Re­place­able You

Econ­o­mist were skep­ti­cal that au­to­ma­tion could per­ma­nently dis­place hu­man work­ers on a large scale. Looks like they were wrong.

Newsweek International - - CONTENTS - BY ELLEN RUPPEL SHELL

Route 9 skims by bos­ton and cuts clear across Mas­sachusetts to Pitts­field, a city of roughly 50,000, the largest in Berk­shire County. Well east of Pitts­field, Route 9 be­comes Worces­ter Road, named for a city that in ear­lier times was the na­tion’s largest man­u­fac­turer of wire— barbed wire, elec­tri­cal wire, tele­phone wire and the wire used in the mak­ing of un­der­gar­ments by the Royal Worces­ter Corset Co., once the largest em­ployer of women in the United States. Older Worces­ter res­i­dents can still re­call the fac­tory bells peal­ing to sig­nal the start and end of the work­day. Now, the bells are silent, and the wire and corset fac­to­ries have been re­placed with three of the na­tion’s largest em­ploy­ers: Wal­mart, Tar­get and Home De­pot.

If this sounds fa­mil­iar, it should. It has been nearly two decades since re­tail over­took man­u­fac­tur­ing as the na­tion’s most im­por­tant job cre­ator, em­ploy­ing roughly one of ev­ery 10 Amer­i­can work­ers—more peo­ple than in health care and con­struc­tion com­bined. That’s a lot of jobs.

Of course, not all re­tail jobs qual­ify as what most of us con­sider good jobs. To­day, the av­er­age hourly wage for a non­super­vi­sory re­tail worker is $11.24, and less than half of re­tail work­ers re­ceive ben­e­fits of any kind. Still, as a na­tion, we’ve come to a sort of un­easy peace with this trend. We know that man­u­fac­tur­ing em­ploys far fewer Amer­i­cans to­day than it once did—that ipads and Macs aren’t made in Amer­ica and nei­ther are many tele­vi­sions, ap­pli­ances, tools, toys or clothes. We also know that shop­ping for these ap­pli­ances, tools, toys and clothes is an all-amer­i­can pas­time: On av­er­age, we spend nearly 45 min­utes a day (more than 270 hours per year) pur­chas­ing goods and ser­vices. Re­tail has be­come the world as we know it, and many of us ex­pect to make our liv­ing work­ing in that world.

And yet, tra­di­tional re­tail is un­der threat—from the same forces that are dis­rupt­ing vir­tu­ally ev­ery sec­tor of the Amer­i­can econ­omy. Last month’s an­nounce­ment of his­tor­i­cally low un­em­ploy­ment num­bers brought cheers but also con­fu­sion: Given what econ­o­mists called the na­tion’s “full em­ploy­ment,” why did so many of us feel left be­hind? Af­ter all, Amer­i­cans have ac­quired more ed­u­ca­tion and are more pro­duc­tive than ever be­fore. Yet, as it turns out, our feel­ings of be­ing ripped off are jus­ti­fied: More than 80 per­cent of us are not reap­ing the bounty

of our own ed­u­ca­tion and pro­duc­tiv­ity. For while un­em­ploy­ment is tech­ni­cally at a his­toric low point, un­der­em­ploy­ment is ram­pant. Fully 20 per­cent of men aged 24 to 55 do not have full-time jobs, and nearly half of all new col­lege grad­u­ates are un­able to find a job that com­ports with their ed­u­ca­tion. (Con­trary to pop­u­lar think­ing, col­lege stu­dents are not im­prac­ti­cal “bas­ket weav­ing” ma­jors—roughly 40 per­cent earn de­grees in “oc­cu­pa­tional” dis­ci­plines, such as busi­ness, le­gal stud­ies and pub­lic ad­min­is­tra­tion, an 80 per­cent in­crease since 1970.) And while Uber driv­ers and free­lance dog walk­ers tech­ni­cally count as “em­ployed,” they are not em­ployed in the sort of oc­cu­pa­tion that typ­i­cally of­fers a liv­ing wage. The bot­tom line is this: Tech­nol­ogy has ad­vanced at a breath­tak­ing pace, while the pol­icy de­signed to help work­ers deal with these changes has lagged far be­hind. Hence, the fi­nan­cial ben­e­fits of tech­no­log­i­cal change ac­crue mainly to the few, while the ma­jor­ity of Amer­i­cans are left with crumbs— pre­car­i­ous, un­sta­ble em­ploy­ment that re­flects nei­ther their abil­i­ties nor their po­ten­tial.

