《BCG：變革的架構師-2023生成式AI在IT服務革命中的作用分析報告（英文版）（69頁）.pdf》由會員分享，可在線閱讀，更多相關《BCG：變革的架構師-2023生成式AI在IT服務革命中的作用分析報告（英文版）（69頁）.pdf（69頁珍藏版）》請在三個皮匠報告上搜索。

1、ARCHITECTS OF CHANGE Generative AIs Role in Revolutionizing IT Services2023GenAI-the new tech frontierWhatare buyerssaying?How are IT/ITeS firms responding to the opportunity?Implications for IT/ITeS sector2-2122-4041-4849-65TABLE OF CONTENTSCHAPTER 1CHAPTER 2CHAPTER 3CHAPTER 4Generative AI is expec

2、ted to achieve 30%share of overall AI market by 2025,reaching$60Bn in TAM BFSI,Consumer,Healthcare&Media expected to be top sectors by TAMPast technologies have shown wide variety of adoption&maturity of usage.Accordingly,Multiple PoVs exist on how big of a shift in technology is Generative AI,its a

3、doption by businesses as well as impact on IT/ITeS Services firms in short to medium termFirst view(Scenario A)states that GenAI is a seminal and fundamental shift(like Internet)in technology,getting adopted faster than any technology before and will have a lasting impact on IT Services firms(claims

4、 of 50-80%of revenues at risk EXECUTIVE SUMMARYin a short-medium term in particular service lines)GenAI uses advanced AI model with 175Bn+parameters(vs.10Mn in any before GPT)that leads to training of advanced conceptual relationships and hence relevance for a high number of use cases It transforms

5、the human-computer interface because of advanced natural language capabilities,which makes it highly accessible&usable for end-users Both these factors have caused wide-ranging adoption of GPT-based toolsChatGPT reached 1Mn users in 5 days and 100Mn in just 2 monthsBard also recorded 50Mn visits in

6、March 2023(within a month of launch)Investments in GenAI have already crossed investments in any other technology in its first 2-3 years saw$20Bn of investments over the past 2-3 years Initial pilots have shown 50%improvement in productivity for tasks like Coding and relevance for up to 80%of custom

7、er service queriesThe second view(Scenario B)states that GenAI is a new technology that despite initial spike,will see humbler pace of industry adoption like technologies that came before it(like Cloud,Automation,IOT)EXECUTIVE SUMMARY GenAI needs immense amount of training data to be useful,demands

8、substantial computational costs&power consumption and faces risks like data privacy,hallucinations,copyright infringement that need to be solved before scaled adoption by enterprisesFor instance,Integration of GenAI in search is expected to add significant dollars in Alphabets costs It takes several

9、 decades for new technologies to have widespread implementation across industriesAdoption rate for big data in 2022 still at 55%while 35%for AI/ML,automation and IoT even with decades of tech availabilityE2E Digital transformation implementation still at only 15%of the enterprises Investors have mai

10、ntained similar multiples for IT Service provider stocks,owing to the continued stickiness of revenuesDip in TSR index witnessed post ChatGPT and GPT-4 launch was recovered in a couple of monthsBPM Services returns still down yet the dip has been called unwarranted by leading investorsGenAI is the l

11、atest inflection point-the fourth generation of AI evolution-and is poised for more success than earlier iterations of AI because of lower barriers to entry for use,cross-functional use cases,easier interfacing,context-based learning as well as availability of high computing&storage capacitiesArchit

12、ects of Change:Generative AIs Role in Revolutionizing IT Services5Generative AI is expected to achieve 30%share of overall AI market by 2025,reaching$60Bn TAM222 2112022433 3 322023865465202410119 915720251412151523932212216191220262027YoY growth of GenAI marketGenerative AI%of total AI64%85%59%52%6

13、1%75%BFSI2189358861121ConsumerNote:1.Other includes Industrial Goods,Energy,and Telecom markets;2.BFSI includes Insurance($2Bn 2025)and Financial Institutions($13Bn 2025)including retail and wholesale banking,asset and wealth management,and private equitySources:AI TAM research,Expert interviews,BCG

14、 analysis131%94%91%74%44%38%10%314%21%27%30%31%Generative AI TAM($Bn)CAGR 2022-2025+66%HealthcareMediaPublic SectorOther1Architects of Change:Generative AIs Role in Revolutionizing IT Services6Past technologies have shown variety of adoption&maturity of usagePast technologies have seen wide variety

15、in adoption,even after decades of being aroundMany of these technologies are yet to see large scale implementations Adoption:Share of organizations with on-going pilots&small-and large-scale implementation;Maturity of usage:Share of firms who have adopted with large-scale implementation of technolog

16、y;Inflection:Signifies a pivotal year of transformative change and advancement(Cloud:AWS launched cloud services;Big Data:Adoption of NoSQL;RPA:BluePrism launched commercial RPA offerings;IOT:US NIC recognized IoT as disruptive civil tech;AI/ML Launch of deep learning&neural networks;AR/VR:ARToolKit

17、 released;Blockchain:Bitcoin went live,Quantum Computing:Google launched a quantum computer)Source:Digital Leadership Survey,2022 by Harvey Nash N=1785 includes senior technology decision makers from Jul 20 to Oct 10,2022;BCG analysis Adoption(%)CloudInternetBig data/analyticsAI/MLRPAAR/VRBlockchain

18、QCMetaverse200406080100LowMediumHighMaturity of Usage200102030406080100Adoption(%)CloudInternetBig data/analyticsAI/MLRPAIoTIoTAR/VRBlockchainQuantum computingMetaverseYears since inflectionArchitects of Change:Generative AIs Role in Revolutionizing IT Services7Similarly,Multiple POVs exist on the s

19、cale&speed of GenAI adoption by enterprisesSCENARIO ASeminal and fundamental shift in technology,leverages advanced AI network with 175Bn+parametersImmense data needs increases computations costs and has inherent risks like privacy concerns,hallucinations,etc.Faster adoption than any other technolog

20、y,ChatGPT reached 100Mn subscribers in 2 monthsWidespread technology adoption takes decades(e.g only 15%enterprises have undergone E2E digital transformation)Significant improvement in productivity and core processes with 50%productivity gains in coding60%enterprises expecting substantial to high im

21、pact on their businesses only in 2-5 years timelineLasting impact on IT/ITeS with claims of 80%of existing revenues at risk in a short-medium termAdoption of prev technologies like Cloud led to overall increase in ITS revenues;65%client cos.still comprehending GenAI&developing adoption strategyWill

22、replace jobs and replace humans in multiple functions,big companies like IBM have already paused hiring in support functionsHuman involvement is crucial in using GenAI technology,and experience confirms the creation of new tech-related jobsConcerns about power consumption and carbon footprint will f

23、ade away as GenAI scales&uses renewable energyTraining GPT-3 consumed 1.287GW hours,equal to the annual energy usage of 120 American housesInvestments in the past three years totaled around$20Bn,five times the investments made between 2018-20Dip in TSRs of IT services stocks recovered,BPO stocks sti

24、ll riding low but found unwarranted by leading investorsShift in technologyAdoptionEnterprisesIT/ITeS sectorEmployeesImpact on SocietyInvestorSCENARIO BArchitects of Change:Generative AIs Role in Revolutionizing IT Services8 Fundamental shift in technology leading to a faster adoption in enterprises

25、 and a lasting impact on IT/ITeS sectorWhat do the leaders sayThe development of AI is as fundamental as the creation of the microprocessor,the personal computer,the Internet,and the mobile phone.It will change the way people work,learn,travel,get health care,and communicate with each otherThe main

26、thing that is interesting about it(ChatGPT)is its generality.That this one system,truly general,is good Bill Gates,Co-founder of Microsoft Sam Altman,CEO of OpenAINever-seen-before like adoption of GPT based tools ChatGPT reached 1Mn subscribers in 5 days and 100Mn subscribers in 2 months post launc

