1、Now decides next:Insights from the leading edge of generative AI adoptionDeloittes State of Generative AI in the Enterprise Quarter one reportJanuary 2024ForewordIntroductionNow:Key findings1 Excitement about generative AI remains high,and transformative impacts are expected in the next three years.

2、2 Many leaders are confident about their organizations generative AI expertise.3 Organizations that report very high expertise in generative AI tend to feel more positive about itbut also more pressured and threatened.4 Current generative AI efforts remain more focused on efficiency,productivity and

3、 cost reduction than on innovation and growth.5 Most organizations are still primarily relying on off-the-shelf generative AI solutions.6 Talent,governance and risk are critical areas where generative AI preparedness is lacking.7 Leaders see significant societal impacts on the horizon.8 Leaders are

4、looking for more regulation and collaboration globally.Next:Looking aheadAuthorship&AcknowledgmentsAbout the Deloitte AI Institute About the Deloitte Center for Integrated Research About the Deloitte Center for Technology,Media&TelecommunicationsMethodologyTable of contents2 2ForewordNow decides nex

5、tThe arrival of generative AI heralds disruption and opportunity across industries.Organizations are exploring how generative AI can be used to unlock business value,supercharge efficiency and productivity,and open the door to entirely new products,services and business models.As business leaders co

6、ntend with this new technology and make decisions about the future of the enterprise with generative AI,it is helpful to keep ones finger on the pulse of adoption.To that end,The State of Generative AI in the Enterprise:Now decides next,captures the sentiments of 2,835 business and technology leader

7、s involved in piloting or implementing generative AI in their organizations.In this inaugural release of the quarterly report series,leaders indicated persistent excitement for using generative AI and many expect substantial transformative impacts in the short term.Yet,they also acknowledged uncerta

8、inty about generative AIs potential implications on workforces and society as the technology is widely scaled,calling for greater investment in talent,governance and global collaboration.From these wave one insights,we can gain a clearer picture of how leaders are using generative AI,challenges,and

9、lessons learned thus far.This helps reveal some of the essential questions leaders should be asking now and actions they should be taking to prepare their enterprise for what comes next.There is still much to discover with generative AI.As it matures and is deployed at scale for a litany of applicat

10、ions,new questions and challenges will become clearer.Our quarterly reports will be available to help you make sense of this fast-moving space,consider practical guidance based on what we have learned,and take a forward-looking view in your business future with generative AI.Learn more about the ser

11、ies and sign up for updates at Dutt,Beena Ammanath,Costi Perricos and Brenna Sniderman3Will generative AI(gen AI)be the greatest,most impactful technology innovation in history?Will it completely transform how humans live and work?Or will it turn out to be just another technology du jour that promis

12、ed revolutionary change but ultimately delivered only incremental improvement?Right now,we cant be certain.What we do know is that many breakthrough technologies of the past have followed a common adoption pattern:initial awareness;excitement that led to hype;mild disappointment as hype met reality;

13、and then explosive growth once the technology reached critical mass and proved its worth.Generative AI seems to be following the same pattern,only much,much faster.ChatGPT was publicly released on November 30,2022,largely as a technology demonstration.Two months later,it had already attracted an est

14、imated 100 million active usersmaking it the fastest-growing consumer application in history.1 Since then,generative AI has continued to advance by leaps and bounds and many new tools and use cases have emergedproviding a powerful glimpse at the technologys vast potential to transform how people liv

15、e and work.Now decides next:Insights from the leading edge of generative AI adoptionIntroduction4During this frenzied period of generative AI advancement and adoption,leaders in business,technology and the public sector are under tremendous pressure to understand generative AIand to figure out how t

16、o harness its capabilities most effectively(or at least avoid being disrupted).They also sense that now decides next;that their decisions and actions today will significantly affect how generative AI unfolds in the future,for better or worse.Its been said that people tend to overestimate the effect

17、of a technology in the short run and underestimate its effect in the long run.This phenomenon has occurred many times in the past and could very well happen again with generative AI.Note here that given generative AIs dizzying pace of change,the gap between the short run and long run might be measur

18、ed in days,weeks or monthsnot years or decades.To help make smart decisions,leaders need objective,timely information about current generative AI developmentsand where things are headed.Which is why Deloitte is conducting this ongoing quarterly survey.Our goal is to take the pulse of generative AI a

19、doption,offer a view of whats happening,track evolving attitudes and activities,and deliver practical,actionable insights that can help leaders like you make informed and confident decisions about AI,strategy,investment and deployment.In this report,we examine our first quarterly survey findings in

20、detail,supported by insights from Deloittes AI-related work with organizations across every major industry and many geographic regions.We also offer a forward-looking view to help you decide what generative AI actions may make sense for your own organization and situation.To help leaders in business