“We’re at a unique point in hu­man his­tory,” Rice Univer­sity com­puter sci­en­tist Moshe Vardi says. “We are sit­ting on the cusp of an enor­mous change.”

In re­tail, this is the chal­lenge: When it comes to prof­its, no brick-and-mor­tar store—no mat­ter how ef­fi­cient—can hold a can­dle to e-com­merce, which since 2014 has be­come the fastest-grow­ing re­tail sec­tor by far. China’s Alibaba Group—asia’s most valu­able com­pany—is the world’s largest player in this keenly com­pet­i­tive arena. But Alibaba has so far failed to gain a foothold in the United States. In Amer­ica, Ama­zon—the na­tion’s fastest-grow­ing em­ployer—reigns supreme.

An­a­lysts pre­dict that by 2020, one-fifth of the mul­ti­tril­lion-dol­lar U.S. re­tail mar­ket will have shifted to the web and that Ama­zon alone will reap twothirds of that bounty. The com­pany al­ready cap­tures one of ev­ery two dol­lars Amer­i­cans spend on­line and is by far the na­tion’s big­gest seller of books, mu­sic, video games, cell­phones, elec­tron­ics, small ap­pli­ances, toys, mag­a­zine sub­scrip­tions and what seems like al­most ev­ery­thing else—hence its nick­name, “The Ev­ery­thing Store.” The com­pany grabs gobs of mar­ket share in nearly ev­ery re­tail cat­e­gory, in­clud­ing its own food line. It pro­duces TV shows and movies; man­u­fac­tures thou­sands of prod­ucts, from bat­ter­ies to baby food; and owns such fa­mil­iar brands as Zap­pos, Shop­bop, IMDB, Au­di­ble and Twitch. Ama­zon Hand­made is chal­leng­ing Etsy with the ar­ti­sanal set, and Ama­zon Busi­ness is threat­en­ing Sta­ples and other in­de­pen­dent of­fice sup­pli­ers. And with ev­ery click, the com­pany gath­ers crit­i­cal in­for­ma­tion—our ad­dresses and credit his­to­ries, as well as ev­ery­thing we’ve ever bought or even looked at on the Ama­zon site—and uses the data to build an even more in­ti­mate re­la­tion­ship with each of us, with the goal of cul­ti­vat­ing still more of our busi­ness.

Thanks to au­to­ma­tion and a killer busi­ness model, Ama­zon is so ef­fi­cient that it reaps nearly twice the rev­enue per em­ployee of Wal­mart, de­spite the fact that Wal­mart, too, has a sub­stan­tial on­line pres­ence. World­wide, Ama­zon has in­stalled over 100,000 ro­bots to la­bor in “per­fect sym­bio­sis” with hu­mans in its ware­houses and has plans to in­stall many thou­sands more. While it’s not clear what con­sti­tutes per­fect sym­bio­sis, the ro­bots are said to save the com­pany $22 mil­lion an­nu­ally, per ware­house. The com­pany’s mas­ter plan of an au­tonomous fu­ture also in­cludes goods de­liv­ered by drones and self-driv­ing ve­hi­cles.

For while Ama­zon con­tin­ues to open ware­houses around the globe and staff them with many

There are more work­ers in re­tail than in health care and con­struc­tion com­bined, and they are un­der threat from the same forces that are dis­rupt­ing vir­tu­ally ev­ery sec­tor of the Amer­i­can econ­omy.