27、hThe race to win for the service providers has already begun Front runners like IBM,Cognizant,Infosys have launched GenAI led suite of offeringsAccelerated investments and registration of patents in GenAI space$25Bn investments in last 3 years,5x of investments in 2018-20 GenAI related patents regis

28、trations in 202113x vs 2016Initial pilots suggest significant impact potential across industries 50%improvement in productivity of coders 29%of the organizations surveyed have already implemented first solutionAdvanced AI network with high accessibility&relevancy across uses cases Enhanced concept u

29、nderstanding via 175Bn+parameters in GPT vs.10Mn previously Highly accessible natural language human-computer interfaceSCENARIO AArchitects of Change:Generative AIs Role in Revolutionizing IT Services9GenAI foundational models have as many parameters as 17,500 traditional AI modelsEnhanced generalit

30、y&relevancy for multiple use cases via 175Bn+parameters in GenAI vs 10Mn previously10K10Mn175BnPower(models parameters)Time Basic analytics Human generatedinsight Not broadly scalableManual Basic statistical models Needs one model per task Human directed insights Access insights via statistical code

31、 Mostly neural networks Needsone model per task Insights based on data-structured patterns(e.g.,images of cats are all cats)Access insights via statistical codeMachine Learning and Deep Learning Basic analytics Human generated insight Not broadly scalableGenerative AIAI TechnologyPre AI EraGenAI Era

32、SCENARIO AArchitects of Change:Generative AIs Role in Revolutionizing IT Services10GenAI is trained on a massive datasets,allowing it to understand the nuances of human languageAbility to generate text/code that is more natural and human-like than traditional AI modelsThe human like understanding of

33、 GenAI makes it much more easily for a broader audience to use it and it doesnt require technical knowledgeWe all remember the old-school travel agent who was a human being,who knows something about you and what you could afford and would do it all for you when putting together your travelThats what

34、 generative A.I.is basically going to do.Youre going to have that conversation again.And its going to be with an A.I.agent Glenn Fogel,CEO of BFind the bug with this code:.for var i=o;i console.log(i).1000).The code contains a bug that will cause it to always print the number 5 to the console,rather

35、 than the expected sequence of numbers from 0 to 4.This is because the set Time out function creates a closure that captures the value of i variable,but the i variable is changed before the callback function is executed.To fix this bug,the code can be modified to create a new variable in the loop th

36、at is passed to the set Time out callback function,rather than using the loop variable directly.This will ensure that each callback function captures the correct value of the loop variable,and the expected output will be produced.Here is an example of how the code could be fixed:YouGPTExplain how el

37、ectricity works to a kid who likes dinosaurs.Electricity is like a group of tiny dinosaurs called Electrosauruses running through special paths called wires.They move together in one direction to create an electrical current,like a dinosaur stampede.The Electrosauruses cross a bridge,called a switch

38、,to make things like lights and toys work.Advanced natural language capabilities allows for higher accessibility for end usersSCENARIO AArchitects of Change:Generative AIs Role in Revolutionizing IT Services11GenAI has witnessed unprecedented levels of adoption by consumersChatGPT reached 1Mn users

39、in 5 days,a fraction of the time it took previous viral hits30Mn550Mn1.7Bn1Bn100Active users recordedDays to achieve 1Mn usersSite visits in March23Site visits in April23Expected customer baseMillion Users recorded in just 2 MonthsBard AIChatGPTBCGAmazonServicenowSalesforceAtlassianMicrosoftMorganSt

40、anley75 days5 days150 daysTop Companies partnered with OpenAI since GPT-3Source:Press Articles;BCG analysisChatGPTInstagramSpotifySCENARIO AArchitects of Change:Generative AIs Role in Revolutionizing IT Services12GenAI startups received$20Bn in investments in the past three years,five times more tha

41、n investments between 2018-2020Source:BCG analysis;BCG Center for Growth&Innovation analyticsPrivate Investments($Bn)#of companies foundedData for Q1 for 202350%growth in annual investments in GenAI space between 2019 and 2023#of GenAI companies founded between 2016-22 have grown at 20%CAGR201320142

42、0152016 201720182019202020212022 20231221194269659510310712281+19%201820192021 2020202220230.22.01.64.83.511.9Data for H1 for 2023OpenAI,Pieces&SambaNova Systems are the top 3 firms with highest investmentsUSA emerging as the biggest hub with 375 companies founded since 2013+56%Architects of Change:

43、Generative AIs Role in Revolutionizing IT Services13GenAI research seeing explosion in patent activity&further reasearch2017 2018 2019 2020 2021 2022#Patent Families28424635384259116276568657764435IBMAlphabetBaiduMicrosoft Chinese Academy of SciencesPing AN InsuranceSamsung ElectronicsAdobeSiemensRo

44、bert BoschNTTHuawei TechnologiesNEC CorpNVIDIA Corp.Capitalone1168074585553474644413530282727Note:3.1K patent families that are related to“Generative AI”domain filed since 2010 are considered for the analysis.Chinese applications are excluded to avoid noise/skewing of dataSource:Derwent Innovation;B

45、CG analysis;BCG Center for Growth&Innovation AnalyticsNo.of patents filed in 2021 about 13x as compared 2016IBM at the forefront of filing GenAI related patents in the last decade13xData incomplete due to publication lag time2010 2011 2012 2013 2014 2015 2016Architects of Change:Generative AIs Role

46、in Revolutionizing IT Services14Initial pilots demonstrate the significant impact potential of Gen AI across various industries1.CarMax puts customers first with car research tools powered by Azure OpenAI Service 2.Jasper Case Study on Mongoose 4.Observations on a real case of deployment of GitHub C

47、opilot on 100 Software Engineers 5.Persado case studies()6.Information Sciences“Using generative adversarial networks for improving classification effectiveness in credit card fraud detection”CarMax replaces time-consuming manual summarization process1,withof content generated in days(with 80%editor

48、ial review approval rate)11 YEARSMongoose leverages Jasper to generate finished content and propose AI-generated edits,resulting inincrease in organic traffic2166%IT organization implemented Git hub Copilot with strict governance and quality gates4overall productivity increase(with 40-50%commits inc

49、reases)50%Insurers generated new app names and descriptions to drive downloads,resulting inincrease in app downloads566%Generative AI Identified a novel drug candidate for the treatment of IPF in(vs.years with traditional methods)3 21 DAYS Synthetic GAN-enhance training set for fraud detection achie

50、ved aaccuracy rate (vs.97%with unprocessed original data)698%AUTOMOTIVESOFTWAREINSURANCEFINANCIAL INSTITUTIONSBIOPHARMAMEDIAWe use ChatGPT to write code Software development that used to take anywhere from eight to 10 weeks now can be done in less than a week Girish Mathrubootham,CEO of FreshworksNa

51、vigating the next AI Frontier:Assessing the Impact of Generative AI in the IT&BPM Services Industry14SCENARIO AArchitects of Change:Generative AIs Role in Revolutionizing IT Services15GenAI is an emerging technology that will see humbler pace of adoption despite initial hypeWhat do the leaders sayWe

52、 have two guidelines that weve established for ourselves in terms of using generative A.I.One is that we are only going to use it on our closed data sets.which means approved data that has been vetted.The second is,its always going to have a human in the loop.You can give it a series of questions wh

53、ich will cause it to believe that up is down,and gravity doesnt exist,and left becomes right-things which are nonsensical to any human being.We dont fully understand the nature of how these large language models work,and we know they make mistakes.Rohit Kapoor,CEO of EXL Eric Schmidt,Former CEO of G

54、oogleExtensive data comes with high computational costs&associated risks Integration of GenAI in search is expected to add significant dollars in Alphabets costs Hallucinations,biased outputs,privacy concerns are some of the associated risksNo immediate impact on clients industries,will amplify grad

55、ually 80%buyers believe GenAI will have moderate to substantial impact in first 1-2 years 60%believe that the impact will amplify in next 2-5 yearsDecades often pass before widespread technology implementation Adoption rate for big data in 2022 still at 55%while 35%for AI/ML,automation and IoT Only

