Will robots really boost productivity all that much?

That is the topic of my latest Bloomberg column, here is one bit:

…consider the general logic of labor substitution. Machines and software are often very good at “making stuff” and, increasingly, at delivering well-defined services, such as when Alexa arranges a package for you. But machines are not effective at persuading, at developing advertising campaigns, at branding products or corporations, or at greeting you at the door in a charming manner, as is done so often in restaurants, even if you order on an iPad. Those activities will remain the province of human beings for a long time to come.

How much is this shift of labor into marketing a step forward? To be sure, a lot of commercial persuasion is useful. Marketing informs consumers about new products and their properties, or convinces them that one product is better for them than another. It was marketing that got me to stop watching baseball and switch to the more exciting NBA. Sometimes the very existence of an ad — even apart from any direct informational value — makes a product more enjoyable. If a particular basketball sneaker is associated with LeBron James, through an endorsement and TV commercials, some people will enjoy wearing that sneaker more.

That all said, a lot of marketing is a zero- or negative-sum game. Each business tries to pull customers away from the other brands, and while the final matching of customers to products is usually closely attuned to what people want, more is spent on these business battles than is ideal for social efficiency. My bank might make me feel better about being a customer there, but its services just aren’t much superior to those of the nearest competitor, if at all. Maybe Coke really is better than Pepsi, or vice versa, but it’s not that much better — and billions are spent trying to persuade consumers to make one switch or the other.

I don’t take the Galbraithian view, but still consumers only enjoy extra marketing so much.  I conclude with this:

Don’t be surprised if you see a lot of robots in daily life, and in news stories, but not huge productivity gains in the published statistics. That’s exactly the American economy right now.

Do read the whole thing.


Coke is definitely NOT a close substitute for Pepsi. Diet Coke, my primary beverage, is delicious, Diet Pepsi is way too sweet. Your point is well taken on your bank.

Except it probably isn't. Anyone who works or studies at GMU can be a member of Apple Federal Credit Union, whose terms and conditions and service (a branch on-campus is nice) are likely to be significantly better than any bank's.

Oddly, credit unions rarely are discussed when talking about banking - possibly because credit unions don't bring in much advertising revenue for the media compared to banks.

Quite agreed that they are under-discussed. My credit union provides every service that a bank does, either for free or at about half the price.

Anyone can join. I have no idea how the banks around here stay in business.

Well, they stay in business because of the marketing Tyler is talking about here. Marketing and inertia.

The railroad ties. They're made from wood. The Amish, they are cool.

The tassles, when the corn grows its arrayed head, their khaki colored. The tassles on your shoes that are tied in pocket of hair, their fine. Their from winesburg ohio.

Great article. There's a lot of triumphalism from those who are woke about automation, they think it will be a great thing* benefiting the "knowledge workers" at the expense of the uneducated and the liberal-arts educated.(same as the uneducated lol.) In fact, in an automated economy it will be the hucksters who are most protected, those with the ability to schmooze the boss or convince the normies to pay 65$ for something that costs 10$ to make, because "I'm too high class to shop at Wal-Mart" or "it costs more so it must be of better quality."

*They of course feign concern about the displaced workers for signalling reasons.


That all seems pretty circular, and ignores that Amazon and Walmart are pretty much the same thing. Perhaps the amusing thing is that some people would never buy a $5 usb cable at Walmart, and some people would never order a $5 usb cable from Amazon.

(Me? Walmart is a bit of a drive, so if I don't really need it now, Amazon.)

"We’re unlikely to see mass unemployment; rather, workers will shift into new economic sectors -- albeit with transition pains -- as has always been the case... Some functions of medical assistants are being automated, but hospitals and doctors are still trying to improve the patient experience and reach new customers."

What hasn't always been the case is having a large percentage of newly unemployed go through a very difficult and long transition process.

Wait until almost all medical jobs, doctors and nurses, are automated by 2027. That will look much different than just some medical assistant jobs currently being automated.

