dis­rup­tion in the mak­ing

Lever­age the po­ten­tial of the AI rev­o­lu­tion, says Kal­pana B, KPMG in In­dia.

The Smart Manager - - Contents - KAL­PANA B IS PART­NER AND HEAD OF RO­BOT­ICS AND COG­NI­TIVE AU­TO­MA­TION AT KPMG IN IN­DIA

Ar­ti­fi­cial In­tel­li­gence (AI) is an idea that has os­cil­lated through many hype cy­cles over many years, as sci­en­tists and sci-fi vi­sion­ar­ies have de­clared the im­mi­nent ar­rival of think­ing ma­chines. But it seems we’re now at an ac­tual tip­ping point. AI, ex­pert sys­tems, and busi­ness in­tel­li­gence have been with us for decades, but this time the re­al­ity al­most matches the rhetoric, driven by the ex­po­nen­tial growth in tech­nol­ogy ca­pa­bil­i­ties (e.g., Moore’s Law), smarter an­a­lyt­ics en­gines, and the surge in data*. A look at how AI has made tremen­dous in­roads into the realm of busi­nesses as well as our ev­ery­day lives.

Busi­nesses these days have leapfrogged by har­ness­ing tech­nol­ogy, and the way they are op­er­at­ing in the present dy­namic and com­pet­i­tive en­vi­ron­ment is stranger than fic­tion. Cus­tomer ex­pe­ri­ence is im­prov­ing as mar­ket­ing tech­niques are get­ting in­creas­ingly sci­en­tific. Soft­ware tools were en­ablers un­til in­no­va­tors cre­ated tech­nol­ogy plat­forms (in a re­cent dig­i­tal episode) that not only tar­get the ‘how’ of the im­prove­ments but also the ‘who’. Ro­botic process au­to­ma­tion em­ploys soft­ware ro­bots that ex­e­cute repet­i­tive and rule-based pro­cesses, while ad­vanced tech­nol­ogy ded­i­cated to dig­i­tal la­bor beck­ons in­tel­li­gent ro­bots to take up a por­tion of judg­ment-based de­ci­sion-mak­ing. The ad­vanced side of this dig­i­tal la­bor con­sists of an ar­ray of tech­nolo­gies such as Ar­ti­fi­cial In­tel­li­gence (AI), ma­chine learn­ing, nat­u­ral lan­guage pro­cess­ing, and so on. We will lay em­pha­sis on the smarter as­pect of the dig­i­tal la­bor spec­trum and on what the fu­ture holds for us.

Moore’s Law sum­ma­rizes the ob­ser­va­tion made by In­tel co-founder Gor­don Moore in the year 1965 that the pro­cess­ing power of com­put­ers will dou­ble ev­ery two years. The Fourth In­dus­trial Rev­o­lu­tion by Klaus Schwab de­lib­er­ates that if this holds true in the fu­ture too, com­put­ers will at­tain the same level of pro­cess­ing power as hu­mans in 2025.

the con­cept

Let us first at­tempt to dis­en­tan­gle the in­tri­ca­cies of tech­nol­ogy and un­der­stand its wider ap­pli­ca­bil­ity in shap­ing the fu­ture of busi­nesses.

AI is the sim­u­la­tion of hu­man in­tel­li­gence by com­put­ers in dif­fer­ent ways such as vi­sion, speech, and de­ci­sion-mak­ing abil­i­ties; it en­com­passes self-learn­ing, rea­son­ing, and self-op­ti­miza­tion. Ma­chine learn­ing is a branch of AI that en­ables com­put­ers to learn with­out be­ing ex­plic­itly pro­grammed and choose an ef­fec­tive course based on the learn­ings and ex­pe­ri­ence.

To un­der­stand this phe­nom­e­non bet­ter, let us as­sume that an academy deal­ing with on­line cour­ses is pi­lot­ing a dig­i­tal mar­ket­ing cam­paign in dif­fer­ent parts of a coun­try, for in­struc­tor-led and self-learn­ing mod­ules. Dur­ing the course of the cam­paign, if the ma­chine learn­ing pro­gram ob­serves that the self-learn­ing course mod­ule is not pre­ferred in a par­tic­u­lar re­gion, then au­to­mat­i­cally the mar­ket­ing at­ten­tion will shift to in­struc­tor-led cour­ses. In the fu­ture, the pro­gram will ex­tend this knowl­edge to other re­gions shar­ing sim­i­lar de­mo­graph­ics too, for ef­fec­tive cam­paigns.

