Experts say demographic shifts, not AI, create the biggest challenges for work
gsb.stanford.edu
gsb.stanford.edu
Imagine a society with two classes - an upper class creating & sustaining 'services' like horse-walking or tinder coaching, a lower class forced to choose between unemployment and being the other class's glorified servants.
We need to be looking at ways to level the playing field - provide opportunity and education to those who were not fortunate enough to be born into it. Celebrating the stability of jobs which make people feel like second class citizens feels like shrinking away from the problem.
I think they're the opposite of glorified. A servant is someone you personally know, that your children may grow attached to, that, ideally, you feel some mutual obligation towards. Workers in the gig economy are faceless, replaceable cogs.
Abuse of servants happens every day all over the world.
The more you reduce human contact and any chance of forming a relationship, the more cog-like workers will be treated. And reducing human contact in the name of efficiency is exactly what all those gig economy services do.
Just because some servants are already abused, doesn't mean gig economy won't make it worse.
Basically, they are saying that AI both creates and destroys jobs. Well no shit.
The problem is, the average truck driver isn't going to retrain and relocate to San Francisco. He will just be out of a job.
Seems like the problem isn't with demographics but the economic system if the problem is perpetual and impossible to correct.
It seems to be qualitatively different than when people left farms to go to factories, or factory jobs to office jobs.
(edit: typo, added last sentence)
There is a fringe subfield of steady-state economics: https://en.wikipedia.org/wiki/Steady-state_economy
But we won't be back to the same demographic issue, because the issue that they people who retire now in the west are the last generations that were bigger than their parents generations (it has a formal name, but you can roughly think of it as many kids born, many kids die transitioning into many kids born, few kids die into what we have now which is few kids born, few kids die).
The argument is obvious in retrospect but, call me a dummy, I hadn't considered it until I heard it.
The act of automating is expensive, regardless of the work involved. You want a big return on investment. so You're going to automate the expensive, high-skill jobs first and work your way down. Radiologists, and financial analysts are expensive so we automate away their jobs.
If you want a decent paying job for the foreseeable future, become a plumber.
[0] https://www.recode.net/2019/1/28/18199981/ellen-shell-job-au...
The thesis was that everyone keeps getting automation wrong, and why the coming revolution will be far more different than what has come before. I even tried to extrapolate and show from the history of AI, just how wrong we've gotten it before and the few things that we keep on systemically underestimating.
Hope you enjoy this!
That Siri slide also just gave me PTSD.
Technology creates new jobs but reduces the number of people and processes needed to do the same job.
Technology creates new value but reduces the value of the contribution humans make.
Technology creates new markets but shrinks the amount winners in those markets
Technology compresses, through digitalization, the number of steps involved in delivering products. (Music and movie industry to name two examples)
Technology increases your technical advantages but forces other competitors to do the same thus reducing the advantage of the individual until you own most of the market.
Globalization creates bigger markets and temporarily offsets the need for technology to provide a more cost-effective competitive advantage.
Outsourcing is the step before automation.
Technology keeps improving while humans don't.
The value added by the individual in a job function is mostly only a fraction of what the human is capable of doing.
Technology is really good at solving these simple tasks and are becoming better and better.
Ironically a cleaning lady will be out of a job much later than a radiologist is.
The jobs that are left for humans are low-value jobs and will be pushed down even more as more and more humans become available to the job market because technology renders them useless in other situations.
Even if all those jobs eventually get automated, at this point they will be done by robots in China, not in America. It's still a big loss for the US.
Since robots cost the same to operate in US and they do in China, factories will shift location closer to the markets they supply.
EDIT: my point was not clear enough. I'm saying in the long-term automation will reduce the cost of labor close to zero, leaving only the cost of inputs and distribution (supply chain). Not to mention the IP risks of locating tech in China, advanced automation will make it more cost effective to shift production to be located next to the markets where those products are sold to consumers. China is well aware of this, which is why the current government policy is focused on boosting the local innovation and technology economy.
Classic "We moved the supply chain to China because it's cheaper --> We can't build stuff in the US - all the factories are in China!"
Infact, I agree with both of your points.
The problem is, supply chains are not as entrenched as you claim.
Given a long enough time horizon and advanced enough automation, robots will move back to USA. Especially consider the IP risks and political concerns WRT China.