thou­sands of hu­man be­ings, es­ti­mates are that ev­ery hu­man on the Ama­zon pay­roll—whether ful­lor part-time—dis­places two hu­mans at tra­di­tional brick-and-mor­tar op­er­a­tions. And that’s a fea­ture, not a bug: As Tim Lind­ner, a vet­eran IT an­a­lyst, con­fided in a note to in­dus­try in­sid­ers, erad­i­cat­ing jobs is the ex­plicit goal of any on­line re­tailer. As he once wrote: “La­bor is the high­est-cost fac­tor in ware­house op­er­a­tions. It is no se­cret that Ama­zon is mov­ing to highly au­to­mated op­er­a­tions within its distri­bu­tion cen­ters, and…it has ad­di­tional tech­nol­ogy that can fur­ther re­duce the num­ber of hu­mans it needs to process cus­tomer or­ders.… You have heard the old pro­gram­mer’s phrase, ‘Garbage in, garbage out.’… [With] the di­min­ish­ing read­ing abil­i­ties of hu­mans on the Re­ceiv­ing dock, find­ing an au­to­mated so­lu­tion to elim­i­nate the ‘garbage in’ prob­lem is the holy grail. Ama­zon may have just patented it.”

By garbage, Lind­ner meant hu­man er­ror, the al­ter­na­tive to which is ap­par­ently robotic pre­ci­sion. And ro­bots can be very pre­cise, es­pe­cially when it comes to rou­tine tasks. Sawyer, an in­dus­trial ro­bot cre­ated by the for­mer Bos­ton-based Re­think Robotics, of­fers an im­pres­sive il­lus­tra­tion of how all-em­brac­ing a ro­bot arm can be. Sawyer is the brain­child of Rod­ney Brooks, the in­ven­tor of both Roomba, the robotic vac­uum, and Pack­bot, the ro­bot used to clear bunkers in Iraq and Afghanistan and at the World Trade Cen­ter af­ter 9/11. Un­like Roomba and Pack­bot, Sawyer looks al­most hu­man—it has an an­i­mated flat-screen face and wheels where its legs should be. Sim­ply grab­bing and ad­just­ing its mon­key-like arm and guid­ing it through a se­ries of mo­tions “teaches” Sawyer what­ever re­peat­able pro­ce­dure one needs it to get done. The ro­bot can sense and ma­nip­u­late ob­jects al­most as quickly and as flu­idly as a hu­man and de­mands very lit­tle in re­turn: While tra­di­tional in­dus­trial ro­bots re­quire costly engi­neers and pro­gram­mers to write and de­bug their code, a high school dropout can learn to pro­gram Sawyer in less than five min­utes. Brooks once es­ti­mated that, all told, Sawyer (and his older brother, the two-armed Bax­ter ro­bot) would work for a “wage” equiv­a­lent of less than $4 an hour.

Ro­bots loom large in dis­cus­sions of work and its fu­ture, a con­ver­sa­tion that can get mired in false as­sump­tions. Un­til re­cently, many econ­o­mists were skep­ti­cal that au­to­ma­tion could per­ma­nently dis­place hu­man work­ers on a large scale. Peo­ple have al­ways shifted away from work bet­ter done by ma­chines, but the eco­nomic prin­ci­ple of “com­par­a­tive ad­van­tage” pre­dicts that hu­mans will main­tain an

Ware­houses are be­com­ing au­to­mated, as are any num­ber of other of other places with jobs once filled by what econ­o­mists call “mid­dle-skill work­ers,” the very peo­ple who once pop­u­lated— and bol­stered—the Amer­i­can mid­dle class.

edge in many fields. Un­der this logic, tech­nol­ogy will not dis­place us but set us free to do less dan­ger­ous, more chal­leng­ing things, es­sen­tially the very things that make hu­mans hu­man.

For ex­am­ple, in 2016, the Na­tional High­way Traf­fic Safety Ad­min­is­tra­tion of­fi­cially rec­og­nized “soft­ware” as a driver of self-driv­ing cars, thereby putting the na­tion’s 4.1 mil­lion paid mo­tor ve­hi­cle op­er­a­tors—driv­ers of taxis, trucks, buses and Uber—on no­tice. The­o­ret­i­cally, this will free these driv­ers to fill new roles—such as ones in Ama­zon ware­houses. But these ware­houses are also be­com­ing au­to­mated, as are any num­ber of other places with jobs once filled by the vast ma­jor­ity of what econ­o­mists call “mid­dle-skill work­ers,” the very peo­ple who once pop­u­lated—and bol­stered—the Amer­i­can mid­dle class. Work­ers like the thought­ful depart­ment store sales­man who helped mea­sure you for the suit you wore to your daugh­ter’s wed­ding, the pa­tient butcher who carved out the chops for the pre-wed­ding din­ner or the travel agent who helped plan the hon­ey­moon.

