This is a big topic, so this will be a long posting; but it still can not cover all the points. Or even most of them. Robotic labor spans from technology to organizational behaviour and management to economics and politics. Of necessity, a lot of that turf will be left unexamined or with a bare mention.
The Politics of it, for example. There will be a lot of heat and shouting about it, especially from the Socialists and Progressive Left. They will try to use it to ram down our collective throats things like “basic income” laws and more collective ownership of the means of production. Volumes will be written covering that as it unfolds and most any specific predictions now will be wrong.
With that intro; on to the topic. Will robotic labor cause a catastrophic collapse of our economy, and especially of capitalism, due to a massive wave of displaced labor that will be unable to compete? My assertion is “No”, and the reasoning will follow.
Will the transition be disruptive? Certainly.
There are huge numbers of folks employed as drivers or operators of vehicles, for example. Trucks, buses, trains, taxis, tractors, etc. etc. Once there are approved and proven self driving cars and trucks, what purpose does the driver serve? Huge numbers will become unemployed and will need to move to some other source of work. I think it will not be as bad as folks are predicting. Simply because for trucks and buses, the driver is not just there to drive. They monitor the passengers, basically keeping order. They assure the load is safe and that it is loaded and unloaded as expected. IF the UPS truck self drives to your house, how does it ask for your signature (or even find the doorbell to ring)? How does it keep local ‘bad guys’ from just getting on board and stealing the load? So many drivers will change to “shipment monitors” and equipment tending.
We have had similar economic transitions in the past, and survived them. The long term results were more prosperity for all. Less unpleasant labor and more leisure time. There is nothing in this transition that is fundamentally different.
Economics is descriptive. It looks at “what is” and attempts to describe it. Often there are several competing descriptions since it isn’t a science and there isn’t always a way to test given theories. Sometimes it attempts to be predictive, often poorly.
Economics looks at the basic question: Who makes What for Whom?
Political Economy adds: Who Decides and a little bit of Why? (Mostly as justification, IMHO).
I note in passing that Engineering covers the How and From What; while Management / Operations Research covers When, How Much, at What Cost; and Marketing covers At What Price.
There is a sub-field of Labor Economics. Most of it is devoted to things like equilibrium wage rates, though some explores the political economy aspects of unions and impacts on labor supply.
When looking at the economic impacts of any change in the “How” (from engineering advances) it is important to remember just what parts of the turf is an economic segment. Why it happens is not in that turf, though political economy can stick an oar in at that point. Just realize that Political Economy is more politics than analysis.
So where is labor used?
1) In Government
2) In the home
3) Private companies
4) Public companies
5) Non-Governmental Organizations
6) Churches and religion
7) Individual activities
8) Military & Armies, public and private
Robotic labor will mostly impact private and public companies in the shipping and trucking industries in the next wave. Manufacturing and assembly has already had major impacts so there isn’t really that much left to happen in them.
The impacts in Government ought to be minor. They simply do far too much that is “desk work” and not simple rote mechanical tasks.
The impacts in the home will also be very minor. Even the Roomba robot vacuum cleaner has had limited impacts. Any impact will be highly beneficial in any case.
NGOs are similar to Government, but act to push it from the outside. Again, little scope for robotic labor to change lobbying, research, recruiting, organizing, pushing papers, etc. etc.
Churches and Religion are essentially an all-people-all-the-time operation. Impact from robots ought to be nearly zero.
Individual activities also will have minor openings for robots, but again entirely beneficial. For example, a home bound person may now be able to go the to movies on their own instead of waiting for a driver. Folks out at a party can all participate and not need a ‘designated driver’ as the car drives itself. People driving long runs cross country can make longer runs, even napping in the car (eventually) while making the trip faster and safer. I, for one, will be quite happy to sit on the porch sipping a mint julep while my robot mower handles the lawn (and I watch out for the cats and kids…)
For the Military uses, we’ve already got drones and remotely operated vehicles. Taking people out of the line of fire is entirely a good thing. Eventually robot tanks and ships will reduce staff risk even more. Again, I see little downside to having fewer dead and wounded in military operations. Having folks driving trucks past land mines and getting legs blown off is NOT a contributor to our economic productivity… and needing fewer artificial limbs is, IMHO, entirely a beneficial outcome.
