Why Soft­ware Hasn’t Won... Yet

AI is not an im­me­di­ate threat to your jobs, but even IT en­gi­neers will have to ad­just to the rise of ‘soft­ware in­tel­li­gence’

India Today - - AUTOMATION - By RO­HAN N. MURTY

SINCE IT WAS for­mally founded in the early 1940s, com­puter science has made ex­tra­or­di­nary progress and con­trib­uted pos­i­tively to so­ci­ety. It has, in many cases, low­ered the bar­rier for ac­cess to in­for­ma­tion, trans­ac­tions, ed­u­ca­tion, and health­care. It has en­abled us to put man on the moon and per­haps even be­yond. It fu­els the most mod­ern progress in science. It has ren­dered dis­tance im­ma­te­rial. It may even help by driv­ing cars for us, de­liv­er­ing pack­ages in cities and fly­ing air­craft. And, in some cases, it even ap­pears that ‘in­tel­li­gent’ soft­ware sys­tems may en­tirely take over from hu­man be­ings. In fact, the birth, ge­n­e­sis and vi­sion of com­puter science has al­ways been to build a ma­chine that can ‘ac­cu­rately’ em­u­late hu­man ‘in­tel­li­gence’.

De­spite all these strides, com­puter science still has a fair bit of dis­tance left to cover be­fore it can pro­duce a soft­ware sys­tem that is func­tion­ally ca­pa­ble the same way a hu­man is. The un­der­stand­ing of ‘in­tel­li­gence’ in com­puter science, though de­fined in aca­demic cir­cles, has been fairly mis­used (and/or mis­un­der­stood) in pop­u­lar re­port­ing, giv­ing peo­ple the im­pres­sion that soft­ware can do just about any­thing a hu­man be­ing can. This no­tion is fun­da­men­tally mis­guided. We are not there. Or at least not yet. The gap be­tween soft­ware and the hu­man mind is still mas­sive.

Soft­ware still op­er­ates, for the most part, on the par­a­digm of GIGO (Garbage in, Garbage out), mean­ing it mostly only does what­ever we tell it to and is bi­ased on the ba­sis of what­ever data we train it with. Though, in some cases, given large amounts of data and hu­man in­put, we are able to get soft­ware to dis­cover some rules on its own. But at the core of it is the fact that we have not in­vented any new magic wand to make soft­ware ‘think’ on its own or have any sense of ‘con­scious­ness’, in the way hu­mans do. In­stead, the most im­por­tant fun­da­men­tal progress that tech­nol­ogy has made over the past few decades is a dra­matic in­crease in com­put­ing power cou­pled with a re­duc­tion in the cost of stor­age. Such progress has en­abled us to run al­go­rithms and code over swathes of data, and at a frac­tion of the cost, yield­ing what some may con­sider as more in­sights. Con­se­quently, we are able to find pat­terns in large amounts of data. This, by some lim­ited def­i­ni­tion, is con­strued as in­tel­li­gence and has been demon­strated in spe­cific classes of ac­tiv­i­ties, such as de­tect­ing di­a­betes from reti­nal scans, car driv­ing, play­ing games, etc. In these cases, soft­ware can em­u­late hu­man be­hav­iour and per­haps even ex­ceed hu­man ca­pa­bil­i­ties.

These gaps help de­flate claims that ar­ti­fi­cial in­tel­li­gence (AI) and au­to­ma­tion are ready to take over peo­ple’s jobs and po­ten­tially spark mass un­em­ploy­ment, par­tic­u­larly in the In­dian IT in­dus­try. This in­dus­try em­ploys a cou­ple of mil­lion peo­ple, largely com­pris­ing our In­dian mid­dle class, and re­cently has been called out by com­men­ta­tors as un­der threat from au­to­ma­tion.