Of course, hu­man work­ers are com­pli­cated. We get tired, hun­gry, dis­tracted, an­gry, con­fused. We make mis­takes, some­times egre­gious ones. Ma­chines lack our frail­ties and bi­ases and are bet­ter equipped to weigh ev­i­dence fairly, with­out prej­u­dice or false as­sump­tions. Per­haps most crit­i­cally, ma­chines can re­tain and process data far more ac­cu­rately than we can, and that data is grow­ing ex­po­nen­tially.

Ev­ery minute of ev­ery day, Google ser­vices 3.6 mil­lion searches in the United States alone. Spam­mers send 100 mil­lion emails. Snapchat­ters send 527,000 pho­tos, and the Weather Chan­nel broad­casts 18 mil­lion fore­casts. This and more data—prop­erly col­lected, cod­i­fied and an­a­lyzed—can be ap­plied to au­to­mate al­most any high-or­der task. Data can also serve as a sur­ro­gate for hu­man ex­pe­ri­ence and in­tu­ition. On­line shop­ping and so­cial me­dia sites “learn” our pref­er­ences and use that in­for­ma­tion to make val­ues-based as­sess­ments to in­flu­ence our de­ci­sions and be­hav­ior. And, in­creas­ingly, ma­chines ex­cel in the tasks once thought uniquely hu­man.

“Com­put­ers are able to see and hear, and have face-recog­ni­tion ca­pa­bil­i­ties that are sig­nif­i­cantly bet­ter than hu­mans,” says Vardi. “Ma­chines un­der­stand the hu­man world far bet­ter than they did just a few years ago. And we haven’t dis­cov­ered any­thing in the hu­man brain that can’t be mod­eled.”

Bart Sel­man is a pro­fes­sor of com­puter sci­ence at Cor­nell Univer­sity and an ex­pert in knowl­edge rep­re­sen­ta­tion—ba­si­cally, trans­lat­ing the real world into terms com­put­ers can un­der­stand and act upon. He cau­tions that com­put­ers do not yet have full hu­man ca­pa­bil­i­ties. For ex­am­ple, they lack “com­mon sense” and an abil­ity to grasp the deep mean­ing of lan­guage. They are un­able to “make mean­ing” in the hu­man sense, and this some­times leads them down the wrong path. Still, he says, these short­com­ings are likely tem­po­rary. “The [ar­ti­fi­cial in­tel­li­gence] com­mu­nity be­lieves that ma­chines will match hu­man in­tel­li­gence within the next 15 to 20 years,” he says.

And ro­bots need not be per­fect, only equal to—or a tad bet­ter than—com­pli­cated and ex­pen­sive hu­mans. And tech­nol­o­gists are work­ing hard to make sure they are a tad bet­ter. For ex­am­ple, in the case of re­tail, it’s be­come clear that many of us avoid the self-ser­vice check­out line—we pre­fer the cashier to punch in our pur­chases rather than do so our­selves. So it seems that the job of cashier—among the

largest re­tail em­ploy­ment cat­e­gories—is not di­rectly at risk. But Zeynep Ton, an MIT man­age­ment ex­pert who fo­cuses on the re­tail sec­tor, says self-ser­vice check­out is only a first step and not a ter­ri­bly smart one. “Cus­tomers rec­og­nized that self-ser­vice check­out is not an in­no­va­tion, but merely a way of out­sourc­ing the job to them, so they didn’t like it,” she says. “But new tech­nol­ogy is com­ing that will make self-ser­vice check­out so much eas­ier and faster, and that will have a real im­pact on re­tail em­ploy­ment.”

Ex­perts cau­tion that the so-called apoca­lypse in re­tail pre­dicted a few years ago has not yet come to pass. In fact, for ev­ery com­pany clos­ing ex­ist­ing stores, two more are open­ing new stores. Re­tail is a highly com­pet­i­tive in­dus­try, and tech­nol­ogy is trans­form­ing not only the way we shop but the way we con­nect with brands—for ex­am­ple, just a few years ago, who would have imag­ined that Ama­zon would open ac­tual re­tail stores? And while e-com­merce has grown to 10 per­cent of re­tail, that still leaves 90 per­cent for brick-and-mor­tar stores. But those brick-and-mor­tar stores, too, are un­der­go­ing rad­i­cal change that has se­ri­ous im­pli­ca­tions for Amer­ica’s work­force.