Never happen. At least not for decades. AMA way too strong, as proven by the state of health care in the US

And the blacksmith unions were way too strong in 1905...

The AMA will have no power once this goes into high gear by the early 2020s. The demand to see an accurate computer like Watson III for very little will far outstrip demand for doctors who make a lot of mistakes.

I can almost guarantee you we won't have the much vaunted self-driving car in any mainstream form by the 2020s let alone automated surgeons.

"In the past decade the use of robots in surgery has become commonplace. Da Vinci, an American-made surgical robot that is used to repair heart valves, among other things, has operated on more than three million patients around the world. ...The market for medical robotic systems will exceed $17 billion by 2020, according to some estimates."


Hey idiot
"Using a joystick and a camera feed, MacLaren guided the arm of the Robotic Retinal Dissection Device, or R2D2 for short, through a tiny incision in the eye, before lifting the wrinkled membrane, no more than a hundredth of a millimeter thick, from the retina, and reversing Beaver’s vision problems. "

So there was a surgeon present to guide the "robot" as in the "robot" is basically a tool and not autonomous and AI-based.

The subject of the thread is robots boosting productivity. You guaranteed no automated surgery by the 2020s. I pointed out that robots are already involved in millions of surgeries and, yes, they are starting to do fully automated surgery as well.

It's late. You might want to call it a night.

I said automated robots won't be replacing surgeons by the 2020s and you sent me an article that did not in any way contradict that. It's only 9 AM.

Plus I said "almost guaranteed" anything can happen but I would not bet on doctors loosing their jobs any time soon as much as envious people like Todd K who probably didn't have the IQ for medicine would like to see it happen.


Better than humans, but not faster yet...

"When it came to the quality of suturing, the robot beat the human surgeon, with more consistent stitching and less mistakes made. But in terms of the amount of time the surgery takes to perform, the machine came in second place. In the test involving live subjects, STAR took 35 minutes at its fastest, whereas the human surgeon only needed 8 minutes."

Image recognition and pattern recognition via machine learning already has them lobbying here in Texas to prevent "unsupervised diagnostic medical practice". Imagine what would happen if the phlebotomist who draws your blood and sends it off to the lab could when the results are known engage in some disintermediation and give you a far more accurate interpretation of the results than what you'd get from the average hematologist. Very unhappy blood docs. They're all worried. Notice how many hopeful clicks this recent article in JAMA has already received: http://jamanetwork.com/journals/jama/fullarticle/2645762?resultClick=1


Automaton has only really affected the lower/lower middle class, and will continue to do so, due to higher class political power. The cuckservatives will occasionally complain about it, but not in the way they practically scream whenever lower like class folks want government protection.

+1 the idea that doctors are going to be replaced by AI is just a fantasy of the resentful low IQ cucks around here who couldn't make it into med school.

JAMRC is correct upthread. The Da Vinci technology is controlled by a surgeon, and merely augments his capabilities-when it doesn't cause problems itself:



The magnitude of magical thinking regarding technology on this blog by both hosts and commenters is inconceivable.

15 years ago we were all supposed to be replaced by cheap overseas labor.

Then they discovered outsourced "accountants" couldn't spot fraud if I walked the front door with a billion dollars.

I bet that with the robot the eye surgeon does more and more complex operations in a week than he/she did before the robot helped..

That is what we are talking about with productivity.not surgery.

It may mean that nationwide we will not need as many eye surgeons doing operations.

Here is a job where a combination of machines and low wage workers has been reducing wages since the 1990s: language translation, especially for Japanese > English translation were the languages are so far apart grammatically.

In the 1990s, a good Japanese to English translator could make $200,000 in today's dollars. That dropped to closer to $100,000 to $150,000 in the 2000s and now closer to $80,000 as machine translation kept improving, especially from 2006 when Google Translate was introduced.

This steep decline in wages would not have been possible without far better MT being an important assistant to lower end/lower wage translators usually in India and at times China. J/E translators complain more and more every three years or so despite them usually insisting machine translation is irrelevant. They just don't see how important it has been to attract low wage translators which pushed down wages.