But au­tomat­ing away this labour force is not an

easy task. Even though, as a com­puter sci­en­tist, I am bred to be­lieve in the su­pe­ri­or­ity of sil­i­con (com­put­ers) over car­bon (hu­mans), my work and study over the past few years in this in­dus­try has given me rea­son to pause and con­sider the unique con­tri­bu­tions and value that hu­mans bring to the ta­ble in en­abling busi­nesses the world over. Au­tomat­ing these unique at­tributes is to em­u­late the same value that hu­mans bring to the ta­ble.

In the case of the In­dian IT in­dus­try, for ex­am­ple, pro­fes­sion­als per­form a range of ac­tiv­i­ties span­ning un­der­stand­ing cus­tomers’ busi­ness re­quire­ments, writ­ing code, main­tain­ing soft­ware sys­tems and pro­cess­ing trans­ac­tions. Work­ing in teams, they per­form a va­ri­ety of ac­tions touch­ing sev­eral in­for­ma­tion sys­tems, co­or­di­nate with each other, nav­i­gate un­cer­tain or chang­ing sce­nar­ios, re­port to their man­agers, take bot­tom-line re­spon­si­bil­i­ties for their re­spec­tive sys­tems, are ac­count­able and serve cus­tomers across the globe. All these ac­tiv­i­ties en­gen­der a deep sense of trust and con­fi­dence among their cus­tomers the world over.

At an in­di­vid­ual level, each has traits such as in­tu­ition, ex­pe­ri­ence, com­mon sense and an abil­ity to iden­tify and fix the un­usual. These are fun­da­men­tal traits that all hu­mans share, though ad­mit­tedly, some more than oth­ers. Even in the most seem­ingly ‘ro­botic’ of tasks, each hu­man has the po­ten­tial and abil­ity to ap­ply these traits with­out nec­es­sar­ily be­ing told to do so a pri­ori, that are in­her­ent to each one of us, and ul­ti­mately cre­ate trust and con­fi­dence in our col­leagues. This trust and con­fi­dence are the fuel all busi­nesses run on. There­fore, it has al­ways been more than just fol­low­ing ‘sim­ple’ rules.

SO ANY AT­TEMPT by ‘in­tel­li­gent’ soft­ware at au­tomat­ing all that work done by a sin­gle in­di­vid­ual or an en­tire team of hu­mans must in­volve recre­at­ing the same trust and con­fi­dence. This nec­es­sar­ily re­quires ad­dress­ing the whole gamut of these ac­tiv­i­ties that hu­mans per­form on a reg­u­lar ba­sis and with sim­i­lar hu­man traits, but in soft­ware. This is no longer about small piece­meal tasks that are easy to au­to­mate. It must in­volve em­u­lat­ing the same traits of com­mon sense, in­tu­ition, team­work, com­mu­ni­ca­tion and an abil­ity to deal with the un­ex­pected.

Con­sider, for ex­am­ple, ques­tions that man­agers could ask their teams to­day. ‘Why did you do that?’, ‘Why did you choose X over Y?’, or ‘Are you sure?’. Or more sim­ply put, ‘ex­plain your­self’. For all the rea­sons ar­tic­u­lated ear­lier, these ques­tions, though straight­for­ward for hu­mans, are very dif­fi­cult for ma­chines to an­swer. There­fore, the next stage of in­no­va­tion in com­puter science will be not just to make soft­ware run faster, but also about how we get soft­ware to em­u­late more hu­man-like traits and ul­ti­mately be­come truly trust­wor­thy (like hu­mans).

All this said, we should not be naive to think that tech­nol­ogy won’t even­tu­ally suc­ceed in recre­at­ing some—if not many—lay­ers of hu­man cog­ni­tion and range of ac­tiv­i­ties. And given this even­tu­al­ity, it is im­por­tant to think about and be­gin plan­ning how In­dia can ex­cel in this new nor­mal. Much of this, it ap­pears, will de­pend on our pre­pared­ness for up­skilling op­por­tu­ni­ties.

