving huge legal databases to find relevant cases much much easier. This replaces the menial legal-aide jobs, but not the creative aspect of figuring out how to tie past cases together, present your case in court, etc. (I'm not super familiar with a lawyer's job, but you get the point.)
In art: Photoshop makes effects take much less work.
I could go on all day. Technology is what creates growth, and those who refuse to learn the new technology deserve to be left behind.
You might be right, but it doesn't matter what people deserve. Unfortunately, you may find that those that are left behind will cause untenable social unrest. If job automation outpaces re-training by too great a rate you can kiss civil society goodbye. No one can hold automation back, but we can work to smooth out the anticipated pain. No matter how optimistic you are about growth, we need this spaceship to hold together!
Perhaps this is a case of one gear (tech progress) going faster than the others (humanity).
Plus I would have to bring in another aspect - the fact that humans age, while machines/tech wont.
When you/people grow older, its quite possible that they wont be nimble enough to jump and twist fast enough onto new platforms.
In which case you/they are screwed. You can't assume that everyone is going to invest and save away and have perfect retirements/medical situations in their old age either.
In about a decade only fundamental services that cannot be automated will be left.
In the past, JOBS were a way to redistribute money and keep the money cycle flowing. The rich who owned companies are rich because the average consumer pays for their products and services. The rich then hire workers and pass money down to them, as well as spend money on other products made by other workers. But as unemployment reaches 20% then 30% then 40% over the next 50-100 years. We'll have to readjust the way we redistribute money in order to keep the money cycle going. Otherwise the rich will get richer and the poor will get poorer, to a point, then the whole system will collapse because who will have the money to buy your products. If the middle and lower class don't have any money to spend on the products that the company using robots makes... Product creators will effective kill off their own wealth machine.
Decades ago 80%+ of everything we bought was made in our own country, now that number has dwindled down to, 20% was it? It was some terrible small number. Outsourcing to China was good in the short term (cheaper more affordable products) and terribly destructive in the long term (disrupting the money cycle and destroying jobs). Now everyone in the US is pissed off about unemployment, we're passing all sorts of job bills and stimuliouses, while we're all ignoring the giant elephant in the room. We pissed our jobs away in the name of cheaper products (consumers are just as guilty as shareholders).
Look at what outsourcing to China has done. Rich get richer, poor got poorer, then the economy tanked because consumers don't have money to spend on the products the rich make. Now imagine what automation is going to do... Driverless buses and subways (already in testing / already in use) that don't go on strike, robots that fix robots (almost there) and never ask for a raise. Robot pharmacists that don't make mistakes (already in testing). Robotic construction workers that don't join unions. The people who own these businesses that employ robots over humans will make tremendous profits and save lives and do a better, faster job too. But they will HAVE to redistribute their money using a new method. Socialism? Who knows.
I'm not afraid of this future where robots take over. In fact I'm looking forward to it. I'm just afraid of our refusal to adjust to it.
The poor in China have gotten considerably richer. India also.
http://www.un.org/esa/desa/papers/2010/wp92_2010.pdf
We've also gotten richer in the US, even while "the Chinese/Indians stole our jobs". The only reason we believe the US has stagnated is because immigration lowers our cross-sectional averages (but not our longitudinal averages).
http://crazybear.posterous.com/did-immigrants-and-simpsons-p...
The bad news however is that most of the growth in family median income is not because wages are going up: it's because there are more two income families. I don't think it's bad that more women are working now and that they're making more money but I'm not sure it's really making us richer. What's missing from this study is any attempt to quantify the "free" labor that women were doing by being home for the kids and taking care of the house. If the extra money being brought in is simply going out to pay for childcare and a maid, we're really running in place, just faster.
In the consumer space, perhaps. The actual manufacturing done on US soil (I am assuming you are American), has continued to grow year after year. Much of that manufacturing has seen automation to the point that the labour force has declined, but that is not related to outsourcing to China.
What technology has allowed is more and more of us to pursue new opportunities. If we were all still toiling in the field, we wouldn't be building hot internet startups. There simply wouldn't be the time. As robots drive labour costs to near zero, brand new opportunities will appear for people to make an income and perhaps even get rich themselves.