56、15%enterprises surveyed had implemented E2E digital transformation IT stocks have maintained consistent multiples due to revenue stability Dip in TSR index witnessed post ChatGPT and GPT-4 launch was recovered in a couple of months BPO returns still down but deemed unwarranted according to leading i

57、nvestorsOrganizations still comprehending GenAI,developing adoption strategy&evaluating internal capabilities Initiatives will be actioned in the due course of timeSCENARIO BArchitects of Change:Generative AIs Role in Revolutionizing IT Services16Economic viability and risk mitigation still a concer

58、n regarding GenAIs adoptionCase Study|Integration of GenAI in search is expected to add significant dollars in Alphabets costsMultiple risks associated with GenAIs adoptionLeaks of proprietary dataPrivacy concerns and security breaches while trainingHallucinationsCan present factually incorrect resp

59、onses in a convincing mannerBiased outputsModels trained on biased data may generate biased outputsCarbon footprintHigh power consumption while both training and usageCopyright infringementCan potentially lead to lawsuits for using copyright protected dataSophisticated phishing and fraudGenerate con

60、vincing phishing emails or deepfakes instantlyQueries handled by GenAIEstimated cost to Alphabet according to number of words generated per queryCosts driven by high power consumptionTraining GPT-3 required 1.287GW hrs which is roughly equal to annual electricity requirement of 120 American houses25

61、 words10%20%30%40%50%50 words75 words100 words$1.2Bn$1.8Bn$2.4Bn$3Bn$2.4Bn$3.6Bn$4.8Bn$6Bn$1.8Bn$3.6Bn$5.4Bn$7.2Bn$9Bn$2.4Bn$4.8Bn$7.2Bn$9.6Bn$12BnSource:Analyst reports;News articles;BCG analysis$1.2Bn$0.6BnSCENARIO BArchitects of Change:Generative AIs Role in Revolutionizing IT Services17Technolog

62、ies typically require decades to mature before widespread implementation across industriesAdoption rate of emerging technologies in organizations worldwide,2022CloudBig dataRobotic process automationAI/MLLoTAR/VRBlockchain/distributed ledger1990s19562000s19822000s1960s2008Year of inception2006201220

63、00s2000s20122009-102010sYear of inflection pointKey milestoneAWS launched cloud servicesAdoption of Hadoop&NoSQL database queriesBlue prism launched commercial RPA offeringsSolving problemsin Computer Vision and Speech TechUS NIC1 recognized IoT as disruptive civil technologyARToolKit,an open-source

64、 software library releasedBitcoin blockchain went live8%12%33%33%34%55%89%1.US National Intelligence CouncilSource:Digital Leadership Survey,2022 by Harvey Nash N=1785;News Articles;BCG analysisSCENARIO BArchitects of Change:Generative AIs Role in Revolutionizing IT Services18Even in case of scaled

65、adoption,very limited large scale implementations seenOnly 15%enterprises able to transition to end-to-end digital transformation even after years of digital tech availability%of respondents1 by focus of business goals in transformation54%15%Digital marketing,customer centricity,personalizationDigit

66、izing/automating operationsInnovation&new business opportunities,product enhancementsEnd-to-end transformation89%59%1.Respondents asked to select specific goals for their last digital transformationSource:BCG Global Digital Transformation Survey Apr-Jun 2020;n=775The timelines for mainstream adoptio

67、n by enterprises in mainstream is often much humbler than initial hypeSCENARIO BArchitects of Change:Generative AIs Role in Revolutionizing IT Services19TSR in IT sector recovered after initial shock of GPT4 launch;Fall in BPO sector called unwarrantedTSR Index:31 March 2022=100Note:TSRs,share price

68、s,and market cap as of 12-May-2023.Based on index median weekly returns.Large IT Services Peers:Infosys Limited,Tata Consultancy Services Limited,HCL Technologies Limited,Accenture plc,Cognizant Technology Solutions Corporation,Wipro Limited,Capgemini SE and International Business Machines Corporati

69、on,LTIMindtree Limited;Mid-size IT Services Peers:Coforge Limited,Persistent Systems Limited and Tech Mahindra Limited Source:S&P Capital IQ;BCG ValueScience Center6080100120Mar 2022Nov 2022May 2022Jan 2023Jul 2022Mar 2023Sep 2022May 20231122Overall,we believe we are too early in the adoption curve

70、to assess the exact impact,but prior trends such as RPA(or cloud in ITS)suggest such tools actually end up being a tailwind for leading vendors,rather than being a headwind,and certainly dont warrant a 16%decline in stock price(MTD)vs.1%lower S&P 500.We remain OW rated JP Morgan on EXL,May 12,202330

71、-Nov-22:ChatGPT 3.5 Launched14-Mar-23:ChatGPT 4.0 Launched98-S&P Global 1200 IT Services -1.5%99-BPO90-Mid-Size IT Services86-Large IT Sevices83-SBI NIFTY IT ETFSCENARIO BArchitects of Change:Generative AIs Role in Revolutionizing IT Services20GenAI is the latest inflection point-the fourth generati

72、on of AI evolutionMachine LearningDeep Learning(via multi-layer neural networks)Generative AI(e.g.,via transformer based model)Rule-Based1K10K10Mn175BnSystems follow a programmed set of rulesSystems automatically learn from data w/o being explicitly programmed,e.g.,to predict values or identify corr

73、elationsLearning from dataClassificationPrediction.Control systemsAnomaly detection.Natural language processingSpeech recognitionMachine visionText&code generationImage&video generation3D model generationSystems use existing EncoderDecoderDeterministic behaviorProbabilistic behaviorArchitects of Cha

74、nge:Generative AIs Role in Revolutionizing IT Services21GenAI is poised for more success than earlier generations of AILeveraging natural language interfaces to generate code,language,or visual responses,GenAI is accessible&user-friendly for every individual with keyboard competenciesUnlike previous

75、 tools that were developed for specific use cases,GenAI operates on a global scale,providing marketplace-based access to expertise and knowledge across all fields(e.g.,millions of libraries)GenAI collaborates with developers through an iterative process,functioning as a partner rather than a mere to

76、ol,and evolving its capabilities as developers improve their usageGenAI empowers human-friendly interaction and facilitates automatic documentation,granting people greater control over the solution than previous technologiesGenAIs transformer model architecture enables context-based learning,which r

77、educes the outputs dependency on the quality of the training set and eliminates the need for labeled dataFrom expertise required to low barriers to entryFrom pre-determined to learn-as-you-go processFrom black-box to glass-box technologyFrom neural networks to transformer architectureFrom specific t

78、o cross-the-board expertiseGenAI the new tech frontierWhatare buyerssaying?How are IT/ITeS firms responding to the opportunity?Implications for IT/ITeS sectorTABLE OF CONTENTSCHAPTER 1CHAPTER 2CHAPTER 3CHAPTER 42-2122-4041-4849-65EXECUTIVE SUMMARYIn a recent GenAI focused IT buyer survey,we heard:85

79、%respondents believe GenAI will significantly improve customer experience&drive efficiency in core processes However,impact will not be immediate but will amplify gradually 60%enterprises expecting substantial to high impact on their businesses only in 2-5 years timeline Race to GenAI has begun:29%r

80、espondents have started implementing a GenAI solution and confident of implementing first use case within a year,prioritizing functions such as customer support,marketing,technology,and product development Talent shortage,lack of leadership understanding,privacy concerns&lack of clear business case

81、are the biggest barriers for GenAI implementation To get started,One-fifth of the buyers have started to partner with GenAI product companies as well as training their employees on GenAI Buyers are open to availing services from IT services providers,however,expect the commercial models to evolve60%

82、of the buyers would be open to work with IT Services firms on data management&analytics,AI model development,Integration&deployment as well as security services80%of buyers anticipate existing commercial arrangements with higher productivity commitments on new contracts as well as alternate commerci