Medicine is a very different thing than just about every other service, literally life and death. I'm not saying tech won't change the nature of medicine or even reduce employment at the margins. But to say "almost all medical jobs, doctors and nurses, are automated by 2027" is laughably untrue.

I realize medicine is about life and death, which is why we want the automation in place as soon as the error rates are lower than for doctors and nurses. (And knowing you humans, the computers will probably have to be quite a bit better.) The doctors and nurses still around in 2027 would be paid less than today as overseas doctors asking lower wages will be part of the medical system.

What will a doctor be able to do better than a 2027 computer?

Interact with "us humans" for one thing. The ones paying for it.

"you humans"? I knew you were a robot.

Computers of 2017 already do better than doctors...


"an AI framework easily outperforms the current treatment-as-usual (TAU) case-rate/fee-for-service models of healthcare. The cost per unit of outcome change (CPUC) was $189 vs. $497 for AI vs. TAU (where lower is considered optimal) - while at the same time the AI approach could obtain a 30-35% increase in patient outcomes. Tweaking certain AI model parameters could further enhance this advantage, obtaining approximately 50% more improvement (outcome change) for roughly half the costs."

Yes, they will have to deal with depression, divorce, insolvency, and a sudden and shocking loss of social capital. And that is just for those able to avoid opiods and alcohol. Then there is the problem of finding a place to live, if you don't own your house outright. There aren't enough spots down by the river and not enough bridges to sleep under. We need more infrastructure.

“We’re unlikely to see mass unemployment; rather, workers will shift into new economic sectors — albeit with transition pains — as has always been the case"

At this point that prediction looks good, for certain values of "mass unemployment." Of course, the future is unknown and new definitions of "mass" may be reached.

I would say that right here, right now, we should look at better systems for unemployment and welfare. If mobility is a problem, how about lump-sum unemployment as a grub stake for a new venture, and not monthly payments tying the displaced in place?

"... persuading, at developing advertising campaigns, at branding products or corporations, or at greeting you at the door in a charming manner ..."

The world would be better with less of all of these things.

Marketers, supervisors, engineers, and an intelectual elite of scientists/scholars

While I have sympathy with the gist of this article, it is not clear it is correct. In all market (non-government) sectors, economic activity comprises both production and sales activities, with marketing one component of sales. Since all goods that are produced have to be sold, there is a Baumol cost disease effect in play: if you improve productivity in production activities more than in sales activities, labour will be diverted to sales activities (and the relative price of sales activities will increase); but if productivity of sales improves faster than the productivity of production, the reverse will happen. Since the mid 1980s it is possible that information technology has improved the productivity of sales activities faster than the productivity of production activities, which pushes labour towards production activities. This is certainly true for the US retail sector - it has had faster than average productivity growth over this period.

When sales activities are more labour intensive than production activities, if productivity increases faster in production than sales activities, the labour share will rise: this is true whether, within each activity, labour is a complement or a substitute for capital. Conversely, if productivity increases faster in sales than production activities, the labour share of output will fall (again, whether labour is a complement or substitute for capital within each separate activity.) This is one explanation why the labour share is declining: automation is improving sales productivity very quickly, reducing the ability of the labour-intensive sales sector to absorb workers. In this case, it is not the case that people increasingly become marketers, but people will increasingly get a smaller share of the economic pie.

The "personal touch" that appears to be desired in the United States can come across as very creepy in other cultures. If a worker in a restaurant I haven't met before appears to be happy to see me, then they are either insane, or lying, because they have no reason to be glad to see me. When I went to the United States I felt as though every restaurant was full of nutters.

In societies that value emotional distance and politeness over disgusting human contact, I don't see how humans can compete with machines, outside of niche markets catering to visiting Americans and people whose parents hugged them when they were children. But even there I don't see how humans can compete for long. A machine can sense a human's physiological and emotional state better than any human as they are not limited to human senses. They will have access to a huge and constantly updated data base that will enable them to select optimal responses based on probability, and they have the capability to remember details about individuals and tailor their responses to them beyond that of humans.