It will take a long time for all industries, some will happen before others, but there will be more vertical integration through automation and it will become more cost effective to have factories (not jobs, there will be no jobs from this) for products next to the markets they serve.
This is what most economists believe, and what the Chinese government believes, which is why they are pushing to develop AI and innovation.
The effect will be the same though: the site of manufacturing will change. But like the Star Trek transporter, no matter actually gets moved.
> wherever costs are lowest -- labor, raw materials, power, regulations, shipping.
Due to automation, labor costs will be less and less of a factor. Ocean shipping is already very cheap, and comprises a negligible fraction of the cost of most products - it's how markets in Europe are full of Chinese garlic. Raw materials can be shipped, so those are also not much of a factor. That leaves only power and regulations as factors. A smart country is likely to make sure those don't work against it, and to outsource only the most environmentally-harmful production steps.
There are still tariffs, but being willing to host polluting industries is otherwise dangerously close to being the only advantage an under-developed country has left.
Nope. Labor, land, resources are all typically cheaper in china. It's also close to other manufacturing. I've read on this elsewhere; it's called "economies of proximity" or some such (I can't recall exactly). But it means you don't have to ship inputs or outputs as far.
What I am saying, in your words, is automation will shift the "economies of proximity" equation to make more sense back in the USA.
This is a big part of the reason why America winning the "AI arms race" is so important.
This article says Chinese median age in 2019 is 37.3 and us is 36.5. That's interesting that we are the same. China has the other problem of producing too many men and not enough women, leading to increasing problems of pressure. That problem of not enough women to match the men seems like it will be a worse problem.
http://www.worldometers.info/world-population/china-populati... https://en.wikipedia.org/wiki/Demography_of_the_United_State...
It seems to me that it is part of the governments job to say "yes this is economically feasible, but not socially acceptable." Outsourcing breaks this dynamic by introducing moral hazard--we can have our cake and eat it too. I think if there were a minimum global standard of living, most cheap plastic shit just wouldn't get made.
With this model, even if some jobs shift to China or are automated, that just means the central bank can step on the gas a little bit more, and employment and standard of living is unaffected.
One problem with this model is structural problems where say jobs are being created but are in a different place than the ones being destroyed. People may be unwilling to move or it may be too expensive. They also could require different skills.
Another issue is that the central banks are limited in their ability to lower interest rates (their traditional way of stimulating the economy and getting out of recessions). Fiscal policy (Gov. borrowing and spending) is not expensive at all now (the Gov. currently pays hardly any inflation-adjusted interest on the debt), but if interest rates were ever to rise substantially having a large debt could be a problem. Spending the money efficiently also could be a challenge and it's politically controversial to do it at all.
Personally I think something like a UBI is probably a good idea to insure against things just imploding. In the event of a recession they could just increase it to stimulate the economy, which probably can happen a lot faster than infrastructure projects etc. And if automation accelerates, it may be tough for large numbers of people to adapt quickly (even if they eventually do).
What does that make the person who demands a better working condition for themselves, but is fine that a percentage of work is outsourced to a sweatshop?
Even the most mundane industries were operating in the US just fine up until the mid 1900s.
You can rightfully critique and regulate the golden goose, but there is a point at which it becomes incapable of producing as many golden eggs as other countries.
What other reasons can we find?
* Product X is made by 3 companies in the US. It costs them $20 to make the product and it is sold for $30-33.
* Government increases minimum wage, which leads the cost to go up to $23 (e.g. cost of raw material is impacted, labor, etc), the producers increase the price to $33-35 while still taking a personal hit in their returns.
* Unexpected economic disruption or shortage in a certain raw material leads to increased costs to $25 and increased consumer price to $36-38
* Government implements a new legislation that impacts customer demand negatively and increases cost of production to $28. This leads to increased consumer price to $39-40
Throughout this process, supply/demand dictates that the increase in prices may impact negatively the demand for the good produced.
* One of the companies realize that it can produce the goods in a country called Melo for $5. It shifts a good portion of its production there, allowing it to sell the goods to consumers for $20.
* Customers sensitive to price will flock to the cheapest good. All the other companies decide to move their production to Melo country to remain competitive.
In part, consumers are willingly exchanging a few jobs here and there so that they can get goods & services at a cheaper price. Economics is full of trade-offs, understanding them is essential for a country's long-term success.
