A regular commenter’s response to my post yesterday on technological unemployment was terse:
1) We produce more…
2) With fewer human workers.
3) How?
My answer is equally terse. The same way it has been for tens of thousands of years.
Technological advances have created surpluses. The surpluses have made new jobs possible or even necessary. Some are in primary production. A century ago practically no gallium or germanium was produced commercially. Now they’re essential to our civilization and new jobs mining and refining them have sprung into being. Thousands of tons of the stuff are produced.
At its height there were hundreds of thousands more jobs in the auto sector than there ever had been in the carriage manufacturing sector. That’s an example of increased production not resulting in technological unemployment but in the creation of many more, better paying jobs.
But by far the most jobs will be in secondary production. What will those jobs be? We have no idea but the Department of Labor says that 65% of today’s schoolkids will work for businesses that haven’t been created yet.
Here’s another example of technology making jobs possible that hadn’t been possible before. Writing. According to the Bureau of Labor Statistics more than a hundred thousand people earn their livings as authors or writers in the United States alone. Four hundred years ago practically nobody did anywhere. What happened? Advances in technology.
People have been complaining that their jobs were becoming obsolete due to machines for 200 years. While there has been some short term technological unemployment, every instance of that is a tragedy for those involved and I don’t want to minimize that, there has been no massive unemployment due to technology. That was true in 1820. It was true in 1920. It was true in 1990.
Is it possible that today really is different? Sure. But there are very real costs associated with guessing wrong about something like that. The costs not only take the form of the direct costs of dealing with massive technological unemployment that never happens but in the opportunity costs of doing that and the deadweight loss that results.
Additionally, there are other explanations for what’s happening right now and those explanations are demonstrable. Among these explanations are foreign competition, immigration, and deadweight loss (deadweight loss alone is estimated to cost millions of jobs). Taxi drivers and coal miners aren’t just victims of technological obsolescence. Policy preference also pays a role. Treating policy preference and technological obsolescence as though they were the same thing is an overreaction as well as making it more difficult to devise solutions.
The historically slow rate of new business formation is one of the factors in slow job growth, too. New businesses produce new jobs. Old, established businesses cut costs by, among other things, cutting jobs. Over the period of the last thirty years most net job creation has been by new businesses. We don’t know why the rate of new business formation has slowed but policy preference seems to be a likely explanation.
Just to head off a possible complaint about some of my observations above, the alternative to the adverse effects of taxation and regulation isn’t no taxation or regulation. it’s better, more targeted taxation and regulation.
Under the circumstances those who are concerned about technological unemployment need to produce much stronger evidence for their theories than they have to date. It’s not enough just to ask how.
The answer to ‘How” is “with machines.” That’s how we increase productivity. More machines doing more jobs more efficiently than the human they replace.
So, without question, X number of jobs have been placed in the “Machine” column and removed from the “Human” column. You point to the past and make the case that we’ve always found new jobs for people. But eight years ago when we were thinking we might need a new WPA we realized that sort of thing doesn’t really work anymore. Why? Because all those ditch-digging jobs are now in the “Machine” category. So, at the very least we have lost a tool we used to have.
But that’s a secondary issue. The problem is that a typical human has an IQ of 100 and the strength of one man. Once upon a time 100% of humans were stronger than machines, and now 0% of humans are stronger than machines.
Now machines are climbing the IQ ladder. They started off as brutes, became toddlers, and are now closing in on duplication of all but the most elevated human intellectual capabilities. Once upon a time 100% of humans were smarter than machines, and now less than 100% of humans are smarter than machines
How many humans can today compete physically with machines? None. We are physically limited. And we are intellectually limited. Right now machines are on the intellectual ladder somewhere in the “dim-witted” zone. They’ve gone from intellectual zero to effectively become the second smartest creatures on earth after homo sapiens.
Yes, I am using intelligence rather loosely, yeah, yeah, I get that AI isn’t the same as HI, but it doesn’t matter to the competition. Because they don’t have to reach IQ 200, or even 100 to wipe out millions of jobs. Machines don’t have to be able to write Hamlet. They are already, by their nature as a ‘species’ 1) stronger, 2) cheaper, 3) more reliable. So as they climb the IQ ladder they will already have three outstanding virtues that make them inherently superior as employees.
You’re pointing to machines climbing the physical strength ladder and finding reassurance because humans still had work. True, that phase of machine evolution didn’t destroy jobs for one species – homo sapiens – but it destroyed the hell out of employment in the horse, oxen and donkey job markets. Because at that point machines were not in competition primarily with us, they were effectively in competition with other animals. Physical strength for physical strength and guess what? The donkeys lost.