Kasey Lobaugh is chief re­tail in­no­va­tion of­fi­cer at Deloitte Con­sult­ing LLP. “The loss of mar­ket share by tra­di­tional re­tail com­pa­nies is not sim­ply an on­line vs. bricks-and-mor­tar bat­tle, with tra­di­tional re­tail­ers los­ing the e-com­merce game,” he says. “Tra­di­tional re­tail­ers are also be­ing chal­lenged by what we call an­kle biters,” small, nim­ble com­pa­nies that—thanks to tech­nol­ogy—can reach con­sumers with­out a mas­sive cap­i­tal out­lay.

As ex­am­ple, Lobaugh cites food trucks, which he says in­creas­ingly pose a threat to many fast-food out­lets. Un­like restau­rants pinned down by a pair of Golden Arches, food trucks are nim­ble—they can home in on ar­eas where cus­tomers are most likely to gather at any par­tic­u­lar time. They can also tai­lor their of­fer­ings to a par­tic­u­lar re­gion or even a neigh­bor­hood, as well as use Face­book or other me­dia to get out the word on their menu items and lo­ca­tions. Small, spe­cialty stores also have far more flex­i­bil­ity than large depart­ment stores. “Tech­nol­ogy has re­duced the cost of en­try into new mar­kets, so in re­tail there are fewer big, mono­lithic com­pa­nies, but more small com­peti­tors,” he says. “Com­pa­nies are di­ver­si­fy­ing to meet the spe­cific needs and de­sires of con­sumers—ev­ery­one’s piece is get­ting smaller, but there are many more pieces.”

Tech­nol­ogy has en­gen­dered a two-tiered re­tail land­scape—with more high-end bou­tique-type stores ap­peal­ing mostly to high-wage earn­ers, and many more dis­count stores ap­peal­ing to price-sen­si­tive cus­tomers. “More than 1,000 dis­count stores opened in the U.S. this year alone, as did a large num­ber of what we call ‘premier’ high-end niche stores,” Lobaugh says. What’s de­clin­ing is what mar­keters call the “bal­anced” store—depart­ment stores and other re­tail­ers that bal­ance qual­ity and price for mid-mar­ket cus­tomers. Per­haps it is not sur­pris­ing that the de­cline of the “bal­anced” store cor­re­lated with the de­cline of the Amer­i­can mid­dle class over the past decade. “Be­tween 2007 and 2017, in­come gains—an av­er­age in­crease of $50,000 in house­hold in­come—went mostly to the top 20 per­cent of earn­ers,” Lobaugh ex­plains. “In fact, this group gained more than 100 per­cent of the in­crease, be­cause the bot­tom 40 per­cent ac­tu­ally lost ground. The mid­dle 40 per­cent gained $10,000 per house­hold. But their ex­penses in­creased—food, hous­ing, trans­porta­tion. Health care sky­rock­eted. On top of that came dig­i­tal ne­ces­si­ties—things like cell­phones and data plans. That leaves most peo­ple very lit­tle money to spend on re­tail, which means they have be­come very, very price sen­si­tive.”

Lobaugh prefers not to spec­u­late about what all this meant for the re­tail worker, other than to say the trend was “trans­for­ma­tive.” But what is clear is that dis­count stores em­ploy fewer work­ers per square foot of store space and tend to of­fer low wages and fewer hours: The num­ber of hours per em­ployee has ac­tu­ally dropped over the past decade. John Chal­lenger, CEO of Chal­lenger, Gray & Christ­mas, a Chicago-based global out­place­ment & ca­reer tran­si­tion­ing firm, said he sees more changes ahead. “I think we’re in the open­ing phase of what hap­pened to man­u­fac­tur­ing in the 1980s and 1990s,” he says. “There’s no ques­tion that store work­ers are vul­ner­a­ble to tech­nol­ogy and that un­told num­bers have al­ready been dis­placed.”