So there are whole swathes of our economy that are outside of corporate production and where robots are either entirely a benefit, or will have little application.
So that leaves us with public and private companies that provide goods and services. In these sectors, there will be significant changes and impacts. Many for the better, a few causing disruption in labor allocation.
To the extent it lowers costs and increases productivity, we have more “stuff” for our enjoyment. (This dodges the question of “for whom” the stuff is made and who can buy it. I’ll explore the ‘for Whom’ more below.) To the extent it doesn’t make more stuff or provide more services or were it to decrease productivity or raise costs, it will NOT be done. So where is the ‘lose’ scenario in that? Only in the displaced workers who need to get new jobs; and that question of who collects the gain.
Basically, robots move labor between tasks and sectors, it does not end the availability of labor.
The Issues and Questions
There are 2 basic questions:
1) How much “stuff” is too much? More automation means more “stuff” is made (and more services provided) with less labor. So what house is too big? How many people have houses that large? How many cars is too many to own? How many folks do NOT want a newer or better car? Or newer and better home appliances? More clothes? Better food? More dinners out and more plays?
Looking at the truly rich, those numbers must be way bigger than the present average person has or uses.
2) Who gets the money from the robot labor to buy that stuff? This is the old “capital vs labor” argument of the Socialists and Communists from the 1800s when factories and manufacturing with machines first replaced hand labor. We have about 200 years of existence proof that yes, we get a large rich capitalist class, but no, we don’t end up with a labor revolt from deep impoverishment.
There is still a case that can be made that due to the tendency for the rich to get richer (since they do not and can not spend all they make but must invest it), that they will get an even larger share of the “robot ownership” and so gain most of the profit from their use. I think it is a poor case, but if it were to ever be shown to be a problem, it would be pretty easy to “redistribute the wealth”. So I see no reason to take action to fix a problem that isn’t manifest yet. But it is a theoretical risk, and has been for 200+ years…
The more reasonable risk, sort of a 2B item, is that the division of the proceeds to labor will have more going to skilled labor and fewer jobs (so less of the share of labor earnings) going to folks who are unskilled. So is there some fundamental limit of skill level that will be unemployable? Well, yes. We have that now. The insane. The infirm. The idiots. The criminals. Children. So I don’t see this as a change of kind, though perhaps a change of degree. Yet we’ve already lost most of the zero skill jobs to mechanization, even if from dumb machines. In the question of skilled vs un-skilled labor, it will hinge on just which skills. Driving will be less valuable a skill. Singing or massage ought to hold their own. Robot Repair will be highly valuable.
Which leads to the point that there will be new jobs needing new skills created. Robot sales. Robot trainer. Robot manager (robot wrangler? at least in films ;-) and of course Robot Repair. Will robots someday be able to repair robots? Maybe someday, but not right now. I doubt I will see it in my lifetime. It is a complicated set of tasks, requiring fine vision and fine motor skills, along with some creativity. All things robots do very poorly and where little on the horizon says a breakthrough is imminent.
How many of those new jobs will there be? Will that be larger than those displaced? Can the displaced be absorbed into those jobs, or other jobs freeing the folks presently in them to take the new jobs? How much re-training work will be needed and how fast? Those are unknowns, but generally the consumption of goods goes up with mechanization until the labor demand is again limiting.
There are some other issues that will come up, but they are more of the form of changes of markets than they are issues of labor supply and disruption.
3) Global Scale manufacturing conglomerates vs Just In Time local & customized manufacturing will have a shift of balance. In many cases, the giant factory + shipping will end up costing more than the local flex manufacturing robot, so we might well see some industries returning to local makers China has built a sock factory sized to meet global demand. What happens when my robotic knitting machine can make all my clothes on demand? Similarly 3D Printing is a kind of robotic manufacturing and it, too, can work on small and local scales. Just how big will the robots be, and can they result in the demise of the Giant Robot Factory?
4) Is there an upper limit to things like Tourism and Baseball Games? Dining out? Shows and live plays? As there are more folks with free time to attend such things and consume their services, their will be a need for more actors, ball players, hotdog vendors, bartenders, tour guides and chefs. To some extent we’ve seen this process already happen with the rise of factory manufactures.
