Sam, your points are valid but I think what it points to is something that we’ve talked about upthread: that the way the unemployment tends to happen is not by suddenly wiping out an entire job category and replacing every worker with automated replacements. It’s more a case of steadily adding automation so that it takes fewer humans to get the same amount of work done.
So I picture for instance with the truck-driving thing, they might have someone at the loading dock who spends the day getting trucks ready to roll. And then maybe there’s someone in a call center who takes over some of those other tasks you talked about, but for 10 or more different former truck drivers. Even if you still have people doing these things, the fact that on the interstate, the trucks are just rolling along night and day without anyone behind the wheel, definitely kills a lot of hours that would have previously gone on thousands or millions of human beings’ time sheets.
But we’re not talking about marginal changes, where workforces are reduced by some fraction due to automation. That’s clearly happening, and will continue to happen. We’re talking about a future where the cumulative effect of these marginal changes is that no one needs to work any more, that robots will ‘take over’, that we will need to provide basic incomes to people because there are no jobs left, etc.
My point is that there are limits. Automated driving might eliminate 10% of trucking jobs, but it won’t make truck drivers obsolete - not for a long, long time. And therefore, the rule of history still exists - human brains have economic value, and since automation only happens where it makes things more efficient, it just has the effect of opening up more opportunities for other jobs humans can do. That’s been the story of the industrial revolution, and I don’t see why it’s different now. The types of jobs lost will be different, but the underlying pattern of job loss/job creation remains.
We could have had this same argument at the turn of the 20th century as mechanization of agriculture began to wipe out tens of millions of farm labor jobs. In fact, some argued for permanent income assistance for displaced farm workers on the grounds that they were uneducated, they lived far from the factories that employed people, and therefore they would starve unless the government provided for them. Instead, they adapted, and whole new industries were created to take advantage of the labor that was freed up by the machines.
But imagine if all those laborers had been given a guaranteed income, so long as they stayed on the farm. Today we still might have an ex-farmer underclass living on subsidies, and their existence would be used to prove that the subsidies were obviously necessary. A self-fulfilling prophecy.
This is not a fantasy - it’s happened in Canada with our fishing industry. It’s happened on native reservations. It’s happened in inner cities.
So there’s a big difference between, “The robots will take over some percentage of current jobs, but this will grow the economy and create new opportunities for the displaced humans”, and “OMG! Robots are taking all the jobs! Soon there will be nothing left for us to do! We need a guaranteed income because all those people won’t have any way to make a living!”
In a broad sense I agree with you, but I don’t think you make your point very well.
I think automation has the potential to take away a lot more than 10% of trucking jobs given how many man-hours are logged on the road. And the way agriculture changed during and after the industrial revolution, shows us just how disruptive such changes can be.
And while there are many technical issues with getting automated trucks on the road, no-one thinks it would take anywhere near as long as the epic changes that happened to agriculture and rippled through society.
So those are the two arguments for an improved social safety net or even guaranteed income:
Firstly, the changes will be very fast. While ultimately employment may recover or even increase thanks to a growing economy, in the short-term a lot of people may be out of work and/or wages may be greatly suppressed for a time (we’re already seeing the latter).
Secondly, in the long term it may be that we can fulfil all our needs barely involving human labour at all. This is A Good Thing, but will need new models of taxation and social security.
The latest episode of the Planet Money podcast is a rerun about UPS. They go into how regimented the job has become, to the point that the computers at HQ know how much backing up each truck has done: where they were when they did it, how many feet they backed up, etc. When asked if they get chided for doing too much of that or doing it improperly, the driver responded “every day”.
The “data guy” they talked to cited things like “one minute wasted by each driver each day costs us $14 million a year”. They have (or had, when this was a new story) just cracked the vexing “traveling salesman problem”, with the help of a lot of Ph.D.'s and a lot of supercomputers. The driver said he was thankful they haven’t gotten to the point of making him like a robot, where he has a voice in his ear saying “turn right, now turn left”, etc. The closing comment by the Planet Money reporter? “The company says they are working on it.” That was a bit chilling: very reminiscent of the story someone shared early in this thread, about a system that turned employees into meat puppets in just this way–headsets that told them where to walk, when to turn, what to do. And then when the robots were ready, they were able to use the already-established system of AI organization, but just swapping in steel and plastic for organic flesh and headsets.