21、,technology and the public sector track the rapid pace of generative AI change and adoption,Deloitte is conducting a series of quarterly surveys.The series is based on Deloittes State of AI in the Enterprise reports,which have been released annually five years running.The wave one survey was fielded

22、 to more than 2,800 director-to C-suite-level respondents across six industries and 16 countries between October and December 2023.Industries included:Consumer;Energy,Resources&Industrials;Financial Services;Life Sciences&Health Care;Technology,Media&Telecom;and Government&Public Services.Learn more

23、 at The State of Generative AI in the EnterpriseIntroductionAll statistics noted in this report and its graphics are derived from Deloittes first quarterly survey,conducted October December 2023;The State of Generative AI in the Enterprise:Now decides next,a report series.N(Total leader survey respo

24、nses)=2,835Generative AI is an area of artificial intelligence and refers to AI that in response to a query can create text,images,video and other assets.Generative AI systems can interact with humans and are often built using large language models(LLMs).Also referred to as“gen AI.”Insights from the

25、 leading edge(cont.)5This first pulse of our generative AI quarterly surveys was completed in December 2023,and included more than 2,800 AI-savvy business and technology leaders directly involved in piloting or implementing gen AI at major organizations around the world.Heres what they had to say ab

26、out sentiment,use cases,challenges and more.Now:Key findings6 6Now:Key findingsNearly two-thirds(62%)of the business and technology leaders surveyed reported excitement as a top sentiment with regard to generative AI;however,that excitement was tinged with uncertainty(30%)(figure 1).The vast majorit

27、y of respondents(79%)said they expect generative AI to drive substantial transformation within their organization and industry over the next three yearswith nearly a third expecting substantial transformation to occur now(14%)or in less than one year(17%)(figure 2).The survey results suggest that ma

28、ny AI-fueled organizations are on the verge of scaling up their efforts and embracing generative AI in a more substantial way.This aligns with what were seeing in the marketplace,where organizations around the world are racing to move from experimentation and proofs-of-concept to larger-scale deploy

29、ments across a variety of use cases and data typespursuing both speed and value capture while managing potential downside risks and societal impacts.In future surveys,we will be closely monitoring progress in this areaparticularly with regard to organizations expertise,capabilities,tangible outcomes

30、,and responses to rapidly emerging advances in generative AI technology.31%of the leaders we surveyed expect substantial transformation in less than one year;48%expect it in one to three years.1Generative AI elicits a range of strong emotionsFascination Trust Anxiety Fear AngerUncertaintlySurpriseCo

31、nfusionExhaustion16%10%8%6%4%1%Excitement62%46%30%17%Excitement about generative AI remains high,and transformative impacts are expected in the next three years.Q:Thinking about generative AI,what emotions do you feel most about the technology?(Oct./Dec.2023)N(Total)=2,835Figure 17 7When is generati

32、ve AI likely to transform your organization?Now:Key findingsQ:When is generative AI likely to substantially transform your organization and your industry,if at all?(Oct./Dec.2023)N(Total)=2,835Figure 2In one to three years48%Less than one year17%Now14%Never1%Beyond three years20%8Many leaders are co

33、nfident about their organizations generative AI expertise.Now:Key findingsA large percentage of our survey respondents(44%)said they believe their organizations currently have high(35%)or very high(9%)levels of expertise with generative AI.This result is somewhat surprising given how rapidly generat

34、ive AI is evolving(figure 3).But within the specific context of our survey,high levels of confidence seem entirely reasonable since we deliberately chose experienced leaders with direct involvement in AI initiatives at large organizations already piloting or implementing generative AI solutions.Howe

35、ver,given how rapidly the field is unfolding,it may be worth questioning the extent to which any leader should feel highly confident in their organizations expertise and preparedness.In fact,even todays foremost AI experts who are personally developing generative AI technologies at times seem genuin

36、ely surprised by their own creations capabilities.2Do some leaders consider their organizations to have high expertise based largely on the knowledge and experience gained from small-scale pilots with a small number of generative AI tools?If so,leaders and organizations might actually become less co

37、nfident over time as they gain experience with the larger challenges of deploying generative AI at scale.In other words,the more they know,the more they might realize how much they dont know.This is a trend weve seen time and again with other technological advancements,and one well be watching close

38、ly in our future surveys.Self-assessed expertise with generative AI runs high2Q:How would you assess your organizations current level of overall expertise regarding generative AI?(Oct./Dec.2023)N(Total)=2,835Figure 344%rate their organizations generative AI expertise as high or very high,but is such

39、 expertise even possible given the pace of the technologys advancement?Little expertiseNo expertiseVery high expertiseHigh expertiseSome expertise10%1%9%35%45%9Rank uncertainty among top emotions feltOrganizations that report very high expertise in generative AI tend to feel more positive about itbu

40、t also more pressured and threatened.Now:Key findingsRelative to other respondents,leaders who rated their organizations overall generative AI expertise as“very high”tended to feel much more positive about the technology;however,they also feel more pressure to adopt itand see it as more of a threat

