《Lucidworks：2024全球企業生成式AI應用現狀調研報告：區分炒作與現實（英文版）（16頁）.pdf》由會員分享，可在線閱讀，更多相關《Lucidworks：2024全球企業生成式AI應用現狀調研報告：區分炒作與現實（英文版）（16頁）.pdf（16頁珍藏版）》請在三個皮匠報告上搜索。

1、The second annual report of the largest global generative AI studyprimary research conducted by Lucidworks,a leader in search and total AI solutionsGenerative AI Global Benchmark Study Vol.22024 STATE OF GENERATIVE AI IN GLOBAL BUSINESSGenerative AI continues to drive massive change around the world

2、and its only the beginning.In 2023,Lucidworks conducted the largest global study to date to understand key areas of generative AI investment and how advanced organizations are in adopting this technology.In 2024,global business leaders remain eager to implement their plans and stay ahead of the comp

3、etition as the technology transforms from hype to reality.As the path to AI success emerges,companies worldwide are enthusiastic about the possible impact on customer experience,automation,and business operations.However,the study found that spending has slowed slightly as businesses more carefully

4、consider the most valuable areas to invest in.A third of business leaders feel like theyre falling behind competitors despite almost everyone struggling to implement this new transformational technology.One of the biggest roadblocks to deploying AI initiatives is a significant uptick in concerns aro

5、und implementation costs.Additionally,data security concerns have more than doubled from last year and concerns around response accuracy have grown by five times.As the hype slowly turns to reality,the race is on to maximize the potential value of generative AI while balancing cost and security conc

6、erns.This report offers a roadmap with best practices to help companies plan and execute their generative AI roadmap.EXECUTIVE SUMMARY2024 State of Generative AI in Global Business WHAT IS GENERATIVE AI?The conversation around generative AI has moved beyond“what could be possible”in 2023.Now,in 2024

7、,business leaders expect it to deliver real value.Unlike traditional AI which excels at analyzing existing data,generative models can learn from vast datasets and use that knowledge to produce novel outputs.Advancements in multimodal AI have allowed generative models to process text,images,and audio

8、,leading to even richer content creation.This technology holds immense potential for streamlining workflows,boosting innovation,and unlocking new ways of working across industries worldwide.2When the inaugural benchmark study was conducted in 2023,there was no standard for generative AI adoption.Bus

9、iness leaders were scrambling to gain a competitive edge by incorporating this new technology into their processes and digital experiencespossibly without a real plan for measuring its true impact.A year later,theres been trial and error as organizations identify the most well-suited use cases and d

10、iscover the most effective and secure way to deploy generative AI.As Gen AI becomes more powerful,trial and error wont survive as a strategy.The question from 2023s study still stands:How do executives know what to do next,why,and how do they compare to others in the race to adopt generative AI?Luci

11、dworks executed the largest global study of Gen AI practices for the second year.The study was conducted between April and May 2024with over 2,500 participants.All participants were affiliated with organizations actively pursuing generative AI initiatives and were involved in decision-making,impleme

12、ntation,and use of generative AI tools.The study included participants from over 1,000 companies with 100 or more employees,14 industries,and nine functional departments.Roughly 23%of the research participants are executives,50%are managers,and 86%are involved in technology decision-making.39%of res

13、pondents are in North America,36%are in EMEA,and 24%are in the APAC region.The study evaluated 80 generative AI best practices across five categories and identified four stages of generative AI development.The research findings support probable generative AI next steps specific to various industries

14、 and provide valuable insights for companies looking to advance their generative AI initiatives.2024 State of Generative AI in Global Business The Need for an Ongoing Global,Generative AI Benchmark SECOND STUDY OF GENERATIVE AI INITIATIVES23%executives&50%managers2,500+globalAI practitioners and sta

15、keholders1,000+companies with active AI initiatives33 continents6 countries14 industries9 functional departments42024 State of Generative AI in Global Business KEY FINDINGS1 Flattened Spending Points to More Thoughtful Planning2 Deployment Delays Stall Anticipated ROI3 Implementation Costs Raise Ala

16、rms4 Practicality Drives Generative AI Adoption5 Leaders Invest in Future-Proof AI Initiatives 4The generative AI landscape has advanced exponentially in the past year.However,applying the technology has presented some unexpected challenges,including dangerous hallucinations and incredibly high impl