The IT in­dus­try is no stranger to up­skilling. In fact, com­pa­nies like In­fosys and a few oth­ers have made un­prece­dented ef­forts in train­ing cam­puses as a way to aug­ment the en­gi­neer­ing ed­u­ca­tion im­parted to the coun­try’s young peo­ple. For ex­am­ple, In­fosys has the ca­pac­ity to train any­where be­tween 40,000-60,000 peo­ple each year at its Mysore cam­pus. Such in­vest­ments in train­ing, per­haps more than any­thing else, give such com­pa­nies an in­sur­ance pol­icy against a fu­ture of au­to­ma­tion. But one may say, isn’t an idea of ‘train­ing’ a copout? In the soft­ware world, learn­ing has been a con­stant fea­ture. Con­sider this: in the 1980s, a FORTRAN/C pro­gram­mer was in vogue, C++/Java in the 1990s, per­haps Python/Go/Ruby to­day, and so on. Hence, con­stantly learn­ing and re­train­ing has been a way of life in this in­dus­try.

But even if the prom­ises of au­to­ma­tion are only half as true as the hype, it sug­gests that just a frac­tion of In­dia’s IT in­dus­try pro­fes­sional s, even with all that com­pany-led train­ing, will be re­quired. Those who are needed will have to learn ar­eas of com­pu­ta­tion that are far more ad­vanced than what the IT in­dus­try presently of­fers. So, do we again look to the IT in­dus­try to lead the up­skill charge, or is a broader pub­lic-pri­vate coali­tion re­quired? In­dia is not alone in this co­nun­drum. Many coun­tries, in­clud­ing the United States, have been grap­pling with the ques­tion of who owns the re­spon­si­bil­ity of up­skilling, es­pe­cially in the face of chal­lenges from off­shoring or au­to­ma­tion. And, as re­cently ev­i­denced, even no­tional at­ti­tudes of a com­mu­nity be­ing over­looked by a chang­ing econ­omy can have sig­nif­i­cant im­pli­ca­tions on po­lit­i­cal out­comes.

De­spite these strug­gles, and those of many other coun­tries, it’s un­clear if a par­a­digm for up­skilling pop­u­la­tions has emerged. Which brings us back to my orig­i­nal point: the re­mark­able ca­pa­bil­ity of the hu­man mind. In an ideal world, this push to­ward au­to­ma­tion would be a cat­a­lyst for up­skilling, which would un­lock re­mark­able hu­man po­ten­tial and usher in a new, more tech­no­log­i­cally ad­vanced econ­omy for In­dia. In the real world, how­ever, this will take a shared mis­sion, care­ful plan­ning and gen­uine co­op­er­a­tion.

THERE IS OVER­WHELM­ING con­sen­sus within the sci­en­tific com­mu­nity, based on in­de­pen­dent ev­i­dence from mul­ti­ple sources, that the on­go­ing global warming is caused by ris­ing lev­els of CO2 in the at­mos­phere. Most alarm­ing is the ac­cel­er­at­ing re­treat of plan­e­tary ice, par­tic­u­larly in the Arc­tic, which hap­pens to be the fastest warming re­gion on the globe. The 3,000 me­tre thick Green­land ice shield has been melt­ing for many years, with two in­de­pen­dent fac­tors in op­er­a­tion ac­cel­er­at­ing the rate of melt: warming of the Arc­tic Ocean due to the steadily di­min­ish­ing seaice cover and the fall in al­ti­tude with de­clin­ing thick­ness of the ice shield. Com­plete melt­ing of the Green­land ice mass will raise global sea lev­els by six me­tres. On­go­ing thaw­ing of the per­mafrost of the cir­cum-Arc­tic tun­dra will re­lease the trapped green­house gases, CO2 and meth­ane, in quan­ti­ties that would greatly ac­cel­er­ate the cur­rent rate of warming.

