AI al­ready in the realm of re­al­ity

Kalpesh Mehta, part­ner, Deloitte In­dia, speaks about the tran­si­tion AI is bring to the fi­nan­cial ser­vices in­dus­try:

Banking Frontiers - - Highlights - Mo­han@bank­ingfron­tiers.com

Kalpesh Mehta, part­ner, Deloitte In­dia, speaks about the tran­si­tion AI is bring to the fi­nan­cial ser­vices in­dus­try

Ar­ti­fi­cial in­tel­li­gence is no longer ar­ti­fi­cial, but in the realm of re­al­ity and an in­te­gral el­e­ment of our daily lives, says Kalpesh Mehta, Part­ner, Deloitte In­dia. “A blend of ma­chine learning, cog­ni­tive com­put­ing and nat­u­ral lan­guage pro­cess­ing, AI has a huge rel­e­vance in bank­ing and fi­nan­cial ser­vices and the applications can range from im­prov­ing work­force pro­duc­tiv­ity, en­hanc­ing the cus­tomer ex­pe­ri­ence, driv­ing in­no­va­tion in fi­nan­cial prod­ucts, launch of new busi­ness mod­els and en­sur­ing reg­u­la­tory com­pli­ance,” says he.

He ex­plains that the range of applications of AI in bank­ing range from anti-money laun­der­ing pat­tern de­tec­tion, chat­bots in man­ag­ing cus­tomer re­la­tion­ships, al­go­rithm-based trad­ing for high fre­quency trades, fraud de­tec­tion and driv­ing cus­tomer rec­om­men­da­tions based on cus­tomer pref­er­ences and past fi­nan­cial his­tory. And In­dian banks have started us­ing AI - for ex­am­ple the use of chat bots for ad­vanced cus­tomer ser­vices or use of nat­u­ral lan­guage pro­cess­ing to help cus­tomers with fi­nan­cial and non-fi­nan­cial trans­ac­tions or get in touch with the bank ex­ec­u­tives for prod­ucts. The other uses are stream­lin­ing sys­tem­atic and data fo­cused bank­ing oper­a­tions for ex­am­ple in risk man­age­ment and man­ag­ing delin­quen­cies, he adds.

CRE­AT­ING NEW MOD­ELS

To­day, says Mehta, dig­i­ti­za­tion of pay­ments and us­ing AI with pay­ments data is lead­ing to cre­ation of new forms of trans­ac­tion­based, close-looped lend­ing mod­els. And this can lead to pen­e­tra­tion of credit to the MSMEs and other un­der banked seg­ments with lit­tle or no ac­cess to credit. “The use of wear­ables for health mon­i­tor­ing is lead­ing to new forms of in­sur­ance prod­ucts where the cus­tomer’s be­hav­ior to main­tain good health is rewarded with low­er­ing of health in­sur­ance pre­mi­ums. In the case of per­sonal finance man­age­ment, use of AI and robo ad­vi­sors is lead­ing to a better man­age­ment of an in­di­vid­ual’s per­sonal fi­nances and al­lows fi­nan­cial in­sti­tu­tions to rec­om­mend fi­nan­cial prod­ucts based on an in­di­vid­ual’s fi­nan­cial goals and be­hav­ior,” he elab­o­rates.

Mehta also be­lieves that AI is def­i­nitely be­yond chat­bots and con­ven­tional in­ter­faces. “It’s true that in­vest­ing in the right AI tech­nol­ogy can have a ma­jor im­pact on a bank’s op­er­a­tional ef­fi­ciency, and that its suc­cess boils down to the cus­tomer im­pact above all else, and like any tech­no­log­i­cal in­no­va­tion, the best re­sults will be re­al­ized only if they are im­prov­ing the end user’s ex­pe­ri­ence. That is a bare min­i­mum, but AI will go be­yond and lead to a bank or a fi­nan­cial in­sti­tu­tion cre­at­ing new fi­nan­cial prod­ucts,” he says.

DE­CI­SION MAK­ING POS­SI­BLE

He cites the in­stance of banks ex­plor­ing the use of dig­i­tal pay­ments data or cus­tomer’s GST trans­ac­tion data to make lend­ing de­ci­sions. This will lead to a better pen­e­tra­tion of credit in the coun­try es­pe­cially into mi­cro and small en­ter­prises. “The main­stream uses of AI. Hence, will be in ar­eas of new prod­uct devel­op­ment, pric­ing of fi­nan­cial and in­sur­ance prod­ucts, work­force pro­duc­tiv­ity, fraud man­age­ment, credit risk man­age­ment, cus­tomer ser­vices and man­ag­ing delin­quen­cies,” he adds.