41、to their business and operating models(figure 4).Analysis showed this group using more modalities,deploying generative AI across more enterprise functions,and pursuing more use cases.As you can see in the figure 4,leaders who reported very high levels of expertise were also more likely to report hig

42、her levels of trust and lower levels of uncertainty.They also tended to show broader interest in generative AI and expected faster transformation for their organizations.At the same time,these respondents greater understanding of generative AI appears to be shaping their perspective on potential imp

43、actspositive and negative.Many reported they viewed widespread adoption of the technology as a threat to how their organizations operate and conduct business,amplifying the pressure and urgency they felt to adopt generative AI and scale it.Rank trust among top emotions felt Very high expertiseSome e

44、xpertiseTrust prevails over uncertainty39%9%11%38%78%38%31%9%33%16%44%25%Say employees show high interest in gen AI Broad interest sparks transformationWidespread adoption generates pressureFeel widespread adoption is a threat to business Say gen AI is already transformative Feel greater pressure to

45、 adopt gen AI 3Expertise with generative AI drives attitudes toward adoptionLeaders of organizations with very high expertise are more likely to view generative AI as a threat to their business and operating models.Figure 4(Oct./Dec.2023)N(Total)=2,835,N(Very high)=267;N(Some)=1,273104Current genera

46、tive AI efforts remain more focused on efficiency,productivity and cost reduction than on innovation and growth.Now:Key findingsThe majority of organizations surveyed are currently targeting tactical benefits such as improving efficiency/productivity(56%)and/or reducing costs(35%).Also,91%said they

47、expect generative AI to improve their organizations productivity,and 27%expect productivity to increase significantly.A smaller percentage of organizations reported targeting strategic benefits such as innovation and growth(29%)(figure 5).This is consistent with past technology adoption patterns.Ini

48、tially,most organizations logically focus on incrementally improving their existing processes and capabilitiescapturing value from low-hanging fruit while building knowledge,experience and confidence with the new technology.Later,they expand or shift their focus to improvements that are more innovat

49、ive,strategic and transformationalusing the new technology to drive growth and competitive differentiation and advantage through capabilities that simply werent possible before.Surveyed leaders that cited higher levels of AI expertise show earlier signs of moving up this curve.They are more focused

50、on uncovering new ideas and insights(23%vs.19%for the overall respondent pool),with less emphasis on efficiency and productivity(44%vs.61%for the overall respondent pool)and cost reduction(26%vs.38%for the overall respondent pool)although those tactical benefits continue to be Key benefits organizat

51、ions hope to achieve with generative AI18%Detect fraud and manage risk56%Improve efficiency and productivity35%Reduce costs25%Increase revenue26%Increase speed and/or ease of developing new systems/software19%Uncover new ideas and insights23%Enhance relationships with clients/customersQ:What are the

52、 key benefits you hope to achieve through your generative AI efforts?(Oct./Dec.2023)N(Total)=2,835Figure 529%Improve existing products and services29%Encourage innovation and growth26%Shift workers from lower to higher value tasks11their bigger focus.In addition,nearly three-quarters of organization

53、s that cited very high generative AI expertise had already begun integrating the technology into their product development and R&D activities,which are key drivers of innovation and growth.As more organizations gain expertise and experience with generative AI,will they reinvest their dividends from

54、improving efficiency and productivity toward pursuing more strategic benefits such as innovation and growth?Or will they use those dividends in other ways?This is another area well be monitoring closely in future pulse surveys.Certainly,productivity and efficiency can be transformational,especially

55、given the massive scale generative AI has the potential to enable.However,the greatest value and strategic differentiation will likely come from using the technology to innovate.First,by helping to generate new products,services and capabilities that wouldnt be possible otherwise.And,second,by enabl

56、ing new business models and ways of working across an enterprise.In addition,organizations that cited very high generative AI expertise were already taking a much more comprehensive approach than average,with significantly higher adoption levels across a broad range of functional areas.In specific a

57、reas such as HR,and legal,risk and compliance,those organizations generative AI adoption rates were nearly three times higher than for the total respondent pool(figure 6).Now:Key findings91%of all organizations expect their productivity to increase due to generative AI.12%of those who are using gene

58、rative AI in a limited or at-scale implementationLevel of generative AI adoptionTotalLittle expertiseSome expertiseHigh expertiseVery high expertise71%73%73%62%61%5%9%10%10%6%7%63%64%60%57%50%57%46%47%37%41%38%34%41%37%35%29%29%28%25%23%21%21%22%16%14%14%13%28%26%Marketing,sales and customer service

59、Product development/R&DStrategy and operationsSupply chain/manufacturingFinanceHuman resourcesLegal,risk and complianceIT/cybersecurity Now:Key findingsFigure 6Q:What is your organizations current adoption level of generative AI across the following functions?(Oct./Dec.2023)N(Total)=2,835;N(Very hig

