*Robots will be able to do the jobs for $15 hour sooner than they'll be able to for $7 an hour. The math will accelerate the move to automation.
At this point, the jobs that are still done by humans are jobs that require judgment and intuition. Those things are very, very difficult to emulate in software, unlike repetitive motions and data lookups.
Physical work -> thinking work -> ???
Also, you forgot that many of the Luddites actually starved to death in the streets.
That was replaced by cars. And while industrial machines automated a lot of what used to be manual blacksmith/tradeskills work, cars were made from thousands of parts, which required a massive new supply chain.
Software automation is completely different, it scales and replicates in ways that don't require new replacement labor. WhatsApp serviced 1 billion users with 33 employees. That's scale.
So when self driving trucks kill 8 million trucking supply chain jobs and self driving cars do reduce the need for car ownership, for parking attendants, valets, meter maids, traffic cops, uber drivers, etc, low skilled people won't suddenly find those millions of jobs replaced with similar jobs with similar skill levels in another vertical. They'll just be....GONE.
And keep in mind, the "gig" economy largely replaced good, reliable, jobs with temporary, inconsistent, jobs. The industrial era, with it's big centralized unions, who regularized pay, the work week, benefits, pensions, et al, was not re-invented for the new service and gig industry.
The idea that technological advancement doesn't create structural unemployment is largely an article of faith. What it means to have a middle class, blue collar, skilled position with predictable work/life balance, income, and benefits has largely evaporated for a good chunk of the country. What comes next after automated logistics, automated factory assembly, automated deliver and self-driving vehicles? Hell, Google Duplex even shows you a lot of the frontdesk services can be reduced along with concierge services to only those instances that go off the rails.
We better start thinking about these problems and stop with dogmatic faith that something big will come along to employ all the displaced people. I'm talking about either universal basic income, government make-work, or something no one's thought of yet.
The next step, I think, is cheap universal energy, solar/wind + storage creating an energy supply that is cheaper and doesn't cause geopolitical resource conflicts. Near-free energy and near-free information is an extraordinary combination that makes amazing things possible.
Amazon used humans because the investment to build full automation was too expensive and take too long to build out compared to spinning up human labor. However now the logistics of the fulfillment center are well understood and they can be rebuilt with full automation.
It was clearly a matter of time and money, not “these jobs are impractical hard to do with automation”
In my area, Amazon has its own trucks delivering packages, for the most part. That's not exactly something that can be automated.
I'm guessing no more than 20 years before all of this is automated. And then what for low skilled labor?
I take it you haven't seen a single item dropped on the floor result in several Kivas playing bumper cars with each other, necessitating human beings to physically drag them away.
Now, let's assume a 0.1% (probably very high) item drop. So about 40 packages will be dropped a day. How many people do we need to run out and get 40 dropped packages? Certainly not more than 40. So at this failure rate, I can reduce the number of employees by a factor of 3600/40 = 90x!
Clearly, handling failure cases is a much lower workload and needs far fewer people in the pipeline. Like I said, Jingdong has a fulfillment center that does 250,000 packages a day they claim, and they say only 5 people are needed to babysit its operation.
Jingdong is doing what Musk talked about: A 3D space filling factory. They've build a Voxel-like 3D positioning system of shelves. The entire warehouse space seems to be used, stacking stuff up to the ceiling. Robot sleds seem to be able to address any X/Y/Z coordinate in this space, moving down to where a robot arm unloads a palette.
So basically, it's Kiva bots, but on 3D degrees of freedom.
They do have Kiva-style bots for the final part that loads the wrapped packages onto the Jingdong delivery trucks.
I don't think Amazon can easily automate its existing warehouses beyond a certain point (I don't know what that is, they probably haven't reached it yet) because so much effort has been put into optimizing them for a human workforce. Obviously, Amazon is going to automate away as many jobs as it can, but it's not clear that full or close to full automation would necessarily be cost effective.
Human labor is both cheap and dependable where automation takes capital and introduces risks. Just look at Tesla who cut back on automation at their factory.
The trend is for companies to do things by hand as they scale, and only automate when it's either required to stay competitive or the business enters a near steady state.
If something can be done by machine, it is done by
machine, and that's been true since the beginning
of the industrial era.
Automating the seasonal jobs would have a payback time 12x longer than automating the non-seasonal jobs. The December robots lock up capital during the 11 months of the year they're idle.
This doesn't make sense. Increasing workers wages doesn't change the rate a which automation technology becomes available. The bottle neck is the availablity of capable automation systems which is unrelated to workers wages.
Even if there was a technology available that cost $15 an hour and worker wages were $7, increasing the workers wages doesn't decrease cost of the technology.
Sure it does. If your current costs increase significantly, the math on funding research to automate the process changes. Not simply because you may fund research to fix your increase in cost, but also because there is now a more viable market for others to do the same.
Lets say there is an industrial machine that TCO is $50,000 per year to operate.
Lets say a human doing the same task, same quantity, and same outcome, for low pay only costs $20,000.
It would be crazy not to use humans. Trying to drop the cost of the machine is likely going to be very difficult because there will be no demand.
Now, give the humans a raise and insurance. Lets say that brings up the total human cost to $48,000. We'll, it is actually worth getting the machines now. Seems counter intuitive even when the machine is $2000 more. But it is almost granted the cost of the machines will drop now there is a mass market for them.
And again, that is where you are wrong.
Economies of scale. Building one machine is expensive. Building 10 is slightly cheaper. Building a thousand lowers the price. A million makes them a commodity.
If the politicians and employees were not complaining they would not have raised wages.