Heat waves in­creas­ing in in­ten­sity, greater fre­quency of freak weather events and ris­ing sea lev­els are in­evitable out­comes of global warming. Sea level rise is the most dan­ger­ous as it will in­un­date coastal cities, push es­tu­ar­ies up­river and dis­place hun­dreds of mil­lions of peo­ple all over the world. Re­set­tling these cli­mate refugees will put enor­mous strain on the moral and ma­te­rial fab­ric of hu­mane civil­i­sa­tion; I hes­i­tate to dwell on that night­mare and pre­fer to think about what could be done now to avert it.

Sea level rise is a new con­cept in hu­man his­tory be­cause coast­lines have been anoma­lously sta­ble over the last 10,000 years. The sea still laps at the shore of Lothal in Gu­jarat, the bronze-age har­bour con­tem­po­ra­ne­ous with the Harap­pan civil­i­sa­tion of 4,500 years ago. How­ever, in the 5,000-year pe­riod be­tween 15,000 and 10,000 years ago, when the last Ice Age came to an end and the ice shields cov­er­ing north­ern North Amer­ica and parts of Eura­sia melted away, the sea level rose by 130 me­tres! In the same pe­riod, at­mo­spheric CO2 con­cen­tra­tions in­creased from 0.18 to 0.28 per cent and re­mained as anoma­lously sta­ble as the sea level un­til the end of the 1900s. Since then, with the dawn of the In­dus­trial Age, they have steadily in­creased, and in 2016 crossed the 0.40 per cent level. In the space of a cen­tury, hu­mans have added more CO2 to the at­mos­phere than ac­cu­mu­lated there in the 5,000-year pe­riod of

the Big Melt. So it is not sur­pris­ing that we have started the next Big Melt which, if left unchecked, will thaw the re­main­ing ice shields and raise sea lev­els by about 100 me­tres. There are maps on the web that show what this does to the planet’s ge­og­ra­phy.

WHEN CON­FRONTED WITH the in­evitabil­ity of sea level rise, the big ques­tion that arises is: how much time do we have? The an­swer de­pends on what global hu­man­ity does in the com­ing years. If we con­tinue burn­ing fos­sil fu­els and do noth­ing about re­mov­ing CO2 from the at­mos­phere, largescale re­set­tle­ment pro­grammes for peo­ple evac­u­ated from the low-ly­ing fringes of coastal cities and deltas will have to start well be­fore mid-cen­tury, i.e. when to­day’s chil­dren reach the primes of their lives. There is hope that the dead­lines can be post­poned if coun­ter­mea­sures are adopted on a war foot­ing as soon as pos­si­ble. Rat­i­fi­ca­tion of the Paris Ac­cord by so many coun­tries, in­clud­ing In­dia, is no doubt a heart­en­ing mile­stone, but its goals fall short of what is re­quired now. At­tempts by the next US ad­min­is­tra­tion to change course on the cli­mate front will be met by stiff in­ter­nal op­po­si­tion from an ad­min­is­tra­tion com­mit­ted to com­bat­ing cli­mate change. This is where In­dia, which prides it­self on be­ing the old­est con­tin­u­ous cul­ture on the planet, should take up a lead­er­ship role and chart the course for a brighter fu­ture.

I per­ceive four fronts along which de­ci­sive ac­tion in In­dia could be taken now:

de­vel­op­ing a mas­sive cli­mate ed­u­ca­tion cam­paign di­rected at peo­ple from all walks of life that ex­plains the na­ture of the dan­gers to so­ci­ety ly­ing ahead, their causes and con­se­quences, and how these can be averted or at least mit­i­gated;

curb­ing green­house gas emis­sions from burn­ing fos­sil fu­els by switch­ing to re­gen­er­a­tive en­ergy sources, in par­tic­u­lar so­lar and wind;

de­vel­op­ing a port­fo­lio of ap­proaches to re­move CO2 from the at­mos­phere;

ex­plor­ing in­no­va­tive tech­niques to en­sure long-term, sus­tain­able food se­cu­rity. I deal briefly with each of these points in the fol­low­ing.