60、h)=267;N(High)=1,003;N(Some)=1,273;N(Little)=2741313The term“unprecedented”is often thrown around when talking about business and technology,to the point of being clich.However,in describing the pace of generative AIs emergence and advancementand its massive potential impact on business(and humanity

61、 as a whole)unprecedented could be an understatement.Generative AI is already widely available to the public and has a running start toward critical mass.Also,similar to smartphones,its easy for an average person to use without much trainingand can help with activities they already engage in every d

62、ayso the barriers to adoption are low.Whats more,generative AI has the strong potential to assist with its own future development,which could trigger a cycle of exponential improvement at exponential speed.Generative AIs speed factor may give organizations less time to ruminate or dabble with small-

63、scale pilotswhile reducing the margin for errorand increasing the consequences of inaction.It also creates opportunities to generate extraordinary business value very quickly.Despite generative AI s greatly accelerated pace,understanding typical adoption patterns based on previous breakthrough techn

64、ologies can provide valuable lessons that leaders can use to help them understand and fully capitalize on the technologys rapid advancement.As in the past,organizations initial efforts will likely center around efficiency,productivity,cost savings and other incremental improvements.This is expected

65、to help the workforce get accustomed to using generative AI,and will show people how it can help make their jobs easier.Also,early wins will likely help produce cost savings and momentum that then can be channeled into higher value opportunities that are more strategic and differentiated in nature,s

66、uch as enabling new products,services,business models and ways of working that simply werent possible before generative AI.Generative AI:Have we seen this movie before?145Most organizations are primarily relying on off-the-shelf generative AI solutions.Now:Key findingsIn line with their current emph

67、asis on tactical benefits from generative AI,the vast majority of respondents were currently relying on off-the-shelf solutions.These included productivity applications with integrated generative AI(71%);enterprise platforms with integrated generative AI(61%);standard generative AI applications(68%)

68、;and publicly available large language models(LLMs)(56%),such as ChatGPT.Relatively few reported using more narrowly focused and differentiated generative AI solutions,such as industry-specific software applications(23%),private LLMs(32%),and/or open-source LLMs(customized to their business)(25%).Re

69、liance on standard,off-the-shelf solutions is consistent with the current early phase of generative AI adoption,which is primarily focused on improving the efficiency and productivity of existing activities.However,as use cases for generative AI become more specialized,differentiated and strategic,t

70、he associated development approaches and technology infrastructure will likely follow suit.When will we see complex,high-value use cases that are truly differentiated and tailored to the specialized needs of specific companies,functions and industries?How will organizations combine internal and exte

71、rnal resources to create customized generative AI tools that enable such strategic differentiation?In particular,will we see off-the-shelf technology offerings be supplemented by private or hybrid public/private development approaches and technology infrastructures capable of delivering and supporti

72、ng those differentiated solutions?71%Productivity applications68%Standard applications61%Enterprise platforms56%Public LLMsWhere off-the-shelf generative AI is used most156Talent,governance and risk are critical areas where generative AI preparedness is lacking.Now:Key findingsIn this initial quarte

73、rly survey,41%of leaders reported their organizations were only slightly or not at all prepared to address talent concerns related to generative AI adoption,while 22%considered their organizations highly or very highly prepared.Similarly,41%of leaders reported their organizations were only slightly

74、or not at all prepared to address governance and risk concerns related to generative AI adoption,while 25%considered their organizations highly or very highly prepared(figure 7).Larger percentages of leaders reported high to very high levels of preparedness in technology infrastructure(40%)and strat

75、egy(34%);however,the survey results show there is still significant room for improvement.Generative AI barriers related to risk and governanceWhen it comes to risk and governance,generative AI is definitely not“just another technology.”The fundamental challenge is how to capitalize on artificial int

76、elligences power without losing control of it.After all,the capability people seem to find most enthralling about generative AI is its ability to so convincingly simulate human thinking and behavior.Of course,human thinking and behavior arent always perfect,predictable or socially acceptableand the

77、same is true for the technology,itself.16Now:Key findingsPreparedness for generative AIFigure 7Q:Consider the following areas.For each,rate your organizations level of preparedness with respect to broadly adopting generative AI tools/applications?(Oct./Dec.2023)N(Total)=2,835Respondents claimed the

78、highest levels of preparation in technology and strategy,while feeling far less prepared in risk and talent.Not preparedSlightly preparedModerately preparedHighly preparedVery highly preparedTalent13%28%37%17%5%Risk&governance13%28%34%18%7%Strategy5%20%41%26%8%Technology infrastructure4%17%38%30%10%

79、17Specific generative AI risks and concerns include inaccurate results and information(i.e.,“hallucinations”);legal risks such as plagiarism,copyright infringement,and liability for errors;privacy and data ownership challenges;lack of transparency,explainability and accountability;and systemic bias.