Ma­chine vi­sion is the abil­ity of a com­puter to ‘see’ just like hu­mans, while speech recog­ni­tion is the abil­ity of com­put­ers to ‘hear’ and un­der­stand dic­ta­tion. Through vi­sion abil­i­ties, the ap­pli­ca­tion can be ex­tended to un­der­stand videos, im­ages, and pic­tures. This opens a huge win­dow of op­por­tu­nity to make sense of vast amounts of data present in the for­mat of im­ages and videos. Speech recog­ni­tion abil­ity al­lows in­ter­pre­ta­tion of hu­man speech and spo­ken lan­guages, thereby us­ing the knowl­edge base to take busi­ness de­ci­sions. An­other tech­nol­ogy that has ex­ten­sive ap­pli­ca­tion is nat­u­ral lan­guage pro­cess­ing (NLP), which can un­der­stand and gen­er­ate lan­guages that hu­mans use nat­u­rally with com­put­ers.

what has hap­pened so far?

Bank­ing, in­sur­ance, re­tail, IT, and tele­com have been early adopters of tech­nol­ogy. Since tech­nol­ogy has been the medium and ba­sis of ex­e­cut­ing trans­ac­tions, for long, data cap­tured over the years acts as a nat­u­ral foun­da­tion to adopt smarter tech­nolo­gies. We can ex­pect these sec­tors to adopt smarter tech­nol­ogy sooner than oth­ers.

A few growth driv­ers of Ar­ti­fi­cial In­tel­li­gence are in­creas­ing num­ber of in­ter­net and smart­phone users, in­creas­ing traf­fic in on­line busi­ness, ad­vance­ments in big data an­a­lyt­ics, cloud com­put­ing, pro­lif­er­a­tion of busi­ness apps, talent, and fierce com­pe­ti­tion. AI has en­tered our day-to-day lives, as busi­ness, per­sonal, and so­cial worlds have shrunk within a de­vice for users. Whether they are au­to­mated chats about in­ter­act­ing with cus­tomers to un­der­stand their prob­lems or in­tel­li­gent seg­re­ga­tion of pic­tures based on weather, ap­pear­ance, and per­son­al­ity traits, AI has nu­mer­ous man­i­fes­ta­tions. Other re­cent ap­pli­ca­tions of AI that one might have ex­pe­ri­enced are speech-to-text con­ver­sion, face recog­ni­tion, age pre­dic­tion, retina recog­ni­tion, and fin­ger­print se­cu­rity scan­ners.

As in­di­vid­u­als ac­cess be­spoke ap­pli­ca­tions, they cur­so­rily per­mit per­sonal data to be sent to a cloud, leav­ing be­hind a dig­i­tal trail. The mag­ni­tude of per­sonal and busi­ness-rel­e­vant data be­ing gen­er­ated is hu­mon­gous and cor­po­rates are us­ing it wisely to tar­get the var­i­ous user seg­ments. Through a shift from data elit­ists to am­a­teurs, ma­chine learn­ing is be­com­ing ac­ces­si­ble to pro­fes­sion­als as it is be­ing pre­sented by in­no­va­tors in a more us­able, as-a-ser­vice pack­age. With the mush­room­ing of star­tups, the idea of mak­ing it a prac­ti­cal so­lu­tion to ac­tual prob­lems is be­com­ing even more re­al­is­tic.