17、ementation costs.In 2024,business leaders are slowing down spending to balance the benefits,costs,and risks of this relatively new technology.5of global companies plan to increase AI spending in the next 12 months,compared to 93%last year Only 63%Planned spending flattened this year,with only 63%of

18、global companies planning to increase AI spending in the next 12 months,compared to 93%last year.The USA leads this list at 69%.Only 49%of Chinese leaders plan to increase AI spending in 2024,a massive drop from 100%in 2023.Across all organizations,36%of leaders plan to keep spending flat,compared t

19、o only 6%in last years survey.Despite the drop,these still sizeable investments demonstrate the necessity of adoption to remain competitive.2024 State of Generative AI in Global Business Flattened Spending Points to More Thoughtful Planning1AI 12-Month Spending Plans20232024IncreaseSpend:93%Decrease

20、 Spend:1%Flat:6%IncreaseSpend:63%Decrease Spend:2%Flat:36%The hype slows as organizations ground applications in reality There is some variability in planned spending across industries.Financial Services moved to the front of the pack,with 70%of organizations planning to increase spending over the n

21、ext 12 months,followed closely by B2B at 68%and B2C retail at 64%.Healthcare and hospitality are laggards in 2024,with only 51%and 50%of businesses planning to increase their investment in the coming months.Manufacturing leaders have also decided to be more conservative.Their planned spending is dow

22、n to 58%from 93%in 2023.2024 State of Generative AI in Global Business 6Spending Plans Vary By IndustryPlans to Increase AI Spend in Next 12 Months50%42%70%68%63%63%63%58%64%51%56%54%Financial ServicesB2BB2C RetailEntertainment&MediaProfessional ServicesOVERALLF&B,CPGEnergyTransportHealth CareHospit

23、alityReal Estate“The initial wave of enthusiasm for generative AI is being met with a more strategic approach.Businesses are recognizing the potential of this technology,but theyre also cautious about the risks and costs.This is reflected in the flattened spending,which suggests a shift toward more

24、thoughtful planning.This planning ensures that AI adoption delivers real value,balancing the need to stay competitive with managing costs and potential risks.”Mike Sinoway,CEO,Lucidworks 6In 2023,global leaders expected to see significant positive impacts across business operations,automation and ef

25、ficiency,and customer experience with generative AI initiatives.The survey examined financial benefits profit and revenue and soft benefits customer satisfaction,employee satisfaction and competitive position.Unfortunately,the financial benefits of implemented projects have been dismal.42%of compani

26、es have yet to see a significant benefit from their generative AI initiatives.The report highlights a key reason for the slowdown in generative AI adoption:the low success rate of getting projects beyond the initial stages.Companies were asked how many initiatives progressed past pilot programs or p

27、roof-of-concept and became fully operational.The findings reveal that only 25%of planned generative AI investments have been fully implemented so far.Tech and retail companies report the greatest success.The tech sector leads with the most deployed initiatives across all industries and half have alr

28、eady realized financial benefits,despite an above average level of delays.Retail is a close second and has achieved the highest deployment of revenue and growth initiatives.1 in 4 Success Rate of Planned and Launched Gen AI Initiatives 72024 State of Generative AI in Global Business of companieshave

29、 yet to see asignificant benefitfrom generative AIinitiatives 42%ONLYDeployment Delays Stall Anticipated ROI2Half of leaders say theyre getting few benefits from Gen AI2024 State of Generative AI in Global Business Despite all the spending,20%of companies report significant delays during deployment

30、with only one in eight planned revenue and growth initiatives and one in six OpEx cost reduction initiatives implemented.Organizations are experiencing a significant learning curve in launching initiatives and companies are twice as likely as they were last year to feel they are falling behind their

31、 peers.More than 85%of business leaders feel theyre behind or only on par with their competition.“The 2023 survey showcased some impressive results:efficiency gains,happier customers,and a solid competitive edge.However,only 25%of Gen AI initiatives have actually reached full deployment.While the te

32、ch and retail sectors are leading the way,most others are struggling with delays and a steep learning curve.The takeaway is clear:succeeding in AI means moving beyond the hype and putting in the hard work.To truly unlock this technologys potential,businesses need to prioritize implementation in real