Smith’s Corollary to Amdahl’s Law
Amdahl’s Law is a rule from computing. You can calculate things one step at a time (scalar computing) or in batches of parallel similar tasks (vector processing). In the ’70s there was a huge debate over which was better. Many asserting that the use of parallel computing would eliminate the need for one big compute engine.
Amdahl postulated his law that, in layman’s terms, basically says that for any given problem, once you vectorize what can be vectorized, the residual scalar portion dominates the remaining work. Basically, you continue to limit on the non-reducible work that takes “one big engine”. Say, as an example, you had a work site and need to move 400 small boxes and one bulldozer to a new work site. It takes two big trucks to move it all. Now you get 400 robot carts. They can move all the 400 boxes all at once. You still need one big truck, and that can’t be reduced.
It is my opinion that a similar process applies to work and labor. There are many jobs that only people can do; or where the customer will only accept a person. I buy NOTHING from robo-calls, for example. You can automate it, but it doesn’t work… I just hang up. Similarly, I don’t want any robot barbers putting sheers near my ears.
So my corollary is just this: Once you have mechanized (automated, robotized) those parts of labor that are amenable to it, the remainder can only be done by people and that dominates the workload.
So let’s say you make plastic molded hoses. If you automate the production portion 100%, you still need clerks, marketing, sales, management, HR, engineering, facilities, legal, janitorial, etc. etc.
This implies some very important things:
1) Until Robots are better than humans at EVERYTHING, there will be a demand for labor.
2) Until everyone on the planet has everything they need and want, there will be unmet demand for more production.
Which together mean that labor will move to the residual non-robotic segments, and demand for goods will grow to consume all available labor (until some far distant time when everyone has everything they want).
We have already seen this happen, by the way. The first rounds of mechanization did exactly that. Moving people from farming and hand craft assembly into sales and factory work.
As noted before, there is a potential issue of the distribution of the fruits of the mechanized production. We’ve already gone through that kind of stress a few times and so far we’ve not had the economy falter, so I’m of the opinion the long awaited collapse (waited for by Communists globally for generations) just isn’t going to happen. It is a belief that has persisted for well over 100 years, despite no evidence, but folks sill make special pleadings for it.
IF it ever arises, it would be simple enough to take taxes from manufacturing industries and apply them to welfare payments, rather like is done for the incompetent today.
Some Historical Perspective
Realize that in very real ways we have already made the transition to automated manufacturing and farming. This happened during the Industrial Revolution. Thousands of manual laborers were moved off of farms and into factories. Later, automation in factories moved many of them again, into computer work and clerical, sales and service. Craft workers were put out of work wholesale by factories.
Take my family history. Named Smith, we were blacksmiths. The industrial revolution eliminated the need for blacksmiths. Granddad moved to the farm and made a good living at it. Then farm mechanization resulted in farm consolidation and much less farm labor needed. Granddad lived out his life on the farm, but his kids when on to be electricians (jobs that did not exist before), real-estate salesmen, restaurateurs and more.
In the ’70s, Cezar Chavez “organized farm labor” with a focus on grape pickers. Now their are mechanical grape pickers, largely a result of his efforts raising the price of labor and, more importantly, making it unreliable at harvest time. A strike at harvest time is simply not an acceptable risk. That was when the last major tranche of farm labor left the fields. While there ARE still farmers, and some farm hand laborers, the numbers are much much less than before. Just like you can find the odd blacksmith still working.
So in large part, the Industrial Revolution has already caused well over 90% of the labor displacement, and it is already behind us. We did have “issues”, but we handled them. The Grapes Of Wrath was about the flood of farm labor leaving the land, driven off by the giant dust storms of the ’30s and by mechanization elsewhere. Their decedents now live in L.A. and work in jobs like making movies, fixing cars, building houses, etc. We’ve already built systems to assure that kind of displacement is not left to decay, as it was then. A new Grapes Of Wrath is just not going to happen in our present welfare state.
Here’s a partial list of surnames and trades that have simply gone entirely out of existence already. In some cases, like sailors, it is a partial. Yet even that partial has eliminated most of the jobs. Giant container cargo ships can carry huge tonnage with a smaller crew than the old sailing ships. We no longer need dozens of ‘riggers’ to handle the sails. On shore, the gangs of longshoreman have been reduced to a few dozen operators of giant cranes. Yet the world did not end…
Gone (or dramatically reduced from pre-industrial levels):
Chandler Smith (several kinds, blacksmith, tinsmith, goldsmith, etc. etc.) Tinker Cooper Weaver Farmer (partial, but high percentage) Taylor Cobler Sailor Shipwright Longshoreman Phone Operators (remember when they were used for all calls? All long distance?) Miners (largely mechanized, few using shovel and pick anymore) Elevator Operators Dishwasher ( I did this once, washing dishes by hand in a restaurant. Glad it's now a machine.) Doorman (other than the ones used for decoration, what with self opening doors)
No doubt dozens more I’ve forgotten. At one time, every town had a local brewer. Then factory beer took over. Now there is a small resurgence of the local “brewpub”, so “brewer” is making a comeback. But more as a choice for preference or show of style than necessity. In my opinion, that, too, matters. As a society becomes more affluent, it spends more on “hand crafted” just for the flamboyance of it. We buy $5 hand crafted beer and $4 hand made coffee “just because”. The world is not always about efficiency and low cost to manufacture. As robotic labor raises general affluence, more folks will be employed providing just that kind of specialty service.
Then there’s all the work that used to be done in the home, now done largely by machines. Clothes washing, carpet beating, even cooking now done in many cases by shoving a factory built meal into the microwave. Not to mention what used to be home-work and is now done by machines. In my closet are a couple of cherished shirts, made by hand by my mother, who had been a seamstress in World War II England making uniforms. Now we have laser cutters that cut whole stacks of fabric in one pass and industrial sewing machines moving fast (though still staffed by people, but in low cost places like China) All that labor displaced from our economy.
In essence, we have already moved through the labor transition for most trades and crafts, for manufactures, and for many services where advanced tools automate the labor portion. Plays no longer happen in each town with a cast and crew, instead we have recorded movies and TV shows. The Big Band era with a band in every town has also moved on. We no longer go to the local restaurant to be entertained by a band of a dozen players. Who needs that when we have an iPod 24 x 7? Musicians are now employed in movie studios, and major venues, and are fewer in number.
The work left to do is now largely dominated by those parts a machine can not do. In many respects, a factory making a product is already a large robot; just one that needs a few people in key places to tend it and where the “programming” is built into the design of the space and machines in it. The real risk of robots in my opinion is to those large single person automations. Why would I buy a factory made cup if my robot can custom make one at home? Which leads us back to the question of who will own the robots, and gain their work product.
What jobs are at risk?
What are the job categories most at risk from the current crop of robotics? I would guess it is those jobs requiring modest levels of manual skill, vision and hearing, and only limited judgments. Jobs with little human interaction (though a little can be done – think of those robotic voice tree “service” lines… “please say your account number”. Where the need to understand speech is very limited and confined to specific question / response patterns).
To me, that says that you might be able to put a robotic receptionist at the directions counter in a mall, but not in your Doctor’s Office; or at any company where image mattered or, the complexity of queries would be large. That is, receptionists are not a good candidate for replacement, outside of a small number of places. But robotic receptionists for phone inquiries are already here.
Much more at risk are tasks like driving cars and trucks. A very limited set of mechanical actions, they are already used to control machines, and the need to see and hear has specific things to look for; like lines on pavement and traffic signs. Not things like a police officer deciding someone looks shady.
With those as examples of my decision process, here’s the jobs I see as most at risk:
Drivers. There’s a strong, active, and progressing effort to make robotic cars. In some ways, this is one of the easier robots to make. The vehicle is already a mechanical device and designed for limited input actions to control it. The major need is to avoid running into other things and that can generally be well accomplished. The problem space is generally well marked with informative signage and lines. GPS provides detailed position information. There is very limited interaction with people (mostly setting destination) and not much in the way of complex manipulation or problem evaluation. We already have working prototypes.
That means that such jobs as Taxi Driver, delivery truck driver, bus driver, and long haul trucker are “at risk”. Yet even there, I think there are reasons many such jobs will remain. Security concerns will likely keep some kind of monitoring presence in some trucks and taxis. Assuring passengers have paid fares will be important for buses, while package delivery has some complicated acceptance and signature issues to resolve along with how to get packages the last 100 feet. The idea that a drone is going to drop off my new stereo after coping with trees and rain while keeping the local thief out of the truck seems a bit far fetched.
For long haul truckers, driving is only part of the job. They also secure the loads, present paperwork at truck stops, assure maintenance is done and monitor for needed repairs. At loading and unloading locations, they assure the right product is handled and placed appropriately in the truck, handle the paperwork, and sometimes even do the loading and unloading. While some of that could be handled by the equivalent of a local pilot, it would still present problems. If all you need to do to rob a robotic truck is put a car in each lane in front of it then slow down, I would expect hijackings to increase. For that reason, I expect the first few years of long haul truck automation to not be replacing the driver so much as allowing them to sleep while the truck crosses Nevada on the interstate.
For trucking and taxis, I would expect conversion to take many years. Plenty of time for old drivers to finish their careers and young folks to pick different ones.
I would expect pilots to be replaced faster. The US Air Force has just had a fully automatic space shuttle land at Kennedy Space Center. Modern passenger jets are essentially robots already. It would be very easy to simple have the airplane fly itself, have the pilot be the “emergency” pilot (what the co-pilot does now), and eliminate the 1/2 of pilots who are presently co-pilots.
Industrial trucks, like coal haulers, have already lost most of their drivers. This was done by making each truck the size of a house. Since they run on a limited course (mine to rail head to mine) and the traffic is strictly regimented, they were some of the first to demonstrate a successful robotic driver. In any case, we are not talking massive job counts here. A very large mine may have on the order of a dozen trucks running at once, and there are not that many large mines. For things like road building and repair, the context is complex, the road markers are, by definition, often gone, and the task is highly variable. They will be slow to automate.
I would expect trains to become robotic first. They run only on a controlled track with controlled crossings. They are already in communications with a central control. The task consists substantially of controlling train speed so as not to exceed what the track can bear and the train can stay upright. Add some collision sensors, some exterior cameras, and some computer control on board and you are pretty much set. But again, the total numbers employed as train drivers is rather small.
Factory assembly work, especially for things designed as a stack assembly, is a prime candidate for automation. So much so that it is already highly automated. Oh, and most of it has gone to China already anyway. It isn’t robots that are a risk to USA manufacturing jobs, it is the smart, fast, and cheap Chinese worker. To the extent we can automate further with robots, factories will return to the USA and save transportation costs.
Also, in many factories, there is a box stacking function at the end of the assembly line. When I worked in the peach cannery, we called it “tail off”. Supposedly since you were at the tail of the line and took the boxes off and stacked them; we told each other it was because you “worked your tail off”… That kind of “pick and place” job is ideal for robots. I’ve also seen film of industrial robots doing that job in at least one factory, so it is already underway.
I’m sure there are other jobs suited to robots, and that I’ve missed. But frankly, I’m not seeing the big problem. There will still be millions of jobs that can not be done by a robot, and those robots will need training and services.
What is not well suited to robots?
We’ve already touched on this to some extent. Let me just start by pointing out that “repair” of most anything does not lend itself well to automation. There are many modes of failure. It is often hard to figure out what is broken, even for people. Every task is unique and often has unexpected problems to solve. (Like removing stuck fasteners or getting broken bits out of a housing.) It will be generations before a robot can easily repair a broken robot. Or a robotic car, plane, or train.
Things requiring creative thinking, delicate manual dexterity, precise vision, human interaction, and work in changing and novel environments are all very poorly suited to robots. A partial list, based on my expectations about what works well with people and poorly with robots follows:
Management Human Relations Sales Marketing Engineering Arborist / tree trimmer Artists Plumber Electrician Painter Piece welding (factory assembly welding is already largely automated) Billing Clerk Filing Clerk (though computerized documents might reduce the need) Complaint Desk / Customer Relations Receptionist Mason / Cement work Lineman Optometrist Doctor Nurse Dentist Dental Hygenist Radiologist Pharmacist Appliance installation Appliance repair Car repair Building maintenance Home repair Robot repair Factory repair Fence buidling and repair Ambulance personnel Fireman Police Photographer Video Camera operator Actor Presenter Author D.J. / Party Planner / Wedding Planner / Event Planner Project Planner Lawyer Politician Chef Server / Waiter / Waitress Teacher ( though automation exists, it still has not replaced the human contact ) Barber / Hair Dresser Butler Maid Jewler Forester / Park Ranger Bike repair / sales / shop Pro Sports Golf Pro / Tennis Pro / Sports Pro and Coaches Sensei and other trainers Real Estate Appraiser Loan Officer Gardener Custom Work - like tailors or counter / kitchen remodels Programmer Systems Administrator
Again, I’m sure there are many many more I’ve missed or left out. Given that much of manufacturing has already moved to China, and what has stayed is often of a custom or specialty sort, the biggest risk of robotic manufacture is to China. not us. The only sector where I see huge disruption possible, is in transports. Even that ought to take 20 years as vehicles age out and are replaced and the technology matures.
Most of the jobs still in the USA are already those where there is little that can be done to automate it.
Is there really a large bunch of folks who want to vacuum the house, mow the lawn, or drive their daily commute? When we have robot lawn mowers, we will still need a human to supervise and make sure the cat is kept safe and kids kept away from the mower. Personally, I’d rather be in the chair on the porch with my mint julep being robot manager than pushing the mower myself.
The real issue from the use of robots is just “Who owns the robotic capital stock?”. That is substantially the same issue raised by factories and automation of manufacturing. It is possible a larger share of GDP Gross Domestic Product will flow to the owners of capital when they have displaced labor with robots. That the concentration of the benefits of robot labor will flow uphill to the few who are the owners of our capital stock.
That leads to the question: Is there a limit to the amount that rich people can spend? To purchase and consume things that do need labor. Yachts, parties, custom cars, butlers, etc.
Similarly we must ask: Is there a cohort of people who can not provide labor for a wage, due to incompetence or infirmity, and who do not own capital stock? Lacking marketable skills. Will that segment grow too big?
In my opinion, crossover to an excess of labor is still far far off. Robots can not understand. They do not see very well, and hove no understanding of what they see. They can do simple things like not run over an object in front of them, or pick up a box; but they can’t tell the difference between a sleeping cat and a towel on the floor. They can barely mimic “social skills” while actually having none at all. They can not handle novel tasks and get flummoxed by complex environments. We can barely get them to deliver a lunch tray to the right hospital room, and then they require a strip on the floor to follow and can’t handle a question like “Where is the salt shaker?” when they do deliver the meal.
Will the advent of autonomous robots be disruptive in some fields? Certainly. I would not want to be a cab driver in the next ten years, for example. But that disruption will be far smaller than the disruptions of the Industrial Revolution and the Mechanization of Agriculture. Or even the impact of computerization on clerical and record keeping tasks in large companies.
The only potential problem would be if the advent of even more capital intensive production resulted in too much concentration of money in the hands of the capital owning class and too little in the hands of working folks. IFF that ever came to be, it would be fairly simple to tax away some of it and use that to pay for training or welfare for the unemployed. Though, frankly, I suspect the rich will find lots of interesting ways to spend their money first. Similarly, the use of stock grants and options for professional workers will mean much of the robotic labor will have the advantage flow to other corporate labor.
So that’s why I’m not all that worried about applying automation in a few areas of the economy where they will help. Since we have an existence proof that it isn’t all that bad and we’ve been through bigger and faster before.
Maybe I’m just not all that sympathetic to the folks Driving for a living as I think it is OK if the Drivers join the Coopers, Weavers, and Smiths doing other work in the future.
I’m certain that most of the folks I know can easily be happy with twice as big a house, another newer car, an added pool in the back, and a boat for weekends. Not one of them is anywhere near maxed out on stuff they would like to have. That means there is not yet a cap on personal demand for goods and services. In that context, we are able to absorb a larger body of work being done to provide those things.
When you put robots into “Who makes What for Whom?” you get “Robots make Goods for Whom?” And that size of future goods desires is very much larger than at present. Which only leaves the question “For Whom?” The few rich who own them? Or perhaps pretty much all of us via a variety of smoothing systems? (From taxes to natural feedback loops). Concentration of wealth is a function of all capitalism and most socialisms. Even communism concentrates the national wealth, though mostly into the hands of The Party. So that problem is a constant, robot workers or no.