CLI­MATE ED­U­CA­TION

Un­der­stand­ing a prob­lem is a pre­req­ui­site to solv­ing it in the long term, and since mo­ti­vated pub­lic par­tic­i­pa­tion is cru­cial for the suc­cess of this gar­gan­tuan un­der­tak­ing, a cli­mate ed­u­ca­tion cam­paign, tai­lored to the sit­u­a­tion in In­dia, is the way for­ward. The type of pub­lic ed­u­ca­tion cam­paign I am ad­vo­cat­ing is anal­o­gous to, but on a much grander and more ur­gent scale than, that im­ple­mented in the past decades for dis­ease con­trol and pub­lic health im­prove­ment: de­spite many short­com­ings, their over­all suc­cess is undis­puted. Earth sys­tem sci­en­tists are mak­ing rapid ad­vances in un­der­stand­ing how the cli­mate ma­chin­ery of our planet works. Their pre­dic­tions of the ef­fects of global warming are be­ing con­firmed by un­fold­ing re­al­ity. The sto­ries emerg­ing from the var­i­ous re­search fields, greatly aided by state-of-the-art com­put­ers, are ex­cit­ing and wor­thy of broad pub­lic at­ten­tion as they make sense in un­ex­pected ways. They can be trans­mit­ted to the pub­lic by a com­bi­na­tion of lec­tures, movies and an­i­ma­tions on the web. The me­dia and en­ter­tain­ment in­dus­try could play ma­jor roles in draw­ing at­ten­tion to the prob­lems ahead and how to solve them now. A ma­jor thrust of the cam­paign would be to teach cycli­cal think­ing, for which the hy­dro­log­i­cal and car­bon cy­cles of our planet pro­vide the blue­print for deeper un­der­stand­ing. Cycli­cal think­ing is the philo­soph­i­cal frame­work of sus­tain­abil­ity and goes far be­yond lin­ear think­ing, which is short-term, lim­ited to the space/time scales of the thinker and the cause of our cur­rent prob­lems. It is worth point­ing out here that the con­cept of cy­cling is deeply en­trenched in an­cient In­dian phi­los­o­phy.

CURB­ING EMIS­SIONS

The vast im­prove­ment in pub­lic health and the qual­ity of life has been driven by burn­ing fos­sil fu­els. In­dia has paid a heavy price in terms of ter­ri­ble air qual­ity and its ad­verse ef­fects on the health of peo­ple, par­tic­u­larly chil­dren. Will the ba­bies born to­day have breath­ing prob­lems when they grow up? Is not the dream of phas­ing out

in­ter­nal com­bus­tion en­gines from the cities and re­plac­ing them with elec­tric en­gines within reach? The process of coun­try­wide elec­tri­fi­ca­tion of­fers enor­mous growth po­ten­tial and em­ploy­ment for a work­force rang­ing from un­skilled labour to highly trained ex­perts. The tech­nol­ogy for im­ple­ment­ing the tran­si­tion to de­cen­tralised en­ergy cap­ture is al­ready de­vel­oped, so why wait for the fu­ture? The vi­sion of a so­lar-pow­ered In­dia is the type of de­cen­tralised self-suf­fi­ciency that Gandhi dreamed of and that is now within reach of the en­tire pop­u­la­tion, ren­dered pos­si­ble by mod­ern, smart tech­nol­ogy. There is enor­mous scope for in­no­va­tion in the field of so­lar-gen­er­ated elec­tric­ity and its stor­age, which will lead to cre­ation of new prod­ucts for new mar­kets. The In­dian di­as­pora could be per­suaded to con­trib­ute to this mon­u­men­tal ef­fort.

