Ezra Klein's podcast with Larry Summers wasn't just about inflation. Ezra also asked Summers to name three books that had influenced him:
Third, a book that will come out in the next several months, Brad DeLong’s “Slouching Towards Utopia,” which is, I think, a really remarkable and powerful placing of all of economic history in perspective, that gives a sense that at some level I had known but never appreciated of how profoundly different the 20th century was than all other centuries and points towards the combined power of science and markets to change the world profoundly, and sometimes, in some ways,
for good, and sometimes, in some ways, for ill. I think anybody who wants to propound about economic policy should read that book.
Now I want to read Brad's book too! I remember giving some remarks a few years ago that touched on the most important events and developments of past and future centuries. I opined that the 19th century was the century of technology: steam engines, germ theory, evolution, electricity, and continent-spanning railroads. The 20th century was the century of market economics: Kuznets and Keynes, active central banking and the end of gold, the decline of tariffs, frictionless international banking, and flows of capital around the globe that would make the old doges weep. Now we're in the 21st century, which will be the century of artificial intelligence.
So yes, we are living today in the aftermath of the Industrial Revolution and the Economic Revolution, which makes the world profoundly different from any previous era. And now, having barely mastered basic science and economics, we're barreling toward the Digital Revolution with hardly a thought about how that will change the world just as profoundly. Mass unemployment will prompt revolts that make the Luddites look like monks and will likely kill off liberal democracy. With luck, we'll avoid being too stupid and greedy about this transition and both liberal democracy and market economics 1.0 will be replaced with something better and far more rewarding for future generations. But there are no guarantees. It's usually not a good idea to bet against greed and stupidity whenever the overclocked apes h. sapiens are involved.
Eh, we've got a pretty heavily aging population that's not even at the replacement level. And in any case, AI is not synonymous with robotics, etc - it's not enough just to have AI, you actually would need the systems necessary for it to replace workers, and for that to be cost-effective compared to hiring workers.
If there really is mass unemployment, then workers are going to be pretty cheap in comparison to the upfront costs on robots. And if you do have robots and AI that can do anything people can do cheaper, then you've got the seeds for your new UI/welfare system in kind - use the robots to provide and support it.
Humans are actually pretty decent self-replicating machines, albeit with high maintenance costs.
Exactly. It's really really easy to make a human being. Making machines that can come anywhere close to replicating the vast majority of tasks that humans can do is extremely difficult.
KD has been conflating the two for years now. And as usual he doesn't provide any sort of meaningful definition of AI, which is something you need if you are going to have a conversation about it.
If by AI you mean something like Siri or Google that can interpret human speech and bang through databases quickly to find info then we are already there. If you mean programs or machines that can perform complex creative tasks reliably based on minimal input and ask questions intelligently to, for example, generate a nice looking UI for a mobile app, well we are a very very long way from that. And if by AI you mean a program or machine with volition, that actually understands and thinks and has a will of its own, that will likely never happen. And the above three concepts have almost nothing to do with one another but can all plausibly be called AI, which is what makes having an intelligent conversation about AI nearly impossible.
Also, everyone should watch the video of the orangutan driving a golf cart. It is certainly much easier to train an ape to drive than to create a self-driving car. Think about that.
> It is certainly much easier to train an ape to drive than to create a self-driving car.
If you ignore navigating, communicating with other drivers and reading road signs. Since you cannot really drive without these, an ape is not going to be useful as a driver. Selfdriving car has some chance.
And if by AI you mean a program or machine with volition, that actually understands and thinks and has a will of its own, that will likely never happen
What do you base this on? Are you suggesting nature has some special sauce we'll never figure out?
"Figuring out" and replicating with a machine are not the same thing. Yeah for sure nature has a special sauce.
'Having worked on it for a billion years' is a special sauce, but...
...we have AI that trained AI to play Go better than any AI before. We have neural nets that can mostly identify objects. And it took us only fifty years of it vs the hundreds of millions it took for nature to do it.
No, I indeed meant "figure out" as in "replicate." So, what's nature's special sauce? I was genuinely curious as to whether you might be referring to some scientific principle I was unaware of. Given enough time, it doesn't seem likely humans wouldn't be able to replicate what nature has done with respect to the human brain (we've done it before countless times in other areas, after all). But sure, it's likely to take a while!
"It is certainly much easier to train an ape to drive than to create a self-driving car."
For certain subcategories of ape, sure. But even the human apes take 16+ years of feeding, sheltering, nurturing, and educating before you can get it to the point of being easier to train to drive a car than a computer. And you have to do that for every single human driver.