83、al models like value/gain-share based contracts for GenAI solutions Majority of buyers still anticipate a potential rise in outsourcing for customer service and application servicesArchitects of Change:Generative AIs Role in Revolutionizing IT Services24FunctionsConsumerEnergyFinancial InstitutionsI

84、nsuranceHealth CareIndustrial GoodsPublic SectorTechMediaTelcoTravel&tourismMarketingSalesSupply chain&operationsResearch&developmentCustomer supportLegalFinanceHuman resourcesPotential impact of generative AIGenerative AI has the potential to transform how clients operate across an array of industr

85、ies and functionsIndustriesArchitects of Change:Generative AIs Role in Revolutionizing IT Services25Our buyer survey focuses on four key themesExpectations on the impact of GenAITimelines for impact to be substantialFunctions to be impacted the mostCurrent usage of GenAITimelines to implement first

86、GenAI solutionFunctions,use cases and applications to be prioritizedBarriers to implementGenAICurrent focus areas to achieve adoptionState of present IT/Digital infrastructureTools/partnerships being leveragedOfferings that IT/ITeS firms should developEstimated impact on outsourcing&off shoringComme

87、rcial models for GenAI driven servicesView on impact of GenAIView on IT/ITeS servicesView on internal capabilitiesCurrent state of adoptionArchitects of Change:Generative AIs Role in Revolutionizing IT Services26Sample covers 44 CDO,CIO,COO&CTOs of client industries to IT/ITeS sector across geograph

88、ies with some knowledge of GenAIShare of respondents(%)Share of respondents(%)HealthcareManufacturing/Industrial GoodsConsumer/RetailInsuranceBankingAutomotive25%18%16%16%14%7%Travel5%United StatesFranceUnited KingdomGermanyIndiaAustraliaChinaSwitzerlandSpainItaly48%11%9%9%9%5%2%2%2%2%82%Very famili

89、ar with GenAI&can discuss industry trends involving GenAI18%Somewhat familiar with GenAI&can discuss current industry trendsNote:All organizations with revenues greater than$1Bn for FY23Source:BCG GenAI Survey 2023,N=44Covers respondents from multiple geographies across key client industries Archite

90、cts of Change:Generative AIs Role in Revolutionizing IT Services27QSource:BCG GenAI Survey 2023,N=44Half of the respondents believe GenAI will improve customer experience,followed by driving efficiency in core processes Share of respondents(%)Significantly improve customer experiences-still in its e

91、arly stages of development50%Drive efficiency in core processes36%Bring out new commercial&monetization models in the industry11%Disrupt my business model significantly2%What are your thoughts on the future of GenAI next 1-2 years?Architects of Change:Generative AIs Role in Revolutionizing IT Servic

92、es28Source:BCG GenAI Survey 2023,N=4460%enterprises expecting substantial to high impact on their businesses only in 2-5 years timelineModerate to substantial impact of GenAI in next 1-2 yearsSubstantial to high impact of GenAI in next 2-5 yearsShare of respondents(%)Share of respondents(%)Minimal I

93、mpactLow ImpactModerate ImpactSubstantial ImpactHigh Impact2%11%48%52%39%14%30%5%Architects of Change:Generative AIs Role in Revolutionizing IT Services29Customer support,IT&marketing are top 3 functions likely to be impacted by GenAISource:BCG GenAI Survey 2023,N=44Substantial ImpactModerate Impact

94、Low ImpactCustomer Support3.9Information Technology3.5Marketing3.4Enterprise Risk Management3.1Sales3.0Workforce Management and Culture3.0Product Dev and Management2.8Supply Chain2.7Finance and Accounting2.7Manufacturing2.6QWhat level of impact do you anticipate Generative AI having on the following

95、 enterprise-wide areas over the next 1-2 years?Scale of Impact 1 No Impact2 Low Impact3 Moderate Impact4 Substantial Impact 5 High ImpactArchitects of Change:Generative AIs Role in Revolutionizing IT Services30In the process of developing GenAI use cases20%Piloting processes in use case with GenAI30

96、%Implementing GenAI use cases for processes within the organization20%Implementing GenAI use cases directly for customers/clients9%Only 29%organizations are in the process of implementing GenAI use casesShare of respondents(%)Source:BCG GenAI Survey 2023,N=44Discovering what GenAI is and how could i

97、t be beneficial?21%To what level does your company currently use Generative AI?QArchitects of Change:Generative AIs Role in Revolutionizing IT Services31Customer SupportOthersMarketingProduct Development&ManagementTechnology60%of the organizations adopting GenAI plan to implement their first GenAI s

98、olution within a yearEarly adopters have implemented a solution;others will follow within a yearFour key priority functions witnessing greater activity for GenAI implementationWhat is your expected timeline to implement your first GenAI solution?In what function(s)have you already implemented or are

99、 in the process of implementing GenAI solutions?Share of respondents,adopting GenAI1(%)1.Includes organizations in the process of developing/piloting/implementing GenAI use casesSource:BCG GenAI Survey 2023,N=44QQ 23%Already implemented at least 1 solution30%Within next 6 months15%Within next 2 year

100、s3%Within next 5 years30%Within next 1 year11%15%16%23%35%Share of responses(%)Architects of Change:Generative AIs Role in Revolutionizing IT Services32Organizations are prioritizing a plethora of use cases across functions(I/II)Source:BCG GenAI Survey 2023,N=44Service DesksAuto RespondChatbotsAgent

101、 co-pilot for guided supportData-based insightsLive Chat AI TriageSentiment analysisAutomating FAQsIncident ManagementCustomer SupportCode GenerationVector EmbeddingsIntegrating and Finding next gen techAutomation infra managementData&analyticsAutomated testingImage based medical diagnosisServer har

102、deningImage visual searchProcess controlSystem ImplementationIntegratingFinding next gen techBug detectionFirewall configurationTechnologyData&analyticsRoboticsAI-Powered consultingContent generation&validationCode base&archtectureRapid product iterationProduct design&validationUX/UI improvementsDev

103、eloping product roadmapsVisual StoryboardsCustomer CX journeyEdge case developmentProduct development&managementWriting copiesLead generationKeyword stemming&SEOAnalyzing sales patterns&campaign effectivenessData&analyticsAd creation&prototypingPersonalizationBuying behaviour analysisPredictive mark

104、eting trendsContent creationAutomation of campaignsMarketingArchitects of Change:Generative AIs Role in Revolutionizing IT Services33SalesOrganizations are prioritizing a plethora of use cases across functions(I/II)Source:BCG GenAI Survey 2023,N=44ManufacturingProcess improvement Workforce optimizat

105、ionSkilled labour supportSupply ChainPlanning&forecastingWarehousing support Managing inventoryTransport route optimizationBuilding demand scenariosFinding new vendorsCosting analysisAutomation of sales processesData&analyticsClient nurturingSelf-service kiosksStandardize CRM entriesPersonalizationW

106、riting emailsCustomer behaviour analyticsProduct portfolio studyProduct positioning recommendationsMonitoring performative linked incentivesResearchContent researchScenario planningMarket analysisRisk analysisReport generationSurvey design and analysisFP&AAccounts receivable and payables reconcilati

107、onAnalysis&reportingVendor management Onboarding new vendors Fraud detectionEnterprise risk managementWorkforce management&cultureEmployee onboarding&offboardingEmployee training qualityDeveloping cyber risk modelData&analyticsFraud detectionBrand securityRisk assessment&remediationArchitects of Cha

108、nge:Generative AIs Role in Revolutionizing IT Services34Language generationInsight generationStructured data creationImage and video synthesisCreative applicationsSimulation and gaming57%54%22%16%5%1%0%Data managementLanguage&Insight generations the top application areas#of responses weighted1 by ra