Ac­cord­ing to him, banks in In­dia are adopt­ing AI in a va­ri­ety of forms – chat­bots for cus­tomer ser­vice, in­no­va­tion fac­to­ries for devel­op­ment of AI tech­nol­ogy and prod­ucts that im­prove cus­tomer en­gage­ment, oper­a­tions au­toma­tion, hackathons for in­no­va­tive prod­uct ideas and so­lu­tions through use of AI etc. “The fu­ture as we see will be that AI will go main­stream across the bank­ing oper­a­tions and in ar­eas such as new prod­uct devel­op­ment, pric­ing of fi­nan­cial and in­sur­ance prod­ucts, work­force pro­duc­tiv­ity, fraud man­age­ment, credit risk man­age­ment, cus­tomer ser­vices and man­ag­ing delin­quen­cies,” he pre­dicts.

Mehta main­tains that use of AI can def­i­nitely lead to ‘cog­ni­fi­ca­tion’, when AI is used to make spe­cific de­ci­sions - for ex­am­ple in ar­eas like mi­cro-lend­ing where the re­liance on non-tra­di­tional sources of data such as pay­ment trans­ac­tions, tax trans­ac­tions is higher. “In such cases, due to the high vol­ume of trans­ac­tion data to be an­a­lyzed, cog­ni­fi­ca­tion leads to a sys­tem an­a­lyz­ing and mak­ing a credit de­ci­sion over a hu­man. With growth in dig­i­tal pay­ments vol­umes and GST re­lated tax trans­ac­tion vol­umes, cog­ni­fi­ca­tion is the way for­ward

in the lend­ing process to make credit de­ci­sions. It is my be­lief that going for­ward with in­creased cog­ni­fi­ca­tion, AI will be used in bank­ing and fi­nan­cial ser­vices to make spe­cific de­ci­sions,” says he.

BROAD US­AGE PAT­TERNS

He out­lines the use of AI in the bank­ing and fi­nan­cial ser­vices sec­tor as:

With dig­i­ti­za­tion of pay­ments, there will be a trans­ac­tional trail for every dig­i­tal trans­ac­tion. Money laun­der­ers use in­no­va­tive and cre­ative means to mask their ac­tions mak­ing the il­le­gal earned money look le­gal. AI helps de­tect pat­terns across trans­ac­tions that are oth­er­wise dif­fi­cult to de­tect. Banks will shift to AI based sys­tems that are in­tel­li­gent in de­tect­ing un­der­ly­ing pat­terns across trans­ac­tions and en­sur­ing AML com­pli­ance.

Trad­ing sys­tems de­ploy so­phis­ti­cated al­go­rithms and mod­els that re­quire in­puts from wide range of data sources (fi­nan­cial mar­kets, com­pany data­bases etc.). With an in­creased trad­ing based on so­phis­ti­cated al­go­rithms, the trad­ing de­ci­sions are au­to­mated, and sys­tem based.

With the surge in vol­umes of on­line trans­ac­tions and a focus on in­creased con­ve­nience to cus­tomers, banks are us­ing AI to in­crease the cus­tomer ex­pe­ri­ence, an early im­prove­ment in fraud de­tec­tion and keep­ing the false pos­i­tives lower.

There can be 3 forms of cus­tomer en­gage­ment that can ben­e­fit from an in­creased use of AI. Firstly, in use of chat­bots where the hu­man chat can be re­placed by a sys­tem yet be­ing able to de­liver greater per­son­al­iza­tion and cus­tomer re­la­tion­ship man­age­ment. Se­condly, in the use of vir­tual ad­vi­sors, ie prod­uct rec­om­men­da­tion en­gines that use the cus­tomers’ fi­nan­cial goals, fi­nan­cial be­hav­ior, fi­nan­cial data and per­sonal data to rec­om­mend fi­nan­cial prod­ucts. Thirdly, in tak­ing fi­nan­cial de­ci­sions to pro­vide a cus­tomer with an instant ex­pe­ri­ence, for ex­am­ple, in lend­ing de­ci­sions.