There will always be short periods of hardship for some people, but in the grand scheme of things, finding an income should not be a problem. Finding a job you love, however, may become more and more difficult. Take agriculture as an example. I know countless people who would love to farm, but it simply is not an option for them. Thanks to technology, very few farmers are needed anymore. Likewise, I imagine there are also people who love working on the factory floor and building a business that employs robots doing who knows what is not at all appealing to them.
The question is, should we prop up jobs that can easily be replaced by robots in order to allow people who love that job to continue to profit by doing it or should we force them into doing work that is not of interest? Programming, for example, is a job anyone could do, but very few, relatively speaking, have interest in actually doing it.
- But not everyone can build internet startups, we can't all have startups. Money, also, is finite. The more someone else has, the less everyone else has. It's impossible for everyone to be rich. In order for someone to win, many have to loose. And making products to sell and creating jobs to make those products were a free market's way of passing the money down from those who have it, to those who don't. Now that those who have it, are starting to automate, more and more money stays in their hands and doesn't get passed down.
"Programming, for example, is a job anyone could do"
- I wouldn't say so. It takes a lot of skill to become a worthy programmer, which not everyone can do. If everyone knew just the basics, that wouldn't mean they could get jobs as programmers. Even if they did, their mediocre talent would mean that they too would eventually be replaced by robots and AI programmers.
Only if you define "rich" in relative terms. In absolute terms, virtually everyone in the US is rich, including most of those the government classifies as "in poverty".
What stops us from all having startups? Cost of labour is one of the biggest reasons. You either have to pour months to years of your time into a product just to see if it will succeed, or pay someone else to do the work for you.
If you could feed your plan into a robot on Friday and have the product in your hand on Monday, all built for pennies, the barrier is lowered to the point that anyone can do it.
More importantly, success need not be large in such a scenario. All you need is one customer to pay you a few dollars and you are ready to move on to solving the next person's problems. You don't have to bank on major or even moderate success.
It takes a lot of skill to become a worthy programmer
Skill is a function of time. It takes a lot of time to become a great programmer, no question. Those who love programming take that time to become great. Those who have no interest in programming do not. However, giving a compelling enough reason, they could do it.
In the US, I'd say health insurance. If you want a second reason, how about children.
But money are only tokens. Products and services created by economy is what makes economy strong, not tokens it uses (imagine all prices and salaries going up 10 times - what would change?).
Limited amount of money can suffice for everyody, if everybody is efficiently doing something useful to others, then limited amount of money circulates in economy, and everybody has it for some time, buing what he needs and selling what he produces. That's theory at least.
Problem is - sometimes nobody wants to buy what you produce at your price - then you are poor and rich keep their money for themselves.
That's really, really optimistic. The only one which seems plausible (and probably a great idea in terms of eliminating human error) in the reasonably near term is dispensing medicine. As in taking a clearly encoded prescription from a verified doctor and dispensing precisely the drug indicated. That still, though, won't eliminate the pharmacist, just a few of the technicians.
As for diagnosing diseases, they were working on expert systems to do that since the 70's. It has hope to be a great tool, but the doctor is still responsible for interpreting the statements of the patient, recognizing symptoms, and feeding the correct data into the system.
I don't see skilled knowledge workers being threatened for quite awhile. If I were going to spread FUD about automation, I would be pointing at RedBox and saying that it's paving the way toward automating low-end service jobs.
Before installing the robot, UCSF needed about half of its more than 100 on-staff pharmacists to administer and check the drugs going out to patients on the floor; now nearly all have been reassigned to different parts of the hospital, where they make IVs, help adjust patients' drug regimens, and perform other tasks that had been neglected when they were simply filling prescriptions
As mentioned there, and later in the article, the pharmacists (and I'm not entirely convinced that they were all pharmacists, as opposed to pharmacist technicians) were freed from their mundane, pill-counting tasks and put out on the floor of the hospital to monitor and "fine-tune" patients' drug regimens.
The article basically lays out how pharmacists had jobs which were largely filled with extremely repetitive tasks -- billing, prescription input, pill counting -- and less of what they're important for, which is counseling patients. A computer isn't going to be able to replace a human in patient interaction, medical demand is rising, and I don't see that he made a case for future pharmacist demand not meeting the projected 17% (if I recall correctly) growth. It just sounds like being a pharmacist sucks a lot less now.