109、nks1 Weights of ranks rank 1 weighs 3,rank 2 weighs 2 and rank 3 weighs 1Source:BCG GenAI Survey 2023,N=44QWhich 3 applications areas of Generative AI do you believe will be most relevant at your company?SCENARIO BArchitects of Change:Generative AIs Role in Revolutionizing IT Services35What are curr

110、ently the biggest barriers to implementing Generative AI within your company?Q#of responses weighted1 by ranksLack of skilled talent to develop&implementIdentifying appropriate modelsPotential risk and/or privacy threatsLack ofinvestmentLack of leadership understanding/strategyNo clarity on specific

111、 use casesLack of clear business careInsufficient technology infra to supportCultural resistance internallyLegal exposureInability to access/leverage data1 Weights of ranks rank 1 weighs 3,rank 2 weighs 2 and rank 3 weighs 1Source:BCG GenAI Survey 2023,N=44Talent shortage,lack of leadership understa

112、nding,privacy concerns&lack of clear business case are the biggest barriers for GenAI implementation48343228241916118538Architects of Change:Generative AIs Role in Revolutionizing IT Services3623%22%17%15%12%11%Share of respondents(%)Note:The rest 1%is others which is driven by discovery and investi

113、gation about GenAISource:BCG GenAI Survey 2023,N=44Partnering with companies developing GenAI based productsCreating trainings&workshops to understand GenAI use casesPartnering with IT services providers using GenAIModifying on-going processes with GenAI to understand its implicationsHiring people w

114、ith GenAI relevant skillsetsUpgrading tech and data stack for GenAIQWhat capabilities have you started building within your teams?Businesses are prioritizing employee training and building ecosystem for GenAI implementationArchitects of Change:Generative AIs Role in Revolutionizing IT Services37Mult

115、iple partnerships&tools being used by organizations to implement GenAI use casesmidjourneygithub-co-pilotbardopenaimicrosoftgoogleawschatgptstable diffusionMicrosoft-coMicrosoft-co-pilotazure-cloudvisual-studiodatabricksdall-e2quillbotservicenowmongodbtitanllamaadobeinviabertdall-envidiabria-aigpt-4

116、jasperpilotilotbradsynthesizazure co-pilotaleph-alphakerastwillsukicurieukggpt-3resnetgptaws-sagemakerpinecone.ioalphabetmoveworksaccenturekpmgtensorflowvggnetsnowflakeeinsteingptgoogle-cloudgoogle-vertexazurepytorchSource:BCG GenAI Survey 2023,N=44Architects of Change:Generative AIs Role in Revolut

117、ionizing IT Services3860%of the buyers prefer to have AI model development,data management&analytics,security&privacy and Integration&deployment services from IT services companiesAI model developmentData management&analyticsIntegration&deploymentUser experience&UI designEthical&responsible AIPerfor

118、mance monitoring&maintenanceContinuous innovation&learningSecurity&privacySource:BCG GenAI Survey 2023,N=4416%16%14%14%11%11%10%8%QWhat should IT services companies offer with reference to GenAI?Share of respondents(%)Architects of Change:Generative AIs Role in Revolutionizing IT Services39Commercia

119、l models expected to evolve in favor of transaction-based or value share-based80%respondents anticipate new commercial models or strengthened commitmentsTransaction-based or value/gain share-based models are preferred for engaging IT/BPO service providers in GenAI servicesUse the same commercial mod

120、el as todayUse the same commercial model as today,but with higher Y-o-Y productivity improvement commitments20%39%41%33%23%16%14%9%2%2%Transaction/usage based include licensesValue/gain share-basedFixed FeeTime&materials basedFTE based CombinationFee atriskSource:BCG GenAI Survey 2023,N=44Share of r

121、espondents(%)Driven by fixed fee and gain-shareArchitects of Change:Generative AIs Role in Revolutionizing IT Services40Outsourcing likely to decreaseOutsourcing likely to increase59%59%50%50%48%45%41%41%50%50%52%55%Most buyers anticipate a potential rise in outsourcing for customer service and appl

122、ication servicesQDo you think the share of outsourcing for some processes will increase using GenAI?Share of respondents(%)Source:BCG GenAI Survey 2023,N=44Customer serviceMarketingApp development&maintenanceF&AProcurmentInfra ManagementGenAI the new tech frontierWhatare buyerssaying?How are IT/ITeS

123、 firms responding to the opportunity?TABLE OF CONTENTSCHAPTER 1CHAPTER 2CHAPTER 3Implications for IT/ITeS sectorCHAPTER 42-2122-4041-4849-65Architects of Change:Generative AIs Role in Revolutionizing IT Services42While GenAI adoption is still at a nascent stage,first movers have started to launch Ge

124、nAI led suites of services for clientsNote:Insights based on the information available in the public domainSource:Analyst Reports;News Articles;BCG analysisFirst movers in IT service providers space have started to launch client facing GenAI led suites IBM introduced Watsonx,featuring Watson.ai,that

125、 enables AI builders to train,test and deploy GenAI capabilitiesestimates 30%of internal back-office jobs to be replaced by AI over time Cognizant unveiled Neuro to automate enterprise business process outsourcing,aiming to reduce ops costs by up to 45%Infosys launched Topaz-AI-first set of services

126、 and solutions LTIMidtree has launched Canvas.ai GenAI platform designed to help enterprises jump start and scale generative AI capabilitiesOther IT services players have also accelerated investment in building GenAI capabilities Accenture established GenAI and LLM CoE with 1.6k professionals and pl

127、ans$3Bn investment over 3 years in its Data&AI practice TCS plans to train 25,000 engineers on Azure Open AI to scale its expertise and launch new offerings HCLTech will train 18,000+engineers on Google Cloud GenAI tech and has also expanded strategic partnership with Microsoft Persistent systems ha

128、s established a dedicated Microsoft Viva practice and has equipped it engineers with Amazon CodeWhisperer Wipro has been investing in GenAI research for 2+years with academic institutions and plans to invest$1Bn into AI over the next 3 yearsWithin BPM space,early movers have started to develop AI po

129、wered suite to incorporate GenAI capabilities EXL has launched an AI platform in collaboration with Google AI,IBM Watson,and Microsoft Azure Genpact has developed a suite of AI-powered solutions and have started prototyping with clients,for health tests,cust.service,etc Teleperformance has developed

130、 its own GP product with a potential to automate 20-30%of its volume in next 3 years TTEC expanded its AI CX solutions,including private language models and managed services trained on proprietary data sets TaskUs has been utilizing AI,with a product called TaskGPT,to enhance productivity and client

131、 workflowsArchitects of Change:Generative AIs Role in Revolutionizing IT Services43Players have started to incorporate GenAI led services into their portfolio(I/V)Launched Watsonx,a new AI and data platform featuring Watsonx.ai,Watsonx.data,and Watsonx.governance Watsonx.ai is an enterprise studio t

132、hat enables AI builders to train,test&deploy GenAI capabilities IBM expects AI to largely automate back-office functions(data entry,HR process automation and repetitive jobs)and estimates 30%of IBMs 28,000 back-office jobs can be replaced by AI over timeIBM Accelerating investment in GenAI using an

133、accelerator process engaging with over 30 clients for prioritizing AI use cases Launched Neuro,an AI platform for simplifying and automating enterprise business process outsourcing Expecting 25-45%improvement in operational costs,30-50%reduction in mean time to delivery&15-30%reduction in FTE Launch

134、ed Cognizant Google Cloud AI University program that will train 25,000 Cognizant professionals on Google Cloud AI technologiesCOGNIZANT Launched Topaz,AI-first set of services&solns.,ensuring ethics,trust,privacy,security,&compliance Developed MasterCraft software for code generation Launched free A

135、I/GenAI certification on Infosys Springboard Virtual Learning Platform Infosys signed a five-year AI deal with$2Bn target spend with an existing clientINFOSYS Launched Quasar,a cognitive AI platform with services like documentation,voice analysis&conversational AI Expanded its cybersecurity services