I absolutely subscribe. I am Chilean-Italian, and going to a convenience store and being greeted with a "Helloooo sunshine, are you having a WON-derful day?" was downright creepy. The first reaction was to look behind me to see if there was somebody else, the second asking "Sorry: do I know you?". My British friends have similar reactions, with the twist that they sometimes find north americans so polite that they have the impression that they are being mocked, so that impression is consistent across cultures with different approaches to formality and contact.

Send them to my chain brick and mortar drive-thru: Jimmies Guns and Liquours. We have some tables outside where we serve chilidogs and such, and we ain't too polite. We are really happy to take your money, and we don't smile much, we got a robot for that.

I think you're mistaken about AI and marketing. Ad companies are embracing AI in a big way. One really interesting company is Persado:

"The idea of creating marketing messages that take the human guesswork and bias out of the equation is particularly appealing to B2C companies, such as American Express, Verizon, Staples and Caesars Entertainment — just a few of Persado's current customers. Plus, it can do it in 23 different languages. The company says that more than 100 global brands use its cognitive content platform and that they've seen a 68 percent improvement in click-through rates and $1 billion in incremental revenues."


People are quite bad at assessing themselves and others. We're governed by bias and lack of knowledge. Low blood sugar too:

" The plaintiffs were asking either to be allowed out on parole or to have the conditions of their incarceration changed. The team found that, at the start of the day, the judges granted around two-thirds of the applications before them. As the hours passed, that number fell sharply (see chart), eventually reaching zero. But clemency returned after each of two daily breaks, during which the judges retired for food. "


We need robot juries - OJ wouldn't be getting out .

Low blood sugar effect is in dispute


"Don’t be surprised if you see a lot of robots in daily life, and in news stories, but not huge productivity gains in the published statistics. That’s exactly the American economy right now."
In other words, America's economy has become a Potemkin village.

"It was marketing that got me to stop watching baseball and switch to the more exciting NBA."
Maybe he should try real football, but no robot wil tell him that.

I'm really surprised I don't see more calls for a Pigouvian tax on advertising. As far as I can tell, there aren't any negatives:

- consumers aren't bombarded with as much advertising
- consumers receive the benefits of whatever the government spends the windfall on
- advertisers don't lose: the vast majority of advertising spending is relative. If Coke spends 2X as much as Pepsi on advertising it doesn't really matter what X is.
- advertising agencies obviously do lose, but that's the point of a Pigouvian tax, so that can hardly be called a negative

The main outcry will be from the third party industries advertising supports: the TV shows, newspapers, blogs et cetera whose primary income source is advertising. But again I believe this is a good thing: market economies work best in a 2 party transaction. Producer sells to consumer, so consumer happiness is the producer's prime objective. But as the saying goes "if you're not paying for it, you're the product". Netflix and HBO are thriving and producing superior TV without advertising even though their competitors are subsidized with cheap advertising dollars.

"It was marketing that got me to stop watching baseball and switch to the more exciting NBA."

What's exciting about watching giants shoot free throws and officials count seconds?

GS/CAVS vs 2016 Chicago Cubs in AS game 7??!! Surely you jest , Tyler.

LOL, the Janis japlon line in E.L. Doctorow's high flung novel about harry Houdini and 1960, yes 1960, the novel without a sinch of novelty, despite the coal train meanderings, was about upflung heels and wild german dancing, swans galloping and evelyn Nesbit. The age of a handful of dust has been covered by garment high sprung of flavors and epaulettes, ricketty garments along the Golden Gate Bridge and tinctures full of esphogaus wax that varnish the statue of liberty.