But once you figure out how to build a car that drives itself, building a second car that drives itself costs much less, and each successive unit thereafter is even cheaper.
(also: there are only ~110,000 orangutans in the world; if the golf cart trade group is to be believed, that many golf carts are sold every year in the US alone, so even if you successfully dragooned the entire global population of orangutans into the golf cart biz, the numbers just won't work out).
You wish! We are far from "below replacement level" unless you only look at a selection of rich countries like Japan or Italy. And even most of those grow beyond replacement level due to immigration from places where the population grows even faster.
...A few? All the purple countries are below replacement.
https://worldpopulationreview.com/country-rankings/total-fertility-rate
Yes my last job was secure because it relied on a variety of dexterous movements and I was much cheaper than a robot at doing. It would take a decade to make the money back if I was replaced by a robot.
It's this principle that makes migrant field workers basically irreplaceable. Low, LOW wage, high skill.
Thank you for recognizing that it takes skill!
Hey, I went to a talk by Gabe Thompson at an Industrial Hygiene symposium years back, and read his book "Working in the Shadows", in which he, a middle class, college-educated white man, tries to do the jobs that immigrants do:
* picking lettuce in Arizona
* delivering food orders and flowers in NYC
* processing chicken in Alabama
It was really enlightening ... not to mention depressing.He barely survived, and his co-workers were impressed that he even tried and held out as long as he did.
We live in an age of specialization. I'm a middle class college-educated white man. I'm getting on in years and am not in great physical shape. There are a wide variety of jobs that I did in my youth, like picking fruit or pulling lumber off the green chain, that I can't do any more. There are also MANY jobs that I have never been able to do; I can't drive an 18 wheeler, or fly an airplane, or do any but the very most basic carpentry or plumbing.
I could still deliver pizza, but the job doesn't pay well. Lots of my clients tell me that they couldn't do my job, and I respond in kind.
At current prices.Technology tends to get cheaper and cheaper as it becomes mature. Look at computers, solar panels, wind turbines, etc. etc. Also, if the robot lasts significantly longer than ten years is might be a good investment even now.
The 21st Century is the century where it all comes crashing down because we will run out of everything. We're like the E. coli colony in the test tube that flourishes up until the moment the food run out. We're smart enough to see it coming but not smart enough to take it seriously.
There is every reason to think humanity will find a way to destroy itself. We seem happy to enable the likes of trump, putin, and all manner of crazy religious fanatics. I’m thinking a catastrophic nuclear war in 2050 would be awesome.
No loss really. I’ll be dead by then anyway… meh… tomorrow is ok too.
I'm not normally a doomsayer but the op-ed on AI in this morning's L.A. Times has me thinking nuclear war in 2050 might be the optimistic case for humankind.
The barrier to entry for rogue states and terrorists acquiring WMD has fallen dramatically with the latest in commonly available machine learning software:
In less than six hours, a commercially available artificial intelligence software typically used by drug researchers to discover new kinds of medications was able to come up with 40,000 toxic compounds instead. Many of these substances are previously unknown to science and possibly far more deadly than anything we humans have created on our own.
https://www.latimes.com/opinion/story/2022-03-30/ai-artificial-intelligence-chemical-weapons
AI? I believe it when I see it. Right now we ought to fix climate change. If we fail AI won't matter in the least--even if--improbably--AI lives up to Kevin's hype.
AI is just a buzzword for software. We've only automated a tiny fraction of what we can automate with current technology. It's just a tremendous amount of work to do it.
Even if no further advances were made in "AI" or algorithms, software improvements would still render the world unrecognizable in a couple of decades.
So....no 3 day work week???
Using computers and other electronics more will hopefully increase productivity, but mechanization, using fossil fuels, increased productivity enormously since the 18th century and it did not lead to "mass unemployment", although there was considerable displacement in certain areas. Most people were farmers through the 19th century whereas now only a few percent are, but somehow jobs were found for all the former farmers.
I don't know why supposedly intelligent people assume that further mechanization will lead to unemployment rather than more stuff for everybody as well as more leisure time (to consume all the stuff), but this is definitely not what has happened in history. There are many other potential dangers such as overpopulation and global warming that are more likely to lead to disruption than too much efficiency.
Sure, "more stuff for everybody" but the folks out of work won't be able to buy it. Soon, that will be nearly everybody, under the "Bright AI/Robotics Future" that many predict.
No work, no money, and here in the US of A, we don't believe in subsidizing layabouts. Those fully mechanized factories and self-driving trucks, railroads and ships will be hauling away American-fabricated goods to the rest of the world.
Well stated; strongly agree.
