At the Foundation for Economic Education Alston Ghafourifar presents some information which is obvious to anyone who’s actually studied the issue but I’m sure will be bitterly rejected by many. Automation is a net creator of jobs, as true today as in the 20th, 19th, and 18th centuries. The most compelling evidence he offers is from a recent study in the United Kingdom:
A Deloitte study of automation in the U.K. found that 800,000 low-skilled jobs were eliminated as the result of AI and other automation technologies. But get this: 3.5 million new jobs were created as well, and those jobs paid on average nearly $13,000 more per year than the ones that were lost.
Positive, worker-friendly outcomes like this illustrate a more complete range of possibilities for automation. Technology is changing the way we work – that’s not in dispute. These changes can improve people’s lives and lead to a more creative, intellectually engaged workforce. AI often means that employees can spend more time on complex tasks for which they are uniquely suited, like interacting with customers or brainstorming innovative new campaigns.
The other examples he cites, Panera whose experience I’ve written about in the past, and Marlin Steel, whose experience is characterized here:
Drew Greenblatt, CEO and owner of Marlin Steel, credits automation with not only providing a lifeline to the company but to its employees as well. “All of a sudden they’re super productive and it’s because we’ve given them the tools – it’s robotics and automation,†he said. “Thank God for robots. If it wasn’t for robots, these guys would be unemployed.â€
Leveraging robotics helped Marlin Steel land major clients by creating higher-quality products. The company’s success suggests yet another positive side effect of automation: the opportunity for companies to enter into higher-margin product lines. Staying competitive means always finding a new edge, a differentiator that inspires people to choose your company over the next.
are both anecdotal but also indicative. The experience of the last 300 years is that automation and efficiency actually produce more jobs than they destroy. That’s why, despite the large number of farming jobs that have been lost over the years, more people are gainfully and productively employed than at any time in human history.
Tellingly, there are no counterexamples over that period. While the people in nearly every country are living better than ever before, no country has experienced persistent massive unemployment due to automation.
That tells us that persistent technological unemployment is a risk to be mitigated rather than an issue. The graver concerns are the factors that are preventing new jobs from being created: governments taking too much money out of the private sector, excessive government regulation, barriers to trade particularly the one way autarkies favored by some of our major trading partners, and protections being given to favored occupations that are prime targets for automation.
What is true is that the sorts of jobs on offer will change. There aren’t nearly as many jobs plowing fields behind a team of mules or hauling loads by driving a team of draft horses as there used to be and I strongly suspect there are few who want to see those jobs return. There are millions more truckers than there ever were of those original teamsters and they’re paid a lot better.
As far as I know, Scalamandre went to computerized looms in South Carolina, keeping its showrooms in major cities:
http://www.scalamandre.com/customer-service.html
“AI often means that employees can spend more time on complex tasks for which they are uniquely suited, like interacting with customers or brainstorming innovative new campaigns.”
Automation is gradually eliminating manual labor which is great for many people, but is perhaps not so good for those who enjoy it or who lack other skills.
“The experience of the last 300 years is that automation and efficiency actually produce more jobs than they destroy. ”
I think that’s true, but I don’t believe it’s some kind of iron law destined to continue in perpetuity and I think it may be starting to change. For all of that 300 year period people could find work with low or no skills. Automation and the trend toward more complex work mean that such jobs are fewer and fewer – in other words the bar for employment is rising. This bar is still low enough that most people benefit but, inevitably, as the bar rising there will be more and more people who are left below it. At the same time, globalization and immigration ae expanding the pool of low skill workers who compete for those fewer low-skill jobs. The end result it that it really sucks to be in the low-skill labor pool and making some kind of living or getting out of is harder than ever.
So how does this end? I’m not sure, but I’m skeptical of the “same as it ever was” argument.
I agree with that. But evidence is lacking that it’s changing. Simply pointing to job losses isn’t evidence. You’ve got to find net job losses and we’re just not seeing that. Might we? Yes, we might. But right now the run-on effects of a lot of poor policy decisions so overwhelm those losses that reversing those policy decisions are a better place to start than trying to remediate the effects of a possibly real possibly illusory permanent technological unemployment would be.
I agree – I’m just thinking out to the future. It’s a hypothesis.
It would be an interesting exercise to develop a set of indicators for how it might play out.
Those were 300 years using asset backed money. Credit backed money will render different results. US manufacturing has increased, but it is creating financial goods and services – the financialized economy.
Automation destroys the highly skilled jobs. It allows the less skilled to produce the same goods and services as the more skilled. Before the automation of manufacturing automobiles, only craftsmen were employed. The assembly line allowed a farmhand to build a car.
Most software developers would be fast-food workers if higher languages had not been developed. Few of today’s software developers are capable of the skills required to use assembly or machine code. Higher level programming languages allow lower skilled workers to produce the same products.
There is no job that technology cannot ‘dumb-down’ enough to allow the less skilled to perform. At some point, technology will allow a cashier to become a nurse, a nurse to become a primary care physician, and a primary care physician to become a brain surgeon, and the brain surgeon will be out of a job.
Unless you actively intervene to protect the nurse, primary care physician, and brain surgeon which is just what we’re doing.
You’re looking at automation in just the right way. The real cost benefit isn’t in automating dishwasher and waiter/waitress jobs. It’s in automating the expensive ones. Jacquard looms only made sense because master weavers were highly compensated. But if you protect the expensive workers you get what’s being called technological unemployment but is actually just bad policy.