But in this case if the driver is simply executing a plan created by some automatic process, then it is a really horrible job and humanity would be better of if a robot can drive the truck.
Or take the long-haul trucker. Maybe the truck needs a human on-site to deal with maintenance, paperwork, ad hoc emergency schedule changes, and so on. No reason for the human to be cranking the steering wheel with his foot on the gas. He can be sleeping in the back, or doing other productive work, while the truck drives itself down the endless highway. Or the single human could be supervising a convoy of trucks.
It’s not a case where every trucker loses his job, it’s a case where one trucker can do the same amount of work that used to take 5 truckers. And so employment as a truck driver hollows out.
Planet Money has another episode about automation, this time the tricky task of making robots able to sew. Turns out they have to have very sophisticated “eyes” that take a thousand frames per second and count every thread as it goes by.
Once this technology is perfected, we may see a lot of clothing manufacturing move back from Third World sweatshops to the U.S. But there will only be a few, relatively highly paid, workers at each plant. So not much of a job boon here, and it may deal a crippling blow to those low wage economies that do the sewing now.
That page also contains links to several other stories they have done about “automation and the future of work”.
NPR has a fun story with a cheeky headline about “our robot overlords” making and delivering fresh hot pizza. As we have often discussed here, it’s another case of not flat out replacing all humans involved in the economic activity, but reducing the numbers massively and leveraging the one remaining person to do the work of many.
Not much, I expect that the promises of Trump that convinced the rust belt to vote for him were poppycock. Those jobs are not coming back.
The loss of those jobs was not only because of trade and other internal developments (like gas beating coal in the market even before any new regulations were in place), automation was not the main factor there because it would had affected many other regions too.
It’s possible that many people see the disappearing jobs and blaming immigrants (I know that H1B visa workers get a lot of the blame for IT workers being put out of their jobs) and don’t realize that a much larger portion of the jobs are automated rather than outsourced.
Yes. I have actually heard offhand comments in news reports several times to the effect of “The U.S. actually does still have a lot of manufacturing–it just doesn’t have a lot of manufacturing jobs” because so much of modern manufacturing is automated. So the ludicrous promises Trump has made to them are almost literally the same as a presidential candidate in 1916 promising to bring back jobs in the buggy whip industry.
Perfect–thanks! Boy, how many people know that manufacturing jobs reached their all-time peak during the Carter Administration? I didn’t.
So manufacturing output has gone up roughly 30% since 1998, while manufacturing jobs have gone down roughly 30% over the same time span. Therefore it’s almost true that manufacturing requires half the human labor it used to. That’s huge. Next comes the other half (or most of the rest anyway).
As we’ve been discussing all through this thread, we’re going to see a lot of turmoil while society digests this change, and I definitely think the rise of Trump is part of it.
I want to see us go to a guaranteed income or “mincome”, but a lot of people really feel something like the “Protestant work ethic”, a fundamental dignity in working. So maybe it will have to be done in a sneakier way, by subsidizing businesses to offer make-work “jobs” that aren’t really needed?
Manufacturing as a percent of all jobs peaked in the 40s. I expect that’s because we’re just looking at civilian workforce (I haven’t actually pulled the BLS dataset) in addition to the war effort. It dropped below 30% in the late 50s and has been in decline ever since. It was only 8.5% this summer.
We’ve been “digesting this change” since my parents were kids. And that was a bigger change than if the remaining manufacturing jobs evaporate over the next 50 years or even twice as fast.
It’s not bigger if the service jobs and many professional jobs disappear (or, more likely, get consolidated to a very small number of humans using AI augmentation to be incredibly productive) at the same time.