80、The latter exemplifies another category of risk in which AI amplifies and exacerbates a problem that already exists,such as propagating and systematizing existing social biases,facilitating and accelerating the spread of misinformation,helping criminals commit crimes,or fanning the flames of politic

81、al divisiveness.According to the business and technology leaders we surveyed during fourth quarter 2023,the biggest concerns related to governance were:lack of confidence in results(36%),intellectual property issues(35%),misuse of client or customer data(34%),ability to comply with regulations(33%),

82、and lack of explainability/transparency(31%).Some of the surveyed organizations were already actively managing generative AI implementation risks through actions such as monitoring regulatory requirements and ensuring compliance(47%),establishing a governance framework for generative AI(46%),and con

83、ducting internal audits and testing on generative AI tools and applications(42%)(figure 8).However,such organizations are in the minority and their actions barely scratch the surface of the challenge.This is especially true given that regulatory requirements typically lag behind the pace of technolo

84、gy innovationalthough a US presidential executive order and the European Unions ambitious Artificial Intelligence Act are clear signs government leaders in many parts of the world are taking the issue of AI risk very seriously.Now:Key findingsManaging generative AI implementation riskFigure 8Q:What

85、is your organization currently doing to actively manage the risks around your generative AI implementations?(Oct./Dec.2023)N(Total)=2,83532%26%21%Single executive responsible for managing generative AI-related risksKeeping a formal inventory of all generative AI implementationsUsing outside vendors

86、to conduct independent audits and testing34%Using a formal group or board to advise on generative AI-related risks36%Ensuring a human validates all generative AI content47%Monitoring regulatory requirements and ensuring compliance46%Establishing a governance framework for the use of generative AI to

87、ols/applications42%Conducting internal audits and testing on generative AI tools/applications37%Training practitioners how to recognize and mitigate potential risks18Now:Key findingsGenerative AI barriers related to talent and workforceGenerative AI has the potential to supplement human workers acro

88、ss a vast array of activities traditionally thought of as uniquely human.As such,its impact on talent and workforce strategies could be immense.How will it affect organizations and their workers in the short and long runs?Which types of skills will be most affected,and when?The vast majority of lead

89、ers we surveyed(72%)said they expect generative AI to drive changes in their talent strategies sometime within the next two years:now(17%),within 1 year(24%),or in 1-2 years(31%)(figure 9).However,less than half(47%)reported that they are sufficiently educating their employees on the capabilities,be

90、nefits and value of generative AIsurvey respondents also cited a lack of technical talent and skills as the biggest barriers to adoption.Generative AI is impacting talent strategies nowFigure 9Q:When do you expect to make changes to your talent strategies because of generative AI?(Oct./Dec.2023)N(To

91、tal)=2,8351-2 yearsNo formal time frame2+yearsNowNeverWithin 1 year31%10%16%17%2%24%19Figure 10Preparing workforces for generative AI:Respondents making a high or very high effort in the following areas.Q:What level of effort is your organization taking regarding the following workforce-related area

92、s?(Oct./Dec.2023)N(Total)=2,835Against this backdrop,some respondents reported making a high or very high effort to:recruit and hire technical talent to drive their generative AI initiatives(42%),educate the workforce about generative AI(40%),and reskill workers impacted by generative AI(36%).Those

93、numbers are much higher for leaders who viewed their organizations generative AI expertise as very high(74%,74%and 73%,respectively)(figure 10).It should be noted,however,that these reported workforce-related efforts might be limited in scope.Deloittes experience suggests that most organizations hav

94、e yet to substantially address the talent and workforce challenges likely to arise from large-scale generative AI adoption.A likely reason for this is that many leaders dont yet know what generative AIs talent impacts will be,particularly with regard to which skills and roles will be needed most.Rec

95、ruiting and hiring technical talent to drive our generative AI initiativesAll respondantsLittle expertiseSome expertiseHigh expertiseVery high expertiseEducating our broader workforce to raise overall generative AI fluencyReskilling workers because of the impact of generative AI to their roles74%74%

96、73%55%55%50%30%27%24%16%14%10%42%40%36%Now:Key findings20“Generating confidence in workers abilities to collaborate with generative AI,now,could elevate creativity and job satisfaction,next.”217Leaders see significant societal impacts on the horizon.Now:Key findingsAlthough the leaders we surveyed w

97、ere generally excited and enthusiastic about generative AIs potential business benefits,they were less optimistic about its broader societal impacts.Specifically,52%of respondents said they expected widespread use of generative AI to centralize power in the global economy,while 30%expected it to mor

98、e evenly distribute global power.Similarly,51%expected generative AI to increase economic inequality,while 22%expected it to reduce inequality(figure 11).Whats more,49%of respondents believe the rise of generative AI tools/applications will erode the overall level of trust in national and global ins