136、 to include threat intelligence,protecting information assets and ensuring privacyCOFORGENote:Insights based on the information available in the public domainSource:Analyst Reports;News Articles;BCG analysisPlayerDescriptionArchitects of Change:Generative AIs Role in Revolutionizing IT Services44Pla

137、yers have started to incorporate GenAI led services into their portfolio(II/V)Partnered with Google Cloud to launch TCS Generative AI for custom-tailored business solutions Enhancing automation suite with large language models TCS plans to train 25,000 engineers on Azure Open AI to scale its experti

138、se and launch new offerings TCS is currently engaged in 50+PoCs/pilots on GenAI and have 100+offers in pipelineTCSTECH MAHINDRAACCENTURECAPGEMINI Launched Generative AI studio as part of its amplifAIO-suite,enabling enterprises to generate high quality customized content faster It is applicable to a

139、ll business agnostic of the size Established Generative AI and LLM Center of Excellence with 1.6k professionals Partnered with EinsteinGBT,and have started assisting clients Predicted that language-based AI will support or augment up to 40%of all working hours Spotted 300+use cases where generative

140、AI can deliver Announced a$3Bn investment over 3 years in its Data&AI practice Capgemini and Google Cloud launch Generative AI CoE to develop a library of 500+enterprise-ready industry use casesSource:Analyst Reports;News Articles;BCG analysisPlayerDescriptionArchitects of Change:Generative AIs Role

141、 in Revolutionizing IT Services45Players have started to incorporate GenAI led services into their portfolio(III/V)WIPROLTIMINDTREEHCLTECHPERSISTENT Wipro has been investing in GenAI research with the University of South Carolina and IIT Patna for 2+years to build language models for the health sect

142、or Plans to invest$1Bn into AI over the next three years Launched Canvas.ai GenAI platform to reduce time&effort for app modernization&cloud migration Intends to invest$4050Mn in AI capabilities and services during the current fiscal year Has been leveraging CoPilot to assist the new junior employee

143、s in coding HCLTech will train 18,000+engineers on Google Cloud GenAI tech to develop joint soln.with Google HCLTech and Microsoft expand their strategic partnership to develop joint solutions Plans to invest in software labs and train employees in Generative AI,including large language models and A

144、I-paired programming Expanded Azure Center of Excellence to include AI-based modern workplace solutions Established dedicated Microsoft Viva practice integrated with Generative AI Strengthens pact with AWS to equip its engineers with Amazon CodeWhisperer(ML-powered code gen.)Source:Analyst Reports;N

145、ews Articles;BCG analysisPlayerDescriptionArchitects of Change:Generative AIs Role in Revolutionizing IT Services46Players have started to incorporate GenAI led services into their portfolio(IV/V)MPHASISTELEPERFORMANCETASKUSEXL Has been able to drive conversations and secure deals on Generative AI I

146、ncorporated earlier versions of ChatGPT into their algorithms Partnered with Kore.ai to transform customer and employee experience for enterprises Launches Mphasis.ai business unit to focus on AI offerings Had an Application Layer to connect with OpenAI since 2021 Developed its own GenAI product,TP-

147、GPT Expect that there is a potential to automate 20-30%of its volumes over 3 years;possibly a slight acceleration to the 5-7%volume automation observed over the past 4 years TASK has already been using AI for nearly 2 years,demonstrated by Task GPT,an OpenAI-powered platform that improves the produc

148、tivity of its own teammates and its clients workflows TASK has seen an uptick in demand and interest from clients for Generative AI Launched a proprietary AI platform that is designed to help businesses accelerate their digital transformation and plans to launch CoE to build a team of 1500+GenAI spe

149、cialists EXL has partnered with many leading technology companies,including Google AI,IBM Watson,and Microsoft Azure.These partnerships give EXLs clients access to the latest AI capabilities and expertiseSource:Analyst Reports;News Articles;BCG analysisPlayerDescriptionArchitects of Change:Generativ

150、e AIs Role in Revolutionizing IT Services47Players have started to incorporate GenAI led services into their portfolio(V/V)TTECGENPACTWNS Has expanded its AI CX solutions,including private language models and managed services trained on proprietary data sets Their Generative AI Empowered Associate e

151、nhances representative performance through conversational knowledge-based support and automated contact summaries Developed a suite of AI-powered solutions that can help businesses automate tasks,improve decision-making,and gain insights from data Seen pipeline intensity enter prototype/experimentat

152、ion phases with clients,with current focus on health tests,customer service,and research work in unregulated industries using GenAI Partnered with Google Cloud for GenAI capabilities Views Generative AI as an additional automation tool which could help in some of the complex cost takeout projects En

153、visions that these advancements will have a positive long-term impact by empowering their workforce to engage in higher-level tasksSource:Analyst Reports;News Articles;BCG analysisPlayerDescriptionArchitects of Change:Generative AIs Role in Revolutionizing IT Services48Firms are also experimenting w

154、ith use cases internally as well as externallyIT/ITeS players have started helping clients on GenAIUse cases being developed across the spectrum,from ops/delivery productivity/orchestration to G&A function effectiveness/efficiencyOffering build-outInternal use cases Data engineering:Data structuring

155、,Deployment of data systems IT Services:Application development,integrations&deployments Consulting services:Identifying use cases,platform selection,model selection Infra services:L1 support aided by co-pilots,System upgrade automation,script generation PoCs for specific use cases:Customer service,

156、Content generation,workforce management,Patient summary generation,Sales proposal generationNote:Insights based on conversations from Apr-Jul 2023Source:Expert Interviews;BCG analysisOperations&delivery Developer co-pilots:Prompt automation,test script generation,infra remedial script generation,UX

157、design&content generation Internal Knowledge repository Agent co-pilots:Auto-response&Email drafting,SummarizationG&A functions Legal/Procurement:Contract comparison/vetting and MSA drafting HR:Recruitment(Resume shortlisting),Training quality,HR service desk optimization GenAI the new tech frontier

158、Whatare buyerssaying?How are IT/ITeS firms responding to the opportunity?TABLE OF CONTENTSCHAPTER 1CHAPTER 2CHAPTER 3Implications for IT/ITeS sectorCHAPTER 42-2122-4041-4849-65EXECUTIVE SUMMARYTo take advantage of the opportunity,IT and ITeS Services players will have to prepare themselves and act n

159、ow Firms must take a holistic approach to GenAI with a clear focus on strategy,outcomes,tech,people,change management&client engagement.In fact,change management(people&clients)could be 70%of the effort required according to some estimates Imperative to build GenAI strategy on how to participate in

160、the adoption curve:Where to play?,How to Win?,How to Execute?Strategy:Potential to play one or more of 5 orchestrator archetypes for customers Advisory:Leverage existing capabilities in consulting to offer platform selection&solution design Data engineering:Leverage capabilities in data engineering

161、to help set up&maintain training data architecture GenAI Model builder:Partner with companies/leverage open source algorithm libraries to train custom models as per client needsFine tuning:Fine-tune existing third-party models to use for clients,via partnerships with relevant GenAI product companies

162、App dev&maintenance:IT Services players can leverage application&infra services to develop,deploy&maintain the GenAI solutionsOutcomes:Service providers must ruthlessly prioritize the industry,sub-sector,customer,function and function specific use cases to invest efforts behind,based on key axes-Imp

163、act(Efficiency,Customer Experience,Risk reduction)vs Feasibility(Privacy,Security,Ethical concerns,Effort to implement&scale)Architects of Change:Generative AIs Role in Revolutionizing IT Services51EXECUTIVE SUMMARYTech:GenAI technology forms additional layers in the end-to-end tech stack(earlier co

164、nsisting of Infra layer,Transaction(ERP)layer,Data layer&Smart business layer)Choice of GenAI architecture archetype is a balance between effort,privacy,and access to training data Different foundational choices across the tech stack blueprint may be required to support these desired use cases right

165、 from specialized hardware to cloud platforms to foundational models(closed-source vs open-source)and apps on top with right UI/UX In the short term,companies will experiment with large GPT models and slowly progress into more cost-effective,fine-tuned models People&Change management:Three important