Trump is a brand not a product: Donald Trump licenses his name to developers who build hotels and condominiums with his brand on them, he doesn't actually build them. In a nation obsessed with celebrity, it was inevitable that a celebrity, a brand, would be elected president. Google and Facebook offer "free" services, but their revenues are derived from advertising; indeed, those two companies capture between 60% and 70% of total revenues from digital advertising (which explains why newspapers and magazines have struggled to survive). That's why I refer to the boy wonders in Silicon Valley as today's Mad Men - maybe they should be called Silly Men (Mad is short for Madison, as in Madison Avenue where the Mad Men worked). On the other hand, yesterday the president and Terry Gou, who controls Foxconn (the company that makes i-phones among other electronics), announced that Foxconn would build a factory in Wisconsin (in Paul Ryan's district) that will employ 3,000 people to make flat screen televisions. As an incentive to locate in Paul Ryan's district, the state is providing $3 billion in tax breaks "and other subsidies" to Foxconn, or $1 million in tax breaks per employee. The president took full credit for this wonderful deal - there's a reason why this company is called Fox conn. Wouldn't it make more sense to employ a few Silly Men (or Mad Men) who would create the advertising for the flat screen televisions (wherever they are made) without any tax breaks "and other subsidies"?


Well, I've seen that Trump directly employs 22,000. Seems like a lot for just branding stuff.

When your table is ready at the Rainforest Cafe the hostess calls your name and announces "You're adventure begins now!" Evidently marketing has been able to transform the consumption of an over-priced hamburger into something akin to the ascent of Annapurna.

Except Annapurna isn't in the rainforest

To make it real they should have mosquitos and leeches. The elite just love to brag about their adventures. For more realism, no bathrooms. Pinch a loaf squattin behind a bush while swatting squeeters and wipe with a leaf. Watch out for the snakes. They DO kill.

Marketing as zero sum game is described as "mature markets". For example, people in Belgium is not going to buy more beer. Marketing only helps to make people choose another beer. In contrast, in developing markets per capita beer drinking is low, thus marketing presents beer to non consumers.

The biggest problem with marketing is that nobody really knows what works (beyond signaling) though many are convinced they do. There's a reason why several of the now famous papers on improper causal inference and the perils of null hypothesis testing appeared in marketing journals. It's because ad firms were among the first to embrace NHST as a way to "prove" to their customers that their radio and TV jingles worked; and they were among the first to recognize, in their professional organizations anyway, that they really didn't. But, 9 out of 10 dentists (who surveys suggest are even more confused about statistical significance and confidence intervals than Lunch Pail Larry) will buy any claim backed by NHST so they kept doing it. Until yesterday. Most people are rational and it turns out that when they read on Facebook that almost all laundry detergent comes from the same unit in a petroleum refinery and that they're paying a steep markup for the penny's worth of lavender scent, surfactant and enzyme that distinguishes SuperBrand from store brand, they switch to No-Name-O ... if they can. Now Proctor and Gamble is in real trouble.

Social signaling is about the only thing that really works. Mom didn't want to run into her friend in the checkout line with No-Name-O detergent in her basket while her friend splurged on the gloriously orange super brand. That signaled hubby wasn't bringing home much bacon. Now that we have the ability to fill our basket online and at a scheduled time pick it up pre-bagged and ready to be loaded my wife is all about store brands. She still buys expensive purses and likely for the same reason Tyler is furiously twerking his woke status in his LeBron's - there's a little chicken in all of us and we pay attention to the pecking order (at least when the other chickens are looking).


"... most people are rational..."

False, demonstrably false...

I eagerly await your Q.E.D.

"The biggest problem with marketing is that nobody really knows what works (beyond signaling) though many are convinced they do."

Agreed, if you're talking about TV and other mass broadcast marketing. Most online marketing is relatively easy to measure. It's impossible to predict what exactly will work in advance, but it's not hard to figure out what works after testing a lot of different ads.