99、titutions.Is this pessimism or realism?Our survey results appear to reflect the broader moral and ethical debates about artificial intelligence that are occurring in every corner of societyeven in the boardrooms of the technology companies driving AI development,where AIs commercial value is being w

100、eighed against its potential value to serve humanity and AIs potential benefits are being weighed against its potential risks.The challenges that generative AI poses in corporate governance and risk parallel those in societal governance and risk.In both domains,the technologys potential benefits and

101、 potential harms are high.National and supranational organizations and governments will likely need to walk the tightrope of helping to ensure that generative AI benefits are broadly and fairly distributed,without overly hindering innovation or providing an unfair advantage to countries with differe

102、nt rules.51%expect generative AI to increase economic inequality.22Distribution of economic powerLevels of economic inequality30%distribute52%centralize51%increase inequality22%decrease inequality5%3%19%27%41%4%25%18%42%10%10%Expected societal impacts of generative AINow:Key findingsFigure 11Q:How w

103、ill widespread use of generative AI shift the overall distribution of power in the global economy?Q:How will widespread use of generative AI tools/applications impact global levels of economic inequality?(Oct./Dec.2023)N(Total)=2,835238Leaders are looking for more regulation and collaboration global

104、ly.Now:Key findingsIn a break from traditional business norms,the unique risks associated with generative AI are prompting many business leaders to call for increased government regulation and increased global collaboration around AI technologies.Among the leaders in our survey,78%said that more gov

105、ernmental regulation of AI is needed,while 72%said there is currently not enough global collaboration to ensure the responsible development of AI-powered systems(figure 12).These results seem to reflect an understanding that generative AI could be too powerful,far-reaching and impactful for individu

106、al organizations to regulate themselves.This isnt meant to suggest that individual organizations be absolved from behaving responsibly;however,relying on them to be the primary gatekeepers for containing AI risk could potentially be dangerous.Figure 12Support for increased regulation and global coll

107、aboration(Oct./Dec.2023)N(Total)=2,835Could generative AI be too powerful and far-reaching for individual organizations to regulate themselves?78%more regulationAgree the widespread proliferation of generative AI tools/applications will require more regulation of AI by governments72%more collaborati

108、onAgree there is not enough global collaboration with respect to ensuring the responsible development of all AI-powered systems24As the first in an ongoing series of quarterly surveys to track the pulse of generative AI,this initial effort was designed to establish a strong baseline to build on.Movi

109、ng forward,our objective is to understand how generative AI adoption is unfoldingand to anticipate where its headed.Throughout this report we posed a number of strategic questions to help organizations think critically about how the actions they take now will best set the stage for what comes next.W

110、e dont have definitive answers to every question yet,nor would we claim to.However,we can offer some questions to spark thinking and some practical guidance based on what weve learned so far.We also note that given the pace at which AI and specifically generative AI is moving,definitive answers that

111、 make sense today may not be relevant in a few months.Next:Looking ahead2525In the race to deploy generative AI solutions,organizational attributes such as adaptation,experimentation and agility will be critical as new models,capabilities and use cases emerge.The key is to maintain a beginners minds

112、etthe belief that no matter how expert you think you are,there will always be much more to learneven as your experience grows.Careful coordination across your organization will be needed to successfully shepherd generative AI transformation in the face of rapid change.Work to improve generative AI l

113、iteracy throughout your organization,and lead using a cross-disciplinary approach.Actively collaborate with partners and third-party organizations.Also,gain experience with a variety of generative AI technologieswith innovation happening so quickly,its nearly impossible to pick a clear winner today.

114、How can my organization build generative AI expertise when things are moving so quickly?Next:Looking ahead26Generative AI projects can potentially be expensive,so leaders are naturally looking for rapid ways to achieve a compelling return on investment(ROI).Benefits might at first arrive slowly,but

115、then ramp up quickly as your organization reaches a critical mass of experience and proficiency.Most generative AI efforts are currently focused on improving efficiency and productivity,and reducing costs.Once you achieve those goals,what will you do with the time and money saved?Will you strategica

116、lly reinvest in more generative AI projects?Will you invest in training and reskilling your workforce?Will you improve your technology infrastructure?Or will you simply boost your bottom line?A deliberate reinvestment strategy for generative AI dividends will help lay the path for continued success.