33、 use cases and push past the pilot phase.Guy Sperry,CTO,Lucidworks 8Success Rate of Gen AI InitiativesOPEX Cost ReductionRevenue&GrowthGen AI Governance1 in 2.51 in 81 in 61 in 61 in 3G&A Cost ReductionService EfficiencyA year ago,data security was the biggest worry,and implementation cost was only

34、a concern for 3%of business leaders.Today,concerns have increased exponentially across the board.The most striking?Concerns about implementation cost increased 14 fold in 2024.Retailers have some of the highest concerns around cost(63%),likely due to the required responsiveness and high number of cu

35、stomer queries.Another significant increase includes a 5x increase in concerns around response accuracy,likely stemming from issues around hallucinations.92024 State of Generative AI in Global Business 20232024Significantly Increasing ConcernsTop Gen AI Concerns 2023 v.20243%43%7%36%17%46%9%35%5%9%2

36、x4x5x14x3xJobDisplacementDataSecurityDecisionTransparencyResponseAccuracyImplementationCostConcerns about implementation cost increased in 202414xImplementation Costs Raise Alarms3High implementation costs have leaders reassessing spend2024 State of Generative AI in Global Business For now,using mor

37、e than one model is the best(and most expensive)way to satisfy the broad needs of commerce,knowledge management,and customer support search queries.Another significant increase includes a 5x increase in concerns around response accuracy,likely stemming from issues around hallucinations.WHY HAVE COST

38、 CONCERNSINCREASED 14X?Businesses arent just using one model to make Gen AI work for themtheyre deploying multiple models to drive performance.The more advanced and aggressive companies are with generative AI initiatives,the more models they employ.A rookie in the space may be using one LLM compared

39、 to two or three LLMs used by their competitors.10Types of LLM Driving Costs Commercial49%Both Commercial&Open Source30%Open Source21%“Security remains a top concern,but a 14x jump in cost worries and a 5x spike in fears about hallucinations paint a new picture of generative AI.Leaders need to under

40、stand requirements around responsiveness,security requirements,and data types to choose the large language models that meet their needs without tacking on unnecessary costs.”Brian Land,VP Sales Engineering,Lucidworks Nearly eight in 10 companies use commercial LLMs such as Gemini and ChatGPT,and 20%

41、have opted for open-source-only such as LLaMA 3 and Mistral.This will likely shift as open source models become more capable and have advantageous features at a lower cost.112024 State of Generative AI in Global Business Practicality Drives Generative AI Adoption4Governance and cost reduction are to

42、p applications of Gen AIDespite the expanding variety of applications for generative AI,global businesses prioritize practical applications of the technology in 2024.Governance and cost reduction are the most successfully deployed generative AI initiatives across all industries.Examples of this incl

43、ude creating defined guidelines around the use of generative AI(governance)and using the technology to help generate first drafts of new computer code(G&A cost reduction).Companieshave deployed of plannedGovernance andCost Reductioninitiatives33%Companies understand the critical need for responsibil

44、ity around data privacy,transparency,and fairness as they adopt new generative AI practices.GOVERNANCEMost Successfully Deployed Governance AI Initiatives:Standard Gen AI tools and models defined to ensure alignment Restricted access to Gen AI tools and data based on role Gen AI guidelines defined a

45、nd distributed to minimize riskToday,with concerns around implementation costs skyrocketing,the need to balance innovation with costs is top of mind for business leaders.GENERAL&ADMINISTRATIVE COST REDUCTION Gen AI for QA testing and debugging code Provide employees with help and FAQs Gen AI generat

46、es first draft of new codeMost Successfully Deployed G&A Cost Reduction AI Initiatives:12122024 State of Generative AI in Global Business The report created a framework to identify patterns in generative AI initiatives.The report organized the most common use cases across four types of applications.

47、Passive:taking the data and passing it along in a different form through summarization or synthesis Active:using generative AI to transform data into insights Qualitative:using text and narrative to provide results or responses Quantitative:adding in numerical tools for applications,including online

48、 personalization and optimizing search resultsQualitative applicationsgenerating FAQs or providing HR support have been the most successful to date with roughly a quarter already implemented.These include some of the less complex applications of generative AI,including retrieving documents and infor

49、mation,extracting the relevant details,summarizing them,and responding to queries.Applications with a quantitative component lag behind,with less than 15%having been successfully implemented.These more complicated applications include optimizing search results,screening candidates for hire,supportin

50、g financial close,and online personalization.The Active/Quantitative quadrant holds some of the greatest potential for generative AI,where the applications are actionable and not simply processing already existing text.QualitativeOnly Synthesize Revise Enhance Generate Retrieve Extract Summarize Res

51、pond Predict Interrogate Activate Iterate Monitor Analyze Prioritize OptimizeQualitative&QuantitativeActivePassive6%13%8%11%“The untapped power of generative AI lies in its ability to quantify qualitative information.By analyzing unstructured signals,Gen AI can reveal hidden patterns and empower bus

52、inesses to take actions based on insights that were previously inaccessible.From optimizing production to predicting customer behavior,quantitative AI applications are the key to unlocking real-world value.”Mike Sinoway,CEO,Lucidworks132024 State of Generative AI in Global Business An LLM alone is n

53、ot a generative AI solution that can drive results.There are dozens of factors that influence decisions around AI initiatives.The first step is to define the use case.Governance and cost reduction are the areas where most businesses prioritize their efforts.Leaders should carefully consider what wil

54、l best serve their unique needs.Once those initiatives are agreed upon,companies must identify the data required before addressing security concerns,selecting an LLM and tracking performance.Leaders Invest in Future-Proof AI Initiatives5of factors influence decisions aroundgenerative AIinitiativesDo

55、zensHere are additional data-related questions to consider:What are the data security requirements?If the LLM will only be used for external needs,a commercial LLM will suffice.If the LLM will be ingesting private,internal data,businesses need a lean LLM behind the firewall to provide the necessary

56、security.What are the expectations around responsiveness?Response time from large commercial models is still measured in seconds,not milliseconds.While that may be fine for employees searching for information internally,its not sustainable for customers doing online shopping.Where is it best to put

57、costs?With commercial LLMs,start-up costs are low initially but costs increase according to the number of API calls.In contrast,an open-source model costs more upfront to create,but costs are lower moving forward.Key considerations to maximize success with Gen AI 142024 State of Generative AI in Glo

58、bal Business Generative AI Path To SuccessGen AI Orchestration EngineDefineUse CaseTrackBusiness CaseDataAcquisitionSecurity&AccuracyAccessControlCostControlLLM Model(s)“A language model alone is not a gen AI solution.Once key considerations around security,responsiveness,and cost have been addresse

59、d,leaders can define KPIs to track the business case over time.This pathway is how you get beyond 1 in 5 successful initiatives.”Mike Sinoway,CEO,Lucidworks Lucidworks clients are more than 2.5x more likely to deploy AI initiatives than their peers 2024 Lucidworks all rights reserved.Lucidworks is a

60、 registered trademark.Learn how Lucidworks can power your path to Gen AI success.Contact us today.GENERATIVE AI IN 2024:REALITY SETS INThe honeymoon phase of generative AI is over.While leaders remain enthusiastic about its potential to transform businesses,the initial euphoria has given way to a mo

61、re measured approach.This years report highlights a shift in focus from hype to practical implementation,emphasizing cost-effectiveness and responsible development.Key findings reveal a slowdown in spending as companies prioritize ROI and grapple with implementation challenges.Concerns around data s

62、ecurity and response accuracy have risen sharply,with implementation costs emerging as a major hurdle.Despite these obstacles,leaders are actively deploying Gen AI for applications with clear financial benefits,such as governance and cost reduction.The study emphasizes the importance of a well-defin

63、ed roadmap for Gen AI adoption.Leaders are advised to prioritize use cases with a high success rate,such as text-based applications.Careful consideration of data security,response time needs,and cost structure is crucial for successful implementation.Companies can maximize the value of this transfor

64、mative technology by focusing on practical applications,responsible development,and a strategic approach to cost management.As open-source models mature and costs come down,the future of Gen AI appears bright,but the path to success requires careful planning and execution.15About Lucidworks Lucidwor

65、ks believes that the core to a great digital experience starts with search and browse.Lucidworks captures user behavior and utilizes machine learning to connect people with the products,content,and information they need.The worlds largest brands including Crate&Barrel,Lenovo,and Red Hat rely on Lucidworks suite of products to power commerce,customer service,and workplace applications that delight customers and empower employees.Learn more at L