If American society insists that we legislate ourselves into a more comfortable environment (higher min wage, better working conditions) then I think we'll have to accept that certain production will just never happen here.
What other reasons can we identify?
It can happen here, if you break with neo-liberal economic dogma, and institute equalization tariffs on imports.
The other thing with tariffs is if you put tariffs on steel that hurts automakers (since foreign ones have cheaper steel). You can then put tariffs on cars, everything, but then everybody puts tariffs on you. Also things like computers are more expensive which might hurt productivity. Most economists don't seem to think the whole thing is worth it, and there's countries like Brazil that do big tariffs and it doesn't seem to help them any.
We're currently pursuing option #1, but neither one sounds fantastic to me.
I think legislation like minimum wage is only a small part of the story too. Lots of people get paid more than minimum wage and we couldn't just put min. wage at $100/hr.
I think the fundamentals we are looking for are productivity and macroeconomic policy. Traditionally, the Gov. and central bank stimulate the economy until it reaches full employment, then slow it down when inflation is high.
Productivity will determine how much people produce/consume (in total) when everyone is working. Of course there's issues of how those goods are distributed (which is probably not quite as bad as the wealth gap, the rich put a lot of money in stocks etc instead of consuming it).
There's also structural problems where the job locations/skills don't match with the workers. While I don't see why in the long-run we can't just produce more of what people want (health care, going out to eat, housing, vacations, etc), if millions of people are displaced rapidly by online shopping, self-driving cars etc I think it will be hard for people to adjust quickly.
Also maybe most worrying is some of the macroeconomic stuff that has been relied on to keep the economy closer to full employment is in uncharted territory. The central bank has less ability to stimulate the economy to get us out of a recession than maybe it's ever had, and the alternative of fiscal policy (gov. spending) is controversial so I'm not sure how much it can be relied on in the US either.
Minimum wage doesn't come into it. No employee earning $20/hour is going to accept that tomorrow she'll get $6/hour for the same job. (One reason is pride. Another is information asymmetry. As an employee being offered a $14/hour wage cut, you can't know whether the company is telling the truth about the equivalent cost of labor in China, or is just trying to cut your wages.)
The problem with the simple model of free trade where workers just lose (why pay $10 in the US instead of $2 in Mexico etc) is that they do not pay workers with dollars in Mexico, they pay pesos. The dollars actually stay in the US. What the total effect of free trade is depends on what happens to those dollars. Are they being used to buy US goods, investments, government debt (then what is the government doing) etc.
The fact that you pay in pesos eventually means if pesos lose value internationally you could pay them less (but dollar cost is the same). But that's not guaranteed.
A trade deficit means overall the other country is buying more investments instead of goods. But those investments are in the US. To actually spend the money somewhere else they would need to exchange it again.
And those investments in the US aren't inherently a bad thing. If it's just buying US government bonds and the government not doing anything productive with that money it might be a bad thing though.
It's cheaper, it buys you flexibility and it makes "re-tooling" for other product lines a matter of switching suppliers.
There are many excuses that allow a company to end-up in a situation where they're little more than finance, product/sales/marketing management, and scm-specialists sitting in cubes with headsets on "hitting their numbers". I find it all very sad and unsustainable in the long term. If China eventually "eats our lunch" economically, we will have richly deserved it as a nation.
https://qz.com/1269172/the-epic-mistake-about-manufacturing-...
Echoing Autor's comments: "It’s a great time to be young and educated. But there’s no clear land of opportunity” for adults who haven’t been to college..."
We live in a world with almost infinite amounts of information, yet how we process and use that information (including labour market information) will be our next challenge. And understanding labour market shifts in the context of multiple forces AND their interactions (i.e. how does a job become automated in practice IN the context of a higher rate of retirement; what does a person actually need to perform well in a job, etc.) will require some novel innovation that I think data + AI can actually help us solve.
I think that demographics will play a huge part for a decade or two. There are two breakthroughs that I anticipate will happen and will change things.
The first thing we need is a really efficient and powerful artificial muscle combined with truly biomimetic mobility. This is the type of thing that needs to be invented to get humanoid robots that are truly dexterous and fast.
Along the same lines we need more general purpose AI to go in those new robots. It doesn't actually need to be fully human-level AGI in terms of generality or capability. It just needs to be more general purpose than current systems and be integrated into the fast dexterous robots.
The future ain't looking bright.