117、How should we invest the dividends gained from improving efficiency and productivity?Experimentation is essential when deploying generative AI.But if you cant scale up your efforts,the high expectations for transformation revealed by our survey likely wont be met.Its fine to focus on a few use cases

118、 at first.However,the most valuable use cases will likely change over time,so its important to focus on improving end-to-end processes,not just narrow tasks.Also,follow the example of organizations that report high generative AI expertise and consider deploying AI broadly across your enterprise as p

119、art of a holistic strategy,rather than focusing narrowly on point solutions and silos.Strive to build platform capabilities that can enable multiple use cases,accommodate new and improved generative AI models,and provide consistent governance and risk management to ensure models produce safe and tru

120、stworthy outputs and content.How can we best scale up and build a foundation for longer-term value creation?Next:Looking ahead27Where companies reinvest the gains they make in efficiency from generative AI,now,may decide their level of transformation,next.28When developing and deploying generative A

121、I solutions,should you buy or build?The answer depends on many factors,including your overall goals and the scale,complexity and uniqueness of your solution and use case.Are you looking to monetize your model?What is your approach to open source?How much control over training datasets do you want?Qu

122、estions like these will help you choose from the broad spectrum of approaches,which include:building large language models or LLMs from scratch,fine-tuning vendor-provided models with your own data,or using enterprise software with generative AI built in.Each approach has its benefits and drawbacks,

123、and you might end up choosing more than one.When deciding,be sure to consider your business strategy,desired investment level,risk tolerance and data readiness.How should we balance buying vs.building?As generative AI adoption rises and the technology becomes a standard commoditywith increased integ

124、ration into common enterprise software,broader availability of specialized tools and models,and standardized data requirementswill first movers lose their advantage?To maximize the value of the technology,organizations should consciously focus on innovation and differentiationcustomizing their gener

125、ative AI solutions to fit their unique needs and data assets,with the goal of building capabilities that create sustainable competitive advantage.Pursuing easy opportunities and quick wins is smart,but not to the exclusion of more strategic opportunities(even though the latter will require more time

126、 and money to achieve and may take longer to achieve ROI).How can our organization use generative AI to create strategic differentiation and a competitive edge?Next:Looking ahead29Respondents expressed a variety of concerns about generative AI risks,including the need to manage hallucinations and mo

127、del bias,assess potential intellectual property issues,and ensure transparency and explainability.These issues underscore the importance of keeping humans in the loop to work with AI,check its accuracy,and address any problems that arise.Additionally,there are open questions about how various regula

128、tory and legal challenges will affect development of the overall market.A large percentage of organizations(47%)reported they were monitoring regulatory requirements as part of their risk management efforts.Many respondents noted their concerns that the widespread use of generative AI will concentra

129、te power and increase economic disparity.As a leader,you will need to consider how your organizations generative AI decisions and actions fit into the larger pictureand will likely need to do so prospectively,instead of waiting for official guidance from lawmakers and regulators.What guardrails does

130、 our organization need to ensure responsible use of generative AI,and how do we stay aligned with shifting societal guardrails?Survey respondents cited talent as their biggest barrier to generative AI adoption.To surmount it,you will likely need to recruit new talent,empower your existing workforce,

131、and build organizational trust.Although the talent market is highly competitive,dont let that deter you from pursuing people with the technical skills to develop and maintain generative AI solutions(e.g.,prompt engineers,AI solutions architects,data scientists/engineers,LLM operators).At the same ti

132、me,invest in training to help your people get the maximum value from generative AI tools and to improve their productivity.Also,prioritize broad workforce education to help allay fears and misconceptions about AI technology.How can we best invest in our people and reinvent how they work with generat

133、ive AI?Next:Looking ahead30Authorship and AcknowledgmentsDeborshi Dutt US AI Strategic Growth Offering Leader Deloitte Consulting LLP Beena Ammanath Global Deloitte AI Institute Executive Director Deloitte LLP Costi Perricos Global Office of Generative AI Leader Deloitte UK cperricosdeloitte.co.ukBr

134、enna Sniderman Deloitte Center for Integrated Research Executive Director Deloitte Services LLP Acknowledgments The authors would like to thank the many talented professionals who brought this research to life:Joe Ucuzoglu,Nitin Mittal,Kevin Westcott,Lynne Sterrett,Rod Sides,Dina Tallarico,David Jar

135、vis,Jeff Loucks,Ahmed Alibage,Natasha Buckley,Jonathan Holdowsky,Siri Anderson,David Levin,Joe Mariani,Sandeep Vellanki,Rajesh Medisetti,Shubham Oza,Gerson Lehrman Group(GLG),Lena La,Kate Schmidt,Ivana Vucenovic,Sharonjeet Meht,Bryan Furman,Lesley Stephen,Stephanie Anderson,Steve Dutton,Justin Joyne

136、r,Jordan Garrick,Karen Hogger,Matt Lennert,Maria Fernanda Castro,Tracy Fulham,Jose Porras,Jonathan Pryce,Sourabh Yaduvanshi,Jessi Hendon,Jamie Palmeroni-Lavis,Melissa Neumann,Tatum Hoehn,Sean Benton,Eric Alons-Cruz,Lancy Jiang,Amber Bushnell,Brandon Gomez,Judy Mills,Marianne Wilkinson,Lou Ghaddar,Li

