《麥肯錫：2024技術趨勢展望報告（英文）（100頁）.pdf》由會員分享，可在線閱讀，更多相關《麥肯錫：2024技術趨勢展望報告（英文）（100頁）.pdf（100頁珍藏版）》請在三個皮匠報告上搜索。

1、Technology Trends Outlook 2024July 2024McKinsey&CompanyMcKinsey&Company is a global management consulting firm,deeply committed to helping institutions in the private,public,and social sectors achieve lasting success.For more than 90 years,our primary objective has been to serve as our clients most

2、trusted external adviser.With consultants in more than 100 cities in over 60 markets,across industries and functions,we bring unparalleled expertise to clients all over the world.We work closely with teams at all levels of an organization to shape winning strategies,mobilize for change,build capabil

3、ities,and drive successful execution.Insights across trends The AI revolution Generative AI Applied AI Industrializing machine learning Building the digital future Next-generation software development Digital trust and cybersecurity Compute and connectivity frontiers Advanced connectivity Immersive-

4、reality technologies Cloud and edge computing Quantum technologies 4 13 14 20 25 30 31 36 43 44 49 54 59 65 66 71 7782 87 88 94 Cutting-edge engineering Future of robotics Future of mobility Future of bioengineering Future of space technologies A sustainable world Electrification and renewables Clim

5、ate technologies beyond electrification and renewables Contents3Technology Trends Outlook 2024Insights across trendsDespite challenging overall market conditions in 2023,continuing investments in frontier technologies promise substantial future growth in enterprise adoption.Generative AI(gen AI)has

6、been a standout trend since 2022,with the extraordinary uptick in interest and investment in this technology unlocking innovative possibilities across interconnected trends such as robotics and immersive reality.While the macroeconomic environment with elevated interest rates has affected equity cap

7、ital investment and hiring,underlying indicatorsincluding optimism,innovation,and longer-term talent needsreflect a positive long-term trajectory in the 15 technology trends we analyzed.These are among the findings in the latest McKinsey Technology Trends Outlook,in which the McKinsey Technology Cou

8、ncil identified the most significant technology trends unfolding today(to know more about the Council,see the sidebar“About the McKinsey Technology Council”).This research is intended to help executives plan ahead by developing an understanding of potential use cases,sources of value,adoption driver

9、s,and the critical skills needed to bring these opportunities to fruition.Our analysis examines quantitative measures of interest,innovation,investment,and talent to gauge the momentum of each trend.Recognizing the long-term nature and interdependence of these trends,we also delve into the underlyin

10、g technologies,uncertainties,and questions surrounding each trend.(For more about new developments in our research,please see the sidebar“Whats new in this years analysis”on page 9;for more about the research itself,please see the sidebar“Research methodology”on pages 1011.)New and notableThe two tr

11、ends that stood out in 2023 were gen AI and electrification and renewables.Gen AI has seen a spike of almost 700 percent in Google searches from 2022 to 2023,along with a notable jump in job postings and investments.The pace of technology innovation has been remarkable.Over the course of 2023 and 20

12、24,the size of the prompts that large language models(LLMs)can process,known as“context windows,”spiked from 100,000 to two million tokens.This is roughly the difference between adding one research paper to a model prompt and adding about 20 novels to it.And the modalities that gen AI can process ha

13、ve continued to increase,from text summarization and image generation to advanced capabilities in video,images,audio,and text.This has catalyzed a surge in investments and innovation aimed at advancing more powerful and efficient computing systems.The large foundation models that power generative AI

14、,such as LLMs,are being integrated into various enterprise software tools and are also being employed for diverse purposes such as powering customer-facing chatbots,generating ad campaigns,accelerating drug discovery,and more.We expect this expansion to continue,pushing the boundaries of AI capabili

15、ties.Senior leaders awareness of gen AI innovation has increased interest,investment,and innovation in AI technologies and other trends,such as robotics,which is a new addition to our trends analysis this year.Advancements in AI are ushering in a new era of more capable robots,spurring greater innov

16、ation and a wider range of deployments.About the McKinsey Technology CouncilTechnology is a catalyst for new opportunities,from inventing new products and services,expanding the productivity frontier and capturing more value in our day-to-day work.The McKinsey Technology Council helps business leade

17、rs understand frontier technologies and the potential application to their businesses.We look at a spectrum of technologies,from generative AI,machine learning,and quantum computing to space technologies that are shaping new opportunities and applications.The McKinsey Technology Council convenes a g

18、lobal group of more than 100 scientists,entrepreneurs,researchers,and business leaders.We research,debate,and advise executives from all industries as they navigate the fast-changing technology landscape.Lareina Yee,senior partner,McKinsey;chair,McKinsey Technology Council4Technology Trends Outlook

19、2024Electrification and renewables was the other trend that bucked the economic headwinds,posting the highest investment and interest scores among all the trends we evaluated.Job postