April 09, 2016
Automation, Robots, Jobs
There are now 260,000 robots working in U.S. factories. My own reaction: Only 260k? I was expecting more. Robot sales are growing at over 10% per year. Amazon alone has 30,000 robots. I think that's a global number.
"Fintech" (financial technology) start-ups look set to provide levels of automation that will become much more disruptive to employment in banking and finance. Employment in banking has peaked due to automation and is in a long term declining trend. Citibank analysts believe "Banking's Uber Moment" has arrived and employment in banking will decline 30% by 2025.
Some fearful people want to stop the robots. That's not going to happen at a global level. If a country puts in place measures to slow automation then some industries will move to other countries. Some socialists welcome robots and want to combine them with a a guaranteed income for all. But as I've previously argued, capital is very mobile and so most robots won't have their labor taxed to support the poor.
Robots and other forms of automation face little opposition in factories and will get rolled out. However, shipping port automation is being slowed by union strike threats in Rotterdam and the US west coast lags far behind Rotterdam because of union opposition to automation. More on union opposition to port automation here.
Autonomous vehicles are another technology that will automate a lot of jobs. But just when autonomous vehicles will hit the market is far less clear. Most of the major car makers and at least 4 Silicon Valley companies are chasing AVs. But there isn't a single cross-over point where autonomous vehicles become better. It depends on the driver's, age, emotional stability (driving while sad or anxious raises accident risk by about a factor of 10), tiredness, and assorted irresponsible behaviors such as text messaging, talking on the phone, or driving under the influence of drugs or alcohol. This suggests that a vehicle could test a driver to decide whether it should take over from the driver.
The relative performance of humans or computers also depends on road conditions such as the weather, the condition of road paint, design of signs and street lights, and an assortment of other environmental factors. We could get autonomous vehicles sooner if transportation departments upgraded physical signals that AVs could use to know what's a road and when to turn or stop. This would likely reduce accidents by human drivers as well.
Long haul trucks could get automated sooner since highways are less complex than city and suburban streets and upgrading their signage and paint for automated vehicles is a much smaller job than doing it for surface streets. Also, not all highways would need to get upgraded for autonomous trucks to start hauling on some of them.
My guess is drones (small autonomous aircraft) are going to take off before autonomous vehicles for the same reason that auto-pilots came to aircraft decades ago: the air is far less crowded and less complex than roads and highways. But autonomous aircraft seem like a smaller economic deal than autonomous cars and trucks. If drones could take a big slice out of UPS and FedEx ground deliveries I'd change my mind. Can they? Or do most deliveries weigh too much?
Randall Parker, 2016 April 09 04:19 PM
Yes, it seems to me that UPS and FedEx are natural companies to experiment with computer-piloted planes.
Another big potential business opportunity I see that doesn't seem to be mentioned anywhere is computer-piloted VTOL passenger aircraft. In a 200-600 mile flight, a tremendous amount of time is spent at the airport. VTOL aircraft allow any large parking lot to become an "airport." Suppose a person wanted to go from somewhere near Boston to somewhere near Washington DC. Say, Newton, MA to Rockville, MD. Right now the trip would involve driving into Boston to the airport, going through security there, flying to DC, picking up checked luggage, and driving to Rockville, MD. With computer-driven VTOL aircraft, it might be possible to get in the VTOL aircraft somewhere outside of Boston with minimal security checks, and fly to the suburbs of Washington DC...maybe even to Rockville itself...and get luggage right out of the plane. I think that will take a real bite out of airline and airport revenue.
Oh, I see you're asking about drones delivering in the "final mile" for UPS and FedEx. I don't see that happening. People won't want drones landing in their yards (or perhaps even more, drones landing in their neighbors' yards). I think computer-driven ground delivery vehicles are much more likely.
When we were younger, we drove from central Ohio to southern New Hampshire and/or northwestern Illinois to visit relatives. The drives are too long for us now, and we are reduced to flying, which I truly hate. Once autonomous vehicles become affordable, we will go back to driving.
Do you think many people will make that choice, to avoid the airline/TSA hassle? What will happen to short hop airlines?
There is an important nuance about the connection between the job market and the robots. Not all robots are destined to steal jobs from humans: the class of robots that displace human workers are only those machines that do what humans do, but there are and there will be many more types of machines that do precisely the kind of work that humans don't do, and the latter types of robots actually would actually create new jobs for humans instead of getting them fired. In fact, robots are not just humanoid taxi drivers and janitors that we see in science fiction movies: gigantic excavation machines and gargantuan automatic bulldozers that would mine the earth a hundred miles below surface, are also robots. Many new automated factories will be manufacturing things that humans can never do, and the latter kind of robots would actually generate new jobs that never existed before instead of eliminating jobs.
However, robots do not get paid in dollars, euros, or bitcoins: they only accept kilowatt-hours or BTUs as payment. This means that in the future, the grid capacity will have to increase dramatically, and I am not even talking about electric cars. Even if all cars and trucks were electric, this would require only an expansion of 10-20 % in the grid capacity, which can easily be achieved within 20 years, just by adding 1 % per year to the grid. The robotic future manufacturing and the mining machines, the giant robots that will build cities within weeks, will require nuclear energy because solar or wind energy is not concentrated enough for such energy-hungry work. The power grid will be expanded 10-fold for those mining machines that go a hundred miles underground, and the futuristic skynet-like manufacturing that will build not only everything in large quantities, but also gigantic things like new subway networks under entire cities all over the place, new giant cities in undeveloped parts of the mountains that are empty.
This is why the key is to develop the neglected thorium molten salt reactors that Nixon eliminated because in the short run the old uranium reactors created more jobs at the time. The fact is that there are working prototypes of thorium molten salt reactors that burn all long term waste, making these nearly 100 times more efficient than uranium reactors, while not causing waste to accumulate. The cost of commercializing these would be much less than a Manhattan project, since the technical barriers are well understood. Even Germans are making progress in fusion research. Nuclear R & D is requires long term investment, which private companies are reticent about, since it would not generate short term profits. This is one areas government is helpful.
Don't get distracted by the idea of robots that look something like humans. That's relatively unimportant, in the grand scheme of things.
The important thing: the vast array of simpler techniques for simplifying and automating human labor that have been in use for hundreds of years, and the wide array of simpler forms of automation and computerization.
One simple example: it's far more effective to find a way to use a simple electric motor in a box to wash clothes (aka a "washing machine"), than to develop a robot with arms that will duplicate the human method of beating clothes on rocks in the river. Another example: assembly lines are technically infinitely simpler than robots, but have had far, far more impact.
Interesting comment on the unions (ILA) slowing automation. Please note they initially did the same when containerization started replacing bulk shipments of cargo on ships. Eventually the ILA accepted the writing on the wall and then international trade exploded because of the reduced unit cost in transportation due to containerization and then the increase in ship size to carry even more containers. This is to say that change happens organically (a transition period) and the so called disruption occurs gradually to reduce labor hours. All new innovations are disruptive when viewed from hindsight. The self propelled tractor was disruptive to agriculture when viewed in this light and now we produce more food than ever with a minimum number of people. All of those people found employment elsewhere as the transition occurred. History suggests that the economy will further diversify as new innovations occur that free up labor to do other things.
The ultimate question should always be what new thing will occupy our time that diversifies the economy? A new growth industry, a new frontier? SpaceX, Bigelow Aerospace...
The hindrance to all economic activity is the parasitical influence of those in government who support themselves through taxes and regulation at the expense of society instead of facilitating human endeavor. Parasitical in that when they don't produce a product or service of value to the customer.