AI used in risk man­age­ment can shift the focus of risk man­agers to a proac­tive risk man­age­ment role. AI based op­er­a­tional risk plat­forms eval­u­ate the cus­tomer data from a wide va­ri­ety of sources, iden­ti­fy­ing the com­plex pat­terns within the data lead­ing to an ac­cu­rate risk pre­dic­tion. The risk mod­els learn from every new data set and im­prove the pre­dic­tion ac­cu­racy over time. A bank’s risk func­tions will see ap­pli­ca­tion of AI in ar­eas such as credit cards, SME lend­ing, un­der­writ­ing de­ci­sions etc.

AI will help reg­u­la­tory com­pli­ance teams of banks to man­age key com­pli­ances such as KYC, AML, rogue per­son de­tec­tion, trade mon­i­tor­ing, trans­ac­tion mon­i­tor­ing etc. The fact that dig­i­ti­za­tion of bank­ing sys­tems will lead to a surge in the vol­ume of data and hence there will be an in­creased reg­u­la­tory focus on the dig­i­tal trans­ac­tions.

WORKER EF­FEC­TIVE­NESS

He also fore­sees that process applications will in­cor­po­rate AI into an or­ga­ni­za­tion’s work­flow to ei­ther au­to­mate pro­cesses or im­prove them by aug­ment­ing worker ef­fec­tive­ness. “Au­to­mated voice re­sponse sys­tems have been used for some years now to re­place hu­man cus­tomer ser­vice agents for first-tier cus­tomer sup­port. The Hong Kong sub­way sys­tem em­ploys AI to au­to­mate and op­ti­mize the plan­ning of work­ers’ en­gi­neer­ing ac­tiv­i­ties, build­ing on the learning of ex­perts. In­sight applications har­ness ad­vanced an­a­lyt­i­cal ca­pa­bil­i­ties to un­cover in­sights that can in­form op­er­a­tional and strategic de­ci­sions across an or­ga­ni­za­tion,” he says.

Mehta out­lines the ex­ten­sive ben­e­fits of us­ing AI as: “Chang­ing cus­tomer ex­pec­ta­tions in an in­creas­ingly dig­i­tized world are dis­rupt­ing the in­dus­try at large. For an in­dus­try rooted in tra­di­tional or­tho­dox­ies, new dis­rup­tive forces are threat­en­ing long-stand­ing en­try bar­ri­ers. Those who sur­vive and grow will rein­vent them­selves and take faster steps to­ward transformation. Or­ga­ni­za­tions can re­al­ize costs sav­ings through the ef­fec­tive use of AI. Other po­ten­tial ben­e­fits - from im­proved flex­i­bil­ity to higher em­ployee morale - can ex­tend the value of cog­ni­tive au­toma­tion.”

MA­JOR AD­VAN­TAGES

Some of the per­ceiv­able ad­van­tages, ac­cord­ing to him are:

• De­creased cy­cle times and im­proved through­put: per­form tasks faster mak­ing 24x7 oper­a­tions pos­si­ble

• Flex­i­bil­ity and scal­a­bil­ity: once a process has been de­fined as a se­ries of in­struc­tions it can be sched­uled for a par­tic­u­lar time and de­ployed to per­form it.

• Im­proved ac­cu­racy: pro­grammed to fol­low rules and do not make ty­pos.

• Im­proved em­ployee morale: The tasks and pro­cesses most suit­able for au­toma­tion are typ­i­cally the most oner­ous and least en­joyed, if hu­mans are re­lieved of these tasks they can focus on more im­por­tant and rewarding work.

• De­tailed data cap­ture: The tasks per­formed can be mon­i­tored and recorded at every step, pro­duc­ing valu­able data and an au­dit trail that can sup­port fur­ther process im­prove­ment and help with reg­u­la­tory com­pli­ance.

Fi­nally, does he think AI will lead to re­place­ment of hu­mans in cer­tain oper­a­tions?

“I don’t see that hap­pen­ing. If any­thing, AI will aug­ment the hu­man role in bank­ing oper­a­tions. It will free up the hu­man time spent on mun­dane or time-con­sum­ing tasks and re­lease hu­mans to focus on more pro­duc­tive oper­a­tions like man­ag­ing cus­tomer re­la­tion­ships which re­quire a more emo­tional and per­sonal touch.”

Kalpesh Mehta strongly be­lieves AI will go be­yond the cur­rent realm of use and will fa­cil­i­tate cre­ation of new fi­nan­cial prod­ucts

IRA 2.0, an ad­vanced hu­manoid from HDFC Bank

Newspapers in English

Newspapers from India

© PressReader. All rights reserved.