And now the machines are no longer looking to replace horses, oxen and donkeys, the machines have already won the battle to replace the animals. Now they’re looking to replace us, and except perhaps at the outer limits of human intelligence they will succeed very quickly in doing just that.
Where to begin?
The problem with the Intelligent Design theory and most Creationist based “science” is that it assumes today was the only possible path possible. This is the problem with the Malthusians and the “here come the robots” crowd.
The Industrial Revolution made it possible to support the number of people alive today. If the robots take over the world, the jobless, penniless, homeless, shoeless, coatless, foodless will starve, freeze, not pro-create, and quickly die. With nobody to purchase their robot produced products, the filthy rich will soon be simply filthy.
Unless society suddenly and catastrophically collapses, it will never happen because there are an infinite number of points between now and the robot caused horrors Robots are only cheap relative to human labor under certain conditions.
The lever, inclined plane, screw, and pulley are a few of the simplest machines, and they are all stronger than any human.
Why is it that almost every person who believes that they have the world’s greatest IQ do not have any idea of what the hell it is? First, it requires Intelligence, and second, it is a quotient.
Computers do not have intelligence. Computers can only think what they are programmed to think. Any “thinking” computer cam be overcome by another computer running through the same algorithms. They are all clones. Even if you make mine slightly different, I can change its programming to match yours, and bingo, I know everything your computer is “thinking”.
True intelligence requires real randomness or a random number generator (RNG), and this is a lot harder than it seems. The bigger problem is that a computer which is capable of random “thoughts” will be capable of psychotic thoughts, and a truly thinking machine will be capable of mass murder.
(If you play computer games, you can pick out the patterns. Some of the games have really good AI simulations, and they will add in an RNG factor to disrupt the patterns. This usually works, but the RNG is not really random. I am not referring to a fixed seed.)
On The Walking Dead, the zombies can live for years without feeding, but in the real world, biology would solve the problem. The robot take-over can be thought of as a pathogen outbreak. With the birthrate declining at the upper end of the scale, it may be occurring already.
If anybody is really concerned about jobs being taken over by robots, illegal aliens, or overseas workers, they should be concerned about the credit backed dollars that make it all possible, but that would mean less Walking Dead time.
Rather worry about robots taking over the US jobs, a better worry might be that the credit backed monetary system has become so over inflated that it would take a generation to pay down all the outstanding debt, but in an attempt to fix the over indebtedness problem, the high IQ geniuses are increasing the amount of debt.
Michael:
I see where you were coming from now. However, let me suggest something for you to think about.
“Machine intelligence” isn’t what you think it is. It’s a lot small, simpler, and less revolutionary. The simplest version is that although you might have a fully autonomous car sometime in the future it still won’t be able to make you a sandwich.
There hasn’t been much in the way of major developments in machine intelligence in decades. What we’re seeing now is a combination of relatively minor elaborations in artificial intelligence and improvements in processor speeds, memory sizes, and network speed and reliability, all at lower costs. The developments aren’t linear. If it takes 10 years to develop a vehicle that’s 68% autonomous, it could take centuries to to come up with one that’s 99.997% autonomous.
A couple of decades ago the Japanese (who have different workforce issues than we do) seriously curtailed how they were using robots in industry. It didn’t make economic sense to do anything else and I suspect we’ll decide the same thing.
Increasingly, robots will take over routine, repetitive jobs. They won’t take all jobs and, if we allow them to be created, there will be lots and lots of jobs for human beings that haven’t even been imagined yet.
On the subject of fully autonomous vehicles, I’m 95% confident that the entire present fleet will not have been replaced in 20 years, 99% confident that it won’t have been replaced in 10 years, and 99.997% confidence it won’t be replaced within 5 years. I’m about 68% confident that it will never be replaced by autonomous vehicles.
You know what the best prospect for fully autonomous operations is? Railroads. Why hasn’t that happened? It could have been done decades ago.
@Dave Schuler
… New businesses produce new jobs. Old, established businesses cut costs by, among other things, cutting jobs. …
I meant to add that this was an excellent post, and this was the best sentence. Actually, this should be the criteria for any legislation considered a “jobs bill”. Cutting taxes for old established businesses will probably not do much, but reworking the Intellectual Property laws will allow new businesses to be created.
Two points on cutting taxes: First, not endorsing cutting does not mean that raising taxes is good or that cutting is never good. As with most things, it depends on the situation.
Second, there may be something that can be done to the tax code to encourage/allow/foster new businesses (or business units) being created. I am not a tax expert, and therefore, I have no idea.