Ma­chine vi­sion is the abil­ity of a com­puter to ‘see’ just like hu­mans, while speech recog­ni­tion is the abil­ity of com­put­ers to ‘hear’ and un­der­stand dic­ta­tion.

ap­pli­ca­tions

Vi­tal and time-in­ten­sive ar­eas such as on­board­ing new can­di­dates are be­ing of­fered through AI-pow­ered re­cruit­ing and hir­ing so­lu­tions. Fields such as telemedicine are be­ing trans­formed by AI so­lu­tions for im­proved di­ag­no­sis and low­ered costs. Cer­tain ap­proaches are fo­cus­ing on en­rich­ing the on­line shop­ping ex­pe­ri­ence through the con­ve­nience of chat­ting on an in­te­grated chat­bot plat­form. Im­age recog­ni­tion tech­nol­ogy finds huge ap­pli­ca­bil­ity in e-re­tail for refin­ing on­line pur­chase choices. An­a­lyz­ing data has been pre­dom­i­nantly done by hu­mans over the years, but in­tel­li­gent plat­forms today are not only an­a­lyz­ing struc­tured data but also gen­er­at­ing re­ports that com­mu­ni­cate in spo­ken English text.

AI plat­forms also help B2C sec­tors such as tele­com, bank­ing, and re­tail in dig­i­tal mar­ket­ing cam­paigns by gen­er­at­ing cus­tomer-spe­cific con­tent through­out the cus­tomer life cy­cle—by an­a­lyz­ing the de­mo­graph­ics and cus­tomer touch­point data. An­other po­tent area of ap­pli­ca­tion is con­tact cen­ter au­to­ma­tion where cog­ni­tive so­lu­tions au­to­mate in­bound and out­bound calls. The ar­eas we dis­cussed are sketchy re­flec­tions of the count­less ap­pli­ca­tions of AI, and this tech­nol­ogy is in­creas­ingly mak­ing a ma­te­rial dif­fer­ence across the op­er­at­ing mod­els of many sec­tors today.

Or­ga­ni­za­tions will have to re­shape them­selves by mak­ing the most of the over­whelm­ing de­vel­op­ment in tech­nol­ogy, to rise above the noise and pro­vide true value to the con­sumer and so­ci­ety at large.

ob­so­les­cence

We no longer use land­line tele­phones, tele­graphs, hand­held cal­cu­la­tors, pagers, floppy drives, and CDs. The next gen­er­a­tion may not even be ac­quainted with these terms. Tech­nol­ogy is chang­ing fast and we have to keep pace with it to avoid be­ing ob­so­lete. As busi­nesses have evolved over past decades, so­ci­ety has also shown keen in­ter­est in adopt­ing newer tech­nol­ogy. Or­ga­ni­za­tions that were slow to rec­og­nize this fact be­came out­dated and fi­nally strug­gled to ex­ist.

We are sur­rounded by AI—whether it is chore­ographed searches or au­to­matic ad­ver­tise­ment feed, we have ex­pe­ri­enced AI in dif­fer­ent sizes and pro­por­tions. Moore’s Law is not just a the­ory that sug­gests the speed of ad­vance­ments, but a thrilling in­di­ca­tion that even so­ci­ety needs to progress in a syn­chro­nized way. Com­pe­ti­tion is no longer with just other in­cum­bents or con­tes­tants in the mar­ket, but with one­self too. Or­ga­ni­za­tions will have to re­shape them­selves by mak­ing the most of the over­whelm­ing de­vel­op­ment in tech­nol­ogy, to rise above the noise and pro­vide true value to the con­sumer and so­ci­ety at large.

what could pos­si­bly help us in the jour­ney

■ ■ ■ ■ ■ ■ ■ ■ ■ ■ Choose the right cap­tur­ing mech­a­nisms [Study] the rel­e­vance and re­li­a­bil­ity of data

Stay data-hun­gry—the ac­cu­racy of these sys­tems im­prove with the va­ri­ety and size of data sets Im­bibe cur­rent learn­ings

In­ter­lock safety with the so­lu­tions—em­bed con­trol mech­a­nisms to re­duce the im­pact of fail­ure Hire the right talent mix and build a mo­ti­vated team Se­lect the right part­ners

Stay up­dated—keep your ear to the ground Be in­no­va­tive—link­ing data sets and ex­tract­ing un­ap­par­ent as­so­ci­a­tions be­tween them could prove pro­gres­sive for your busi­ness

Make AI an en­ter­prise-wide pri­or­ity ■

Newspapers in English

Newspapers from India

© PressReader. All rights reserved.