Asked where all these re­tail work­ers had gone, he says it was likely many had found new jobs in truck­ing, ship­ping, distri­bu­tion—that is, ware­hous­ing. And af­ter all, in Oc­to­ber, Ama­zon an­nounced its de­ci­sion to raise min­i­mum wage at its ware­houses and re­tail out­lets to $15 an hour, a big jump for many re­tail work­ers.

But de­spite what it pre­dicts will be a ban­ner hol­i­day sea­son, this year Ama­zon took on far fewer sea­sonal em­ploy­ees than usual—100,000 em­ploy­ees ver­sus 120,000 the pre­vi­ous two years. And while an Ama­zon spokes­woman in­sisted that au­to­ma­tion is not a fac­tor in this re­duced work­force, oth­ers seem to not agree. In a re­cent re­port, Mor­gan Stan­ley an­a­lyst Brian Nowak soothed the fears of Ama­zon share­hold­ers con­cerned with the wage in­crease by point­ing out that au­to­ma­tion had al­ready and would con­tinue to re­duce the call for la­bor, and there­fore re­duce over­all costs. When asked about this, Lobaugh again tact­fully de­clined to com­ment—other than to say that while the re­tail sec­tor had lost less ground than most peo­ple as­sume, re­tail em­ploy­ees were an­other mat­ter. “There are win­ners,” he says, “and then there are losers.”

Hod Lip­son, a pro­fes­sor of me­chan­i­cal en­gi­neer­ing at Columbia Univer­sity, di­rects the Cre­ative Ma­chines Lab, where he and his stu­dents train ma­chines to be re­flec­tive, cu­ri­ous and, yes, cre­ative—in­clud­ing in the kitchen. When we spoke, he was putting the fi­nal touches on a de­vice that uses soft­ware to con­coct beau­ti­fully com­posed gourmet de­lights from a jum­ble of pastes, gels, pow­ders and liq­uid in­gre­di­ents. From the looks of it, this ma­chine could com­pete with a three-star Miche­lin chef and her en­tire staff. When I ran this thought by Lip­son, he groaned. He says sci­en­tists and engi­neers like him­self have a re­flex­ive urge to au­to­mate al­most ev­ery dif­fi­cult task. The whole point of en­gi­neer­ing, he says, is to al­le­vi­ate drudgery and in­crease pro­duc­tiv­ity; in the past, that was al­most al­ways the right thing to do, the good thing to do. But now he’s not so sure.

“Au­to­ma­tion and AI will take away pretty much all of our jobs,” he says. “If not within our life­time, then within our grand­chil­dren’s life­time. This is a new sit­u­a­tion in hu­man his­tory, and we’re not pre­pared for it. Maybe we think we are, but we’re not.”

A high school dropout could pro­gram Sawyer in min­utes, and it would work for the equiv­a­lent of un­der $4 an hour.

SAY HELLO TO MY LIT­TLE FRIEND The Ama­zon ful­fill­ment cen­ter in, New Jersey still hires hu­mans—for now. The com­pany cur­rently em­ploys 100,000 ro­bots in its ware­houses world­wide. Op­po­site: Sawyer of Re­think Robotics.

GOOD­BYE TO ALL THAT Clock­wise from top: A Wal­mart cashier; women em­ployed by the Royal Worces­ter Corset Co. in Mas­sachusetts in 1902; New York taxi driv­ers protest­ing re­cent in­roads by the car ser­vice Uber, which, among other things, has launched a ground­break­ing self-driv­ing car.

STUCK IN THE MID­DLE From top: Dog walk­ers may not be at risk from ro­bots and au­to­ma­tion, but “mid­dle-skill” jobs like long-haul truck­ing and ware­house jobs will; bot­tom, ro­bots and vi­sion sys­tems at Ama­zon.

BIG LOSSES, SMALL WINS From top: Even as it opens brick-and-mor­tar stores (the first, here, in Seat­tle in 2015), Ama­zon is hir­ing fewer sea­sonal em­ploy­ees for the hol­i­days this year; Sawyer in ac­tion; food trucks, which are nim­bler and more me­di­asavvy than fast-food chains. “Every­body’s piece is get­ting smaller, but there are more pieces,” Lobaugh says.

Newspapers in English

Newspapers from UK

© PressReader. All rights reserved.