Mechanization created new employment for farmers and those in horse-related occupations in the 20th century because the explosion of new products resulted in an explosion of manufacturing. As the decades passed, ALL workers benefited because, A- more and more of these new products had to be manufactured to meet the demand of a growing population, and B- almost all of this manufacturing took place in this country. Even ‘labor saving’ equipment had to be manufactured here, so the result was a net plus for workers. The process of transitioning from a mechanized economy to a digitized economy has demonstrated that this benefit becomes a negative for workers when the labor saving tech can be created once by a software engineer (rather than manufactured one at a time by factory workers), and most of the ‘new jobs’ (like, say, smart phone hardware manufacturing), can be easily shipped off to low wage countries. This process will get worse as more and more jobs held by people who don’t have advanced STEM degrees from top universities are eliminated. Self driving cars and trucks, for example, will devastate the entire auto manufacturing, repairing, selling and (particularly) driving ecosystem, which, aside from law enforcement, represents probably the largest pool of full-time employment opportunity for non college educated Americans - particularly American men. The profound dislocation felt by millions of people has so far only resulted in one term of Trump and a failed insurrection. Unless the political system figures out a way to deal with this honestly and effectively, we will not be so lucky next year or the year after that. We can’t and shouldn’t slow innovation and disruption, but we also can’t assume that capitalism will somehow magically take care of average people like it did more than a century ago. The last 30 years have shown us that it doesn’t work like that, and if we think economic inequality and its sociological consequences are bad now, we have much worse to look forward to.
Thirty-five years ago, Robert Solow said “You can see the computer age everywhere but in the productivity statistics.” https://www.brookings.edu/articles/the-solow-productivity-paradox-what-do-computers-do-to-productivity/ Since then, productivity growth has actually been slower than in earlier decades. Today, unemployment is under 4%. Manufacturing output in the US is higher in absolute terms, though it is a smaller fraction of GDP (you can look all this up). So, no, it hasn’t been demonstrated that “transitioning from a mechanized economy to a digitized economy becomes a negative for workers
Oops, quotation contains an elision: “transitioning from a mechanized economy to a digitized economy … becomes a negative for workers”.
That's because you're still looking at mechanization, not automation.
Automation is why there are fewer bankers today than thirty years ago, but mechanization is why there's fewer tellers and check clerks.
I don't know why supposedly intelligent people assume that further mechanization will lead to unemployment...
I'm skeptical* of the mass unemployment thesis myself. But those who advance the idea note that the difference between "mechanization" up to now and what's coming in the future (AI) is that, unlike the former (mostly a substitute for the physical weakness and frailties of humans and animals) the latter will replace our brains. If AI really does get sufficiently perfected—all the creativity, commonsense, understanding and intelligence of the smartest humans with unfathomably more processing speed and no need to sleep or take a break—there simply won't be a need to ever employee actual humans, at least not in numbers anywhere near enough to provide income to the bulk of the population.
*Kevin and those with similar views seem to think this mass unemployment situation is very near. I'm skeptical in the sense I doubt this timeline is likely. But the basic idea behind the perfection of AI doesn't seem off to me. So, five years from now, no. But sixty? Eighty? I could see it coming to pass.
First Tyler Cowen, then Larry Summers, and now Brad DeLong. Looks like Kevin went dumpster diving and came up with a trifecta of nincompoops today.
Jus because they're wrong some of the time doesn't mean they aren't important to read. We read Kevin so we don't have to read the nincompoops.
Sure, Kevin reports on their latest tomfoolery, but:
IOW, I don't need Kevin to read them so I don't have to; I'm quite sure I already know what they're going to say before they say it. I'll concede that I'm only right nineteen out of twenty times. OTOH, is it really worth investing any time and energy going after that last five percent? I think not.
This is a challenging time. On the one hand, I hate to see anyone suffer. On the other hand, humanity is so fucking stupid it deserves to suffer.
https://www.nytimes.com/live/2022/03/29/world/ukraine-russia-war#the-chief-of-the-un-food-agency-warns-of-a-crisis-not-seen-since-world-war-ii
“Countries like Egypt and Lebanon are over 80 percent reliant on Ukrainian grain, Mr. Beasley said, adding that the war’s impact on agriculture was no longer just skyrocketing food prices, but also a potential availability problem if countries were not able to offset the losses from Russia and Ukraine. If the conflict does not end, the world risks famine, destabilization of countries and mass migrations, Mr. Beasley said, adding that the agency had already been forced to cut food rations by half and possibly to zero in places like Yemen.”