137、sa Iliff and Michael Lim.We would also like to thank additional Deloitte subject matter specialists who contributed to the development of the survey and report:Rohit Tandon,Mike Segala,Bjoern Bringmann,Kellie Nuttal,Ed Bowen,Oz Karan,Lou DiLorenzo,Ed Van Buren,Amelia Dunlop,Ashley Reichheld,Maggie F

138、letcher,Elizabeth Powers,Baris Sarer,Dany Rifkin and Laura Shact.31About the Deloitte AI Institute The Deloitte AI Institute helps organizations connect all the different dimensions of the robust,highly dynamic and rapidly evolving AI ecosystem.The AI Institute leads conversations on applied AI inno

139、vation across industries,using cutting-edge insights to promote human-machine collaboration in the Age of With.The Deloitte AI Institute aims to promote dialogue about and development of artificial intelligence,stimulate innovation,and examine challenges to AI implementation and ways to address them

140、.The AI Institute collaborates with an ecosystem composed of academic research groups,startups,entrepreneurs,innovators,mature AI product leaders and AI visionaries to explore key areas of artificial intelligence including risks,policies,ethics,future of work and talent,and applied AI use cases.Comb

141、ined with Deloittes deep knowledge and experience in artificial intelligence applications,the institute helps make sense of this complex ecosystem and,as a result,delivers impactful perspectives to help organizations succeed by making informed AI decisions.About the Deloitte Center for Integrated Re

142、searchThe Deloitte Center for Integrated Research(CIR)offers rigorously researched and data-driven perspectives on critical issues affecting businesses today.We sit at the center of Deloittes industry and functional expertise,combining the leading insights from across our firm to help leaders confid

143、ently compete in todays ever-changing marketplace.About the Deloitte Center for Technology,Media&TelecommunicationsThe Deloitte Center for Technology,Media&Telecommunications(TMT Center)is a world-class research organization that serves Deloittes TMT practice and our clients.Our team of professional

144、 researchers produce practical foresight,fresh insights,and trustworthy data to help clients see clearly,act decisively,and compete with confidence.We create original research using a combination of rigorous methodologies and deep TMT industry knowledge.Learn moreLearn more32To obtain a global view

145、of how generative AI is being adopted by organizations on the leading edge of AI,Deloitte surveyed 2,835 leaders between October and December 2023.Respondents were senior leaders in their organization and included board and C-suite members,and those at the president,vice president and director level

146、.The survey sample was split equally between IT and line of business leaders.Sixteen countries were represented:Australia(100 respondents),Brazil(115 respondents),Canada(175 respondents),France(130 respondents),Germany(150 respondents),India(200 respondents),Italy(50 respondents),Japan(100 responden

147、ts),Korea(11 respondents),Mexico(101 respondents),Netherlands(75 respondents),Singapore(76 respondents),Spain(101 respondents),Switzerland(50 respondents),the United Kingdom(200 respondents),and the United States(1,201 respondents).All participating organizations have one or more working implementat

148、ions of AI being used daily.Plus,they have pilots in place to explore generative AI or have one or more working implementations of generative AI being used daily.Respondents were required to meet one of the following criteria with respect to their organizations AI and data science strategy,investmen

149、ts,implementation approach,and value measurement.They:influence decision-making,are part of a team that makes decisions,are the final decision-maker,or manage or oversee AI technology implementations.All statistics noted in this report and its graphics are derived from Deloittes first quarterly surv

150、ey,conducted October December 2023;The State of Generative AI in the Enterprise:Now decides next,a report series.N(Total leader survey responses)=2,835Methodology33About DeloitteDeloitte refers to one or more of Deloitte Touche Tohmatsu Limited,a UK private company limited by guarantee,and its netwo

151、rk of member firms,each of which is a legally separate and independent entity.Please see for a detailed description of the legal structure of Deloitte Touche Tohmatsu Limited and its member firms.Please see for a detailed description of the legal structure of Deloitte LLP and its subsidiaries.Certai

152、n services may not be available to attest clients under the rules and regulations of public accounting.This publication contains general information only and Deloitte is not,by means of this publication,rendering accounting,business,financial,investment,legal,tax,or other professional advice or serv

153、ices.This publication is not a substitute for such professional advice or services,nor should it be used as a basis for any decision or action that may affect your business.Before making any decision or taking any action that may affect your business,you should consult a qualified professional advis

154、or.Deloitte shall not be responsible for any loss sustained by any person who relies on this publication.Copyright 2024 Deloitte Development LLC.All rights reserved.Endnotes:1.Krystal Hu,“ChatGPT sets record for fastest-growing user base analyst note,”Reuters,February 2,2023,https:/ January 3,2023.2.Will Douglas Heaven,“Geoffrey Hinton tells us why hes now scared of the tech he helped build,”MIT Technology Review,May 2,2023,https:/ January 3,2023.