First, read and absorb this statement: The experience of the last 300 years is that automation and efficiency actually produce more jobs than they destroy. That’s why, despite the large number of farming jobs that have been lost over the years, more people are gainfully and productively employed than at any time in human history.
Now, look at an electoral map that breaks things down to county level — the red areas are pretty much all the areas where jobs have been lost.
The Automation Revolution in agriculture has wiped out hope in a huge swath of the country (no jobs, houses worth less than when people bought them, etc), and all of the new jobs are compacted into the largest cities (small cities are not doing so well).
There are winners and losers in the race towards automation, and we haven’t done enough to cushion the blow for the losers, and give them another chance.
Now, look at at an electoral map by state — huge swaths of America is in empty states filled with the losers, and by a quirk of our electoral processes, their votes matter more than those in the densely populated, more successful areas.
(If the GOP weren’t waging a war on minorities, the maps would likely have more red, as there are economic losers in cities too, but the GOP is really going after the white, rural losers)
TastyBits: Most software developers would be fast-food workers if higher languages had not been developed. Few of today’s software developers are capable of the skills required to use assembly or machine code. Higher level programming languages allow lower skilled workers to produce the same products.
High level programming languages allow mediocre programmers to write mediocre code quickly. Mediocre turns out to be good enough for the vast majority of cases.
Before then, you had higher skilled workers producing higher quality products much more slowly. It wasn’t cost effective for a lot of problems.
It’s not the same product.
The higher level programming languages aren’t displacing the people at the top — it’s making them more effective (higher quality products, not as slowly, in the areas where higher quality is important) — while expanding opportunities for the mediocre. (The mediocre engineers still aren’t going to do great things, but slapping together a simple website is easy enough). And, it’s a very good thing, since there’s more than enough work to go around in the field.
The same applies to the brain surgeons — former Primary Care Physicians aren’t going to become brain surgeons, or even regular surgeons. They will, however, move up a half step to handling full-body-care-and-interactions while nurses handle the minor care.
And, in medicine, that’s a good thing, since we don’t have enough PCPs to handle the work, when people are going to their PCP for a sinus infection.
“Now, look at an electoral map that breaks things down to county level — the red areas are pretty much all the areas where jobs have been lost.”
The claim that the red (GoP majority) counties are where “all the jobs have been lost” is an extraordinary one and demands some evidence.
“Now, look at at an electoral map by state — huge swaths of America is in empty states filled with the losers, and by a quirk of our electoral processes, their votes matter more than those in the densely populated, more successful areas.”
The “quirk” was entirely intentional. Google the “Connecticut Compromise.”
High level programming languages allow mediocre programmers to write mediocre code quickly. Mediocre turns out to be good enough for the vast majority of cases.
I would like to see evidence proving the statement. Complicated systems have been chosen to be written in high level languages such as python for YouTube and erlang for WhatsApp. The programmers at google and Facebook are many things, but knowing some the last thing I would them is mediocre.
Well, as your friendly process engineer turned manufacturing company owner, the assertion that automation is a net job creator is as obvious as the nose on your face. However, there simply are few advocates for the less visible newly created jobs, and plenty of advocates for the very visible people losing jobs.
I don’t see any evidence we have reached some magic tipping point where automation has tipped toward a net job killer. I see no evidence of companies preferring to hire legions of minimum wage surrogates.
Instead of automating. However, I do think our neglect of a national education strategy that addresses the increasing requirements a more automation intensive environment implies is shameful. In fact, from where I sit, we are moving in the opposite direction.
The cause of job losses is not automation. It is a financialized economy. With government policy and intrusion, job creation has been shifted to the finance, education, and healthcare industries.
Restricting foreign governments purchasing treasuries would force the price of imported goods and/or the purchase of exported goods to increase. Either of these would increase job creation.
(I know – monetary system, blah, blah, blah …)
@Gustopher
I believe that the people you reference are the creative people, and creative people will always be needed.
Assuming that you are not being trite, a higher quality good is the same kind as the poorer quality version. A Rolls Royce and a Kia are not the same quality, but most people would agree that they are the same product. A top fuel dragster is nominally an automobile, but most people would consider it a different product.
If you mean innovation, I agree that they do not produce the same products, but innovation is about creation. The higher level software languages were developed by the creative types, and they would produce different products up to their maximum creative ability.
The IT world takes great liberties with the concept of engineer. Engineering is about rules. It may not seem so, but these rules have been automated using standards and codes. Engineering is about remaining compliant, but a creative engineer may be able to bend the rules. New materials and techniques do require new rules.
Any process that is rules based can be automated, and most likely, brain surgery can be automated. A mechanical hand can be limited in the amount of pressure or movement allowed. A mechanical eye can ‘see’ the entire electromagnetic spectrum. Once the technology has been developed, a car mechanic or librarian will be able to perform brain surgery.
One or two hundred years ago few people had enough skills to work today. It is possible that the human animal is able to evolve quicker than any other animal, but to me, this seems unlikely.
@CuriousOnlooker
I somewhat agree with the thrust of @Gustopher statement, but it is really just technological snobbery. Coding Facebook using assembly is possible, but I doubt there are many people capable of doing it. For enhancements, the base code could be reused, but it would eventually become a mess.
Even the most skilled developer uses a compiler because it is substantially easier. Of course, it is possible that the non-mediocre developers are using machine language, but I doubt it.
High level programming languages is a bit unlike say an assembly line or a mechanical plow because it’s a productivity multiplier but not a productivity leveler.
Good programmers are still orders of magnitude better then mediocre ones when using compilers; that’s why there’s still a pretty hot market for good ones.