IN THIS BRIGHT NEW LIGHT, in­vest­ing in nu­clear power plants is re­peat­ing the same mis­take made at the dawn of the fos­sil fuel era—dis­re­gard of the fate and fu­ture im­pact of the waste prod­ucts. Ra­dioac­tiv­ity is a form of en­ergy hu­mans can­not feel, hence eas­ily mis­judge. Ger­many is cur­rently fac­ing the oner­ous task of dis­man­tling its re­ac­tors and dis­pos­ing of their wastes. Apart from the enor­mous costs to be borne by the tax­pay­ers, deciding where to bury them within the coun­try is go­ing to cause strife. In­dia should be writ­ing a new verse for Surya Na­maskar (salu­ta­tion to the sun) rather than bur­den­ing com­ing gen­er­a­tions, labour­ing un­der in­creas­ing pres­sure from global warming, with spent nu­clear re­ac­tors.

RE­MOV­ING CO2

Other than plant­ing trees, re­mov­ing an­thro­pogenic CO2 from the at­mos­phere is an un­pop­u­lar topic in the sci­en­tific com­mu­nity, par­tic­u­larly as it’s be­lieved that re­search in this field will dis­tract from the pri­mary task of cut­ting emis­sions. The anal­ogy that comes to mind is that of a leak­ing ship al­ready list­ing be­cause the of­fi­cers have de­cided not to let the crew start bail­ing out the wa­ter un­til the leak is re­paired. That CO2 re­moval can never be an op­tion for curb­ing emis­sions is il­lus­trated by the mag­ni­tude of the prob­lem. The mass of ex­cess car­bon in the at­mos­phere, as CO2, that would have to be re­moved to re­store ini­tial con­di­tions (the dif­fer­ence be­tween 0.28 and 0.4 per cent) is equiv­a­lent to 250 bil­lion tonnes. For com­par­i­son, the to­tal amount of car­bon present in all the vis­i­ble veg­e­ta­tion present on the con­ti­nents, in­clud­ing crops, grasses and forests, is 500 bil­lion tonnes. Where will the space, wa­ter and nu­tri­ents to in­crease veg­e­ta­tion cover by that much come from? For me, the an­swer lies in the vast deserts of the ocean—the sub­trop­i­cal gyres that cover half the plan­e­tary sur­face but barely con­trib­ute to food sup­ply or car­bon se­ques­tra­tion.

SUS­TAIN­ABLE AGRI­CUL­TURE

Al­ter­na­tive, se­cure sources of food will also need to be de­vel­oped. My ex­per­tise lies in the oceans, and that is where hu­mankind will have to turn to for cre­at­ing ar­ti­fi­cial ecosys­tems by aqua­farm­ing in the ocean’s deserts. These are vast lenses of warm wa­ter about 200 me­tres deep, float­ing on nu­tri­ent-rich cold wa­ter that fills the oceans. The oases would be main­tained by pump­ing nu­tri­ent-rich deep wa­ter to the sur­face layer where, af­ter it has warmed, plank­ton productivity would feed fish and could also be used to grow sea­weeds for con­sump­tion as well as for car­bon se­ques­tra­tion. Local en­ergy sources could ac­com­plish the task. Such an un­der­tak­ing will en­tail enor­mous in­fra­struc­ture of pipes and an­chored islets and will be an en­gi­neer­ing chal­lenge wor­thy of the smart era of tech­nol­ogy that we have now en­tered. The al­ter­na­tive sources of food pro­vided by aqua­farm­ing on the open ocean will not only re­lease pres­sure on the present agri­cul­tural land and en­able de­con­ges­tion of cities, but also pro­vide space for nat­u­ral ecosys­tems to ex­pand and se­quester car­bon. In­stead of dream­ing about space travel, we should be direct­ing our at­ten­tion to the un­used in­ner spa­ces on our planet. Need­less to say, In­dia is well placed to launch such a ven­ture.

Illustration by ANIR­BAN GHOSH

Newspapers in English

Newspapers from India

© PressReader. All rights reserved.