166、 aspects of People&change management:Roles&Op Model:While GenAI improves productivity in current roles,the new layer will also lead to requirement of new skills and roles.Organization will need to evolve a new op model(Design,governance,KPIs,incentives)Talent&Skills:GenAI will lead to changing skill

167、 requirements&hence higher need of reskilling,as well as revamped talent management processes Purpose,culture&Change management:Boosting individual adoption,while activating&embedding change through a structured programClient engagement:Service providers should focus on a continuous engagement loop

168、to become an indispensable strategic partner for its clients right from Client prioritization,to Outreach with conviction,to Solution design,to Continuous engagement Prioritize&sequence outreach in a structure manner Enable sales partner with the right pitchbooks&right pre-sales capabilities Develop

169、 PoCs,conduct co-creation&solutioning workshops Explore alternate models of commercials including equitable value sharingArchitects of Change:Generative AIs Role in Revolutionizing IT Services52Four key types of GenAIs impact on IT/ITeS services providersCreation of code snippets,code completion&rev

170、iewDesign prototypingTest case generationDocumentation generationService desk automationInfra ops insightsData strategy&services Content moderation servicesApplication services(deploy&maintain)Cloud services for GenAI solutionsCustomized industry specific foundational models for use&deploymentFine t

171、uning existing GenAI models for use casesE.g.,Personalized onboarding journeys or claims applications in Financial institutionsHigher use of outcome oriented commercial modelsHigher revenue per FTE because of shift to more skilled FTETransform core processAlternate commercial modelNew industry speci

172、fic offeringsNew opportunities in existing servicesArchitects of Change:Generative AIs Role in Revolutionizing IT Services53GenAI revolution will widen the gap(almost irretrievably)between innovators&laggardsDigital leaders to reap significant financial impactfrom digital transformationfrom scaled d

173、igital solutionsHigher returnsCost savings Revenue growthSuccess at scale requires immediate vulnerability management&value deployment while concurrently building long-term organizational AI capabilities(ROI)from digital initiatives3X15-20%15-20%Source:BCG analysisArchitects of Change:Generative AIs

174、 Role in Revolutionizing IT Services54IT/ITeS firms need to take a holistic approach to GenAI with clear focus on strategy,outcomes,tech&peopleBCGs integrated approach to build an GenAI-driven IT Services company and modernize its operations Deployment strategy Prioritized use cases Business impact

175、Where to play?Right to win How to play?Organizational redesign&op model Individual role and skills redesign Enablement&change management Client prioritization Systematic outreach Continuous engagement Tech and data partners Data architecture Proprietary analytics and toolsClient engagement&impactPeo

176、pleTech&DataAI OutcomesStrategy10%20%70%StrategyAI OutcomesTech&DataPeopleClient engagement&impactArchitects of Change:Generative AIs Role in Revolutionizing IT Services55GenAI implementation in any use case has multiple stepsFoundational Model deployment,monitoring&hypercarePlatform selection&solut

177、ion designData readiness&engineeringAlgo development&model trainingModel testingCustomization,configuration&enhancementInfra provisioning&maintenanceApp development&maintenanceSelection of appropriate models based on business needsProprietaryCompanys own dataConfigured/enhanced ModelsModels develope

178、d over foundational models with specific use casesCloud platformsComputer hardware exposed to developers in a cloud deployment modelBERT variants(e.g.,RoBERTa,distilBERT,alBERT etc.)Open-source foundational modelsModels released as trained weightsClosed-source foundational modelsLarge scale pre-trai

179、ned models exposed via APIsModel hubsPlatform/Repositories to share modelsCustomers dataClients own datasetPublicly available informationWikipedia,comman crawl,reddit,.HUGGING FACEREPLITGITHUBBERTAZUREAWSGOOGLE CLOUDCOREWEAVELIGHTONBLOOMADEPTUSHURDEBUILDFront end appsEnd user facing B2B&B2C applicat

180、ionsEnd-to-end appsEnd user facing applications with proprietary modelsCHATGPTOPENAIUSHURJASPERMAGE.SPACEADEPTRUNWAYDHIWISEMIDJOURNEYOPEAN AIGPT-2GLAMDAHUGGING GPTGOOGLECO;HEREAL21LABSBARD AIOPEAN AIGPT-3/4GITHUB COPILOTSTRATEGYArchitects of Change:Generative AIs Role in Revolutionizing IT Services5

181、6Five different orchestrator archetypes identified for IT/ITeS playersDescriptionOrchestrator archetypes24543215Platform selection&solution designData readiness&engineeringAlgo development&model trainingModel testingCustomization,configuration&enhancementInfra provisioning&maintenanceApp development

182、&maintenanceTypically smaller players will find their niche in one of these roles;larger companies will tend to play more than one roleCustom-buildTrain a custom modelUse an existing modelOff-the-shelf13Own offerPartner offer,orchestrationAdvisoryBased on client needs,help select right GenAI model&r

183、ight solution architecture(including daisy chaining of multiple models like Jasper)App development&maintenanceLeverage digital engineering skills to build applicationsDeploy,host&maintain applicationsData engineeringLeverage existing data engineering services for building the right datasets/datalake

184、s to be used to build,customize or configure a GenAI foundational modelGenAI Model builderCo-lead motion with Partner with proprietary algorithms to train custom foundational model based on client data Fine tuningPartner with product cos.with proprietary models&fine-tune it as per client needs with

185、client specific/industry specific dataSTRATEGYArchitects of Change:Generative AIs Role in Revolutionizing IT Services57AI Outcomes|Ruthless prioritization of industry,sector,client and function specific use casePriority 1:need to experiment now to scale fast;Priority 2:need to investigate rapidly,ru

186、n a few POCs to confirm value/feasibility;Priority 3:need to assess potential feasibility within the next 6-12 months to confirm proper time horizon(e.g.refactor legacy code,it would be good to run a small test in 2023&assess what the real maturity of this is Note:all use cases will still involve hu

187、mansDoing nowStarting to doWait and seeRefactor legacy codePersonalized content creationAutomated code ulnerability assessmentHighHighLowLowGenerative code healingRegression test automation Conversational troubleshootingSelf-service chatbotAssit.Code generationAutomated ticket routing Predictive mai

188、ntenanceConversational knowledge accessData CleansingSimplified data queryingCode doc.generationPrompt-based fast wireframingUnit test generationApp MaintenanceInfra&OperationsApp DevelopmentImpact Productivity:How much faster am I completing tasks?Time:How much time can I re-allocate to value-added

189、 tasks?Quality:How much better is my output from GenAI?Feasibility Tech:How ready is my tech stack to support use cases?Are market solutions mature?Data:How available are the data/data platform capabilities required?Risk:What is our risk appetite level?How confident are we on our safety&security mit

190、igations?People:How is the availability of the required talent/skillset as well as op model?AI OUTCOMESArchitects of Change:Generative AIs Role in Revolutionizing IT Services58Tech&Data|Generative AI will inevitably become another part of the architecture1.Return on Investment 2.Business as usual Pa

191、stSiloed AI/ML/RPA effortsHighMedLowHighLowLowUnclearHigh(but focused)Very high(at scale and BAU2)CostBarrier to entry Business value(ROI1)CurrentSmall scale GenAI implementationFutureFull architecture integrationGenAI assetsOrg architectureTECH&DATAArchitects of Change:Generative AIs Role in Revolu

192、tionizing IT Services59GenAI technology forms additional layers in your end-to-end tech stackOmnichannelERPOn-premCloudHybridData ProductsIngestion&DistributionRepository&StorageBusiness ServicesOther SystemsOperational Data ServicesCore Transaction LayerData LayerOn-premCloudHybridTPU-GPUData produ