What can you even say about this? Globalization supply chains are doing just f8ne? No worries! 😂
In this case, however, it's a supply chain that goes back millennia, not some newfangled neoliberal scheme. The region between the Don and Dnieper rivers has supplied Central Asia and the Mediterranean with grain since antiquity. The Black Sea-Mediterranean trade routes, led by rat-infested grain ships, carried the Black Death to the Middle East and Europe, in fact.
Russian TV Host Calls For ‘Regime Change’ in US – and ‘Our Partner’ Trump to Be Put in Power,
https://www.mediaite.com/news/russian-tv-host-calls-for-regime-change-in-us-and-our-partner-trump-to-be-put-in-power/
"Mass unemployment will prompt revolts that make the Luddites look like monks and will likely kill off liberal democracy."
Explain it to me like I'm a 6 year old. What does it look like?
Did you see what damage just a tiny, single-digit minority of truckers and nuts did in Canada? Think that, plus they decided to not heed the police. Or the police joined in.
Yeah. They screwed up traffic in a city of 1 million people for a couple of weeks, and a border crossing for a couplr of days, brcause the authorities (for political reasons) decided to treat them with kid gloves.
Contrast the damage done by the truckers in Washington.
When it becomes clear that AI and robotics will solve the climate crisis, I'll be convinced. Technological progress has always come at a price, and the reckoning for the Industrial Revolution in already under way.
Kevin assumes a lot of things will remain as is in the 21st century.
THAT would be wrong
I agree that the 19th century was a great techno-time
It's the 20th century that changed the world. In spite of two world wars and minor and major skirmishes the population of the world exploded both here, and abroad
Now we are seeing population declines. This will change everything. Demand will slacken and anyone in any supply business will then have to weigh the up front costs of AI versus declining sales due to lack of people.
Already Apple is seeing and planning for less demand of some cell phones. This is due to everyone having one already and fewer and fewer people becoming old enough to use one.
Housing is starting to slow due to cost increases and interest rates rising. As the boomers die off second homes THEY own will come on the market putting further pressure on pricing.
The old measurements are now meaningless.
Why do we need NEW roads? (We don't). Sure replace existing bridges but new ones? Hardly needed. We are moving away from NEEDING things to just "wanting" to spend $. All for convenience.
Less people = less demand. We will NEED to adjust to this over the next 2 decades. All of our economic modelling is based on boomer birth numbers and are now useless
"It's the 20th century that changed the world."
Really things changed much much more in the 19 century than the 20th.
Tend to agree with this. Railroads, steamships, telegraphy, transatlantic cables increased the speed of transmission of information and the movement of goods by an order of magnitude, or more.
Roads wear out, even if you don't use them.
But where is population crashing? Russia.
It's been shown that educating girls and women reduces the birthrate, because women want / need fewer children. The advances in medical technology and infant survival rates help further this.
But look at the increasing use of IVF. Researchers are finding links between diminished human fertility and the widespread use of plastics in everything, permeating the environment, getting into the food chain and even water. (Let's not even talk about the rest of the creatures of the ecosystem, poor things!)
We may be controlling our own growth, albeit unconsciously, by our products. Kind of like the mold colony poisoning its substrate.
Just not women, same for men.
THIS!
Eventually we will have a UBI in order to placate the masses. Won't be enough work and even the poors get angry after a certain point. And there are a LOT of them.
Capitalism is debt+technology. It worked better than expected, but the digital revolution doesn't have the raw investment to service the debt. That was what 2008 told us.
The future is debt liquidation and destruction of faux "Western Living standards" which are partly myth due to how they were atained. When that goes, the tribal impulses of society's factions will be ugly and war will be waged like in corded ware times.
I really didn't need that last paragraph to start my day.
https://www.thedigradio.com/podcast/china-boom-w-ho-fung-hung/
https://www.thedigradio.com/podcast/clash-of-empires-w-ho-fung-hung/
KD: “having barely mastered basic science and economics …”
We mastered economics? I’m ecstatic! Wait, why is my IRA down so far?
Barely.
My 18-year-old kids already say they won't want children of their own because of climate change and the rest. Kevin concentrates on technology and capital. I think hope is more important, and I can't think of a time when there was less of it.
Yes.
If they try to make an artificial intelligence that thinks like a human being, we are most certainly doomed.
Starting to think I should start skipping this blog, KD is just repeating his greatest hits over and over these days.
Poor optimistic Kevin.
We will not cope. Not only will representative government not survive, nor will scarcity economics be replaced with something better.
We are witnessing the operation of the Great Filter, the answer to the so-called Fermi Paradox. The reason why we see no evidence of intelligence anywhere in our galaxy: every intelligent species that grows powerful enough that they might make themselves visible over interstellar distances, or travel such distances themselves, destroys itself somehow before it can do so.