"Don’t be surprised if you see a lot of robots in daily life".
I see robots all the time. The ATM that you mentioned in the first paragraph is a robot. Just a special purpose robot that doesn't happen to look like the one in the picture that heads your article. The same could be said for all the other robots from washing machines, to microwave ovens, to the flat-ish rectangular slabs that beep at us from our pockets. Then, there is industry where factories have been full of capable robots for over 20-30 years in my field, just getting smarter and more integrated with every generation. And there again the robots don't look like humans and they never will. Thats not the effective form factor for many of their tasks.
The robot generation is already here and it has been here for a good portion of most of our lifetimes.

Robot != machine

"Robot != machine"

If you say so, but even if that's the case, who cares? What's the significant difference between a packaging machine that employs rollers versus one with a robotic arm?

If our ancestors had had such a bleak view of human potential, our human world would never have been built. If they had an inkling about the unworthiness of their descendants, they may not have bothered.

If you've got such a high view of human potential, why rush to replace them with robots?


Gettin old sucks.

Beats the alternative though.

Speaking of your impending doom, NR human trial results from the U of Colorado were discussed last week at a conference but for some reason no information leaked. Both ChromaDex and Elysium have known the results of their 140 and 120 person trial for weeks and months respectively, but they aren't talking yet. ChromaDex said they will release their results sometime this summer.

"We see robots everywhere but in the productivity statistics"


I work in the automation field. For me the article is a classic example of the Murray Gell-Mann Amnesia effect.

Reading comprehension isn't your thing, huh?

_1, jagged edge=kierkegaard

"Briefly stated, the Gell-Mann Amnesia effect is as follows. You open the newspaper to an article on some subject you know well. In Murray's case, physics. In mine, show business. You read the article and see the journalist has absolutely no understanding of either the facts or the issues."

You provide nothing other than your alleged occupation, and you are bereft of any facts to support your denial of reality:

"There is no evidence of an upsurge in automation in the last 10 to 15 years that has affected overall joblessness. The evidence indicates automation has slowed."


Repeat: The magnitude of magical thinking regarding technology on this blog by both hosts and commenters, of which you are a sterling example, is inconceivable.

"more is spent on these business battles than is ideal for social efficiency."

Tyler comes out of the closet as a Bernie bro and marches merrily along to the COMINTERN diktat that "You don't necessarily need a choice of 23 underarm spray deodorants or of 18 different pairs of sneakers when children are hungry in this country" canard.


Interesting article. I like the idea that more jobs will shift from technical work to marketing work. But the focus on advertising and whether it's good or bad for society seems a bit off topic. I think your focus is really on marketing in more of an abstract sense - workers increasing the percentage of time they spend doing some form of persuasion or influencing that humans are uniquely good at and machines cannot easily be trained to do. This type of work in theory could be more meaningful and rewarding than technical work, though I suppose it depends on the person. But either way, that transition will happen independent of how much brands spend on advertising.

This would imply that the spam filter is one of the greatest productivity-enhancing AIs available.

Here is my fear - "high touch" service/marketing jobs are more susceptible to discrimination (racism/sexism/sexual orientationism/creedism/nationalism).

No one really cares who makes your stuff in a factory - you never see them. Perhaps there is a bit of nationalism if you know where something is made, but most people value price far more than nationalism.

On the other hand, people who are discriminatory may have more of a visceral reaction to direct "high touch" of a marketer. For instance, a born-again Christian goes into a bank to be sold financial products by a cross-dresser.

Could this lead to a "black bank" and a "white bank"? A "gay bank" and a "Christian bank"?

Visiting the US in the 1830s, when the 'Bank of the United States' controversy was raging, the British actress Fanny Kemble saw a hat store with a sign proudly proclaiming the store to be 'an anti-Bank hat store'.


I don’t find robot good at all, as I feel machines (software) are not really going to work on most situation. Being a Forex trading, I always give priority to manual trading and I find it a lot easier with broker like OctaFX. They are awesome with having incredibly low spreads at 0.1 pips for all major pairs, zero balance protection and then there is even daily market forecast analysis too, so that really helps me with working nicely and allows me to make profits.

Turing Test fail.

viable sex robots are what I'm waiting for!!


Comments for this post are closed