193、ctsOperational Data ServicesRepository&StorageIngestion&DistributionChatCognitive ServicesWebCopilot.App builderCognitive AppsBusiness ServicesOmnichannelModel GardenAI AgentAI MarkletplaceAI GuardrailsAI ModelAI OpsTraditional stackGenAINewImpactedImpactedImpactedSmart Business LayerCore Transactio

194、n LayerInfrastructure/CloudInfrastructure/CloudData layerSmart Business LayerAI LayerTECH&DATAArchitects of Change:Generative AIs Role in Revolutionizing IT Services60Choice of GenAI architecture archetype is a balance between effort,privacy,and access to training dataManaged secure1Hybrid privateFu

195、lly private1.For example,Azure OpenAI Service 2.Model topology only 3.Regarding IP/copyright of training dataSource:BCG analysisCost driver(usage)Within boundaries of enterprise IT,service contractsOpenAI announced March 1st they also keep data privateAPI pricingAPI pricingHosting/computeHosting/com

196、puteDevelopment effortsLittleLittleDevelopment effortsDevelopment effortsData kept privateNoYesYesYesLegal status3RiskyRiskyTrainingFine-tuningRiskyControllablePublic researchSolution providerClientModelTopo.2Enterprise contextFully publicModelUser/AppUser/AppUser/AppUser/AppModelModelModelModelTECH

197、&DATAArchitects of Change:Generative AIs Role in Revolutionizing IT Services61In the short term,companies will experiment with large GPT models and slowly progress into more cost-effective,fine-tuned modelsBuild proof of concept and demonstrate value to business Identify a targeted business use case

198、 and understand how GPT models can address itDevelop a proof of concept to showcase the feasibility and potential value of GPT modelsOptimize models for cost and prepare to scaleStart research on more cost-effective infrastructure,such as Google TPU and Nvidia GDX GPU and smaller more-cost-effective

199、 modelsFine-tune GPT model architecture and parameters for improved performance and efficiencyContinuously fine tune foundational modelsContinuously optimize cost by fine-tuning smaller and special purpose language models(10B parameters)to solve functional tasksMonitor real-world performance and col

200、lect user feedback for iterative model improvementEstablish a feedback loop and data-driven process for ongoing model fine-tuning and enhancementEnd state:Prepare foundational models to integrate into SaaS providers or be a separate SaaS offeringMedium TermShort TermLong TermTECH&DATAArchitects of C

201、hange:Generative AIs Role in Revolutionizing IT Services62People|New roles will emerge requiring new skills,and the overall organization will have to evolve its operating modelPurpose,culture&change managementGreater productivity&redesigned workChanging skillset requirements and new career pathsNew

202、roles to govern&unlock AI potentialRevamped HR processes(e.g.,talent acquisition,performance management)Changing how people work togetherIncreasing need for fluid talentRedesigned organizations reflecting changed workPersonalized trainingBehavioral science,nudges at scaleworkExtensive change managem

203、ent needsRoles,operating models&org structuresTalent&SkillsPEOPLEArchitects of Change:Generative AIs Role in Revolutionizing IT Services63While GenAI improves productivity in current roles,it will also lead to requirement of new skills and roles1.MIT research(Experimental Evidence on the Productivit

204、y Effects of Generative Artificial Intelligence,Shakked Noy and Whitney Zhang 2.BCG publication-CEO guide to GenAI revolution 3.Josh Bersin Note:Background image generated by Midjourney AIMassive value will be unlocked as a workforce integrates GenAI into operationsNew GenAI skills&roles likely to b

205、e neededProductivity ImprovementSuch as using generative AI to generate and summarize content in real-timeWork Reductionfrom organizational efficiencies by automating or streamlining end-to-end processes(e.g.,customer service issue resolution,marketing content creation)Improved Employee Engagement b

206、y reducing repetitive tasks and offering real-time supportGenAI SkillsGenAI algorithms&methodsFoundational models training&validationGenAI model fine tuningAPIs&SDKs 3rd party models knowledge Prompt engineeringGenAI RolesChief AI OfficerAI Ethics&Compliance OfficerAI/ML Governance SpecialistGenAI S

207、oftware EngineersPEOPLEArchitects of Change:Generative AIs Role in Revolutionizing IT Services64Continuous engagement for engagement certaintyDevelop PoCs,conduct co-creation&solutioning workshopsEstablish regular feedback loopClient negotiations using alternate commercial modelsTrack&resolve ongoin

208、g issuesInclude contract clauses for equitable sharing of value,risk mitigationTrack&demonstrate value deliveredAlign funding&saving retention planIdentify dynamic client needs,new use casesEstablish need for continuous change through communicating industry best practices,peer benchmarking etc.Servi

209、ce providers should focus on a continuous engagement loop to become indispensable strategic partner for clientsClient selection&prioritizationClient outreach&convictionSolution design&contract closureSelect sub-sectors and function basis burning platform&Clients offeringClient prioritization framewo

210、rk:Client nature of work,contract statusClient strengths&client relationshipsEstimated value&retained savingSequence client outreachDemonstrate capability fit for client needsBuild conviction on equitable value sharingAlign requirements from client-end(e.g.,tech access etc.)Enable sales partnerGTM p

211、itchbook,value proofRight teamingCLIENT ENGAGEMENTArchitects of Change:Generative AIs Role in Revolutionizing IT Services65Demostrate case for GenAI journeyDemostrate case for GenAI journeyBaselineBaselineDefine GenAI visionRun/support POCsFrame the journeyExplore the realm of possibilitiesMap curen

212、t GenAI capabilitiesFrame expected POC outcomes/requirementsSet the ambition,confirm KPIs,impact on op modelConfirm value,enablers,risk,limitations and requisites to scale POCPrioritize use cases&size GenAI impact across orgDefine transformation roadmap and frame delivery approachIgnite scale transf

213、ormationAccelerate value deliveryBuild capabilitiesExpand beyond front-runnerEnsure feedback loop between overall strategy&POC build outHow to get started|Two track approach has resonated well with CIOsArchitects of Change:Generative AIs Role in Revolutionizing IT Services66Rajiv GuptaManaging Direc

214、tor and Senior Partner,Boston Consulting GroupSumit SarawgiManaging Director and Senior Partner,Boston Consulting GroupVikash JainManaging Director and Senior Partner,Boston Consulting GroupSambhav JainPartner,Boston Consulting GroupMEET THE AUTHORSArchitects of Change:Generative AIs Role in Revolut

215、ionizing IT Services67Boston Consulting Group publishes reports on related topics that may be of interest to senior executives.Recent examples include:Mastering the Generative AI Revolution:Keeping the Focus on the Userhttps:/ CEOs Guide to the Generative AI Revolutionhttps:/ AI Is Under Growing Scr

216、utiny.Heres What That Means.https:/ to Put Generative AI to WorkResponsiblyhttps:/ Generative AI Is Already Transforming Customer Servicehttps:/ AIhttps:/ FURTHER READINGArchitects of Change:Generative AIs Role in Revolutionizing IT Services68ACKNOWLEDGEMENTSVikash Jain is a Managing Director and Se

217、nior Partner,in the New Jersey office of Boston Consulting Group and leads the TMT practice areaSambhav Jain is a Partner,in the New Delhi office of Boston Consulting Group and leads the TMT practice areaThis study was undertaken by Boston Consulting Group(BCG).We would like to thank Raghav Sally,Ak

218、shit Jhamb and other BCG research teams including the entire global GenAI team for their assistance in writing this report.We are thankful to Jasmin Pithawala and Sucheta Desai for managing the marketing process as well as Jamshed Daruwalla,Saroj Singh,Sujatha Moraes,Vijay Kathiresan,Abbasali Asamdi

219、,Soumya Garg,and Yashika M for their contribution to the editing,design and production of this report.ABOUT THE AUTHORSNOTE TO THE READERRajiv Gupta is a Managing Director and Senior Partner,in the New Delhi office of Boston Consulting Group and leads the TMT practice areaSumit Sarawgi is a Managing Director and Senior Partner,in the New Delhi office of Boston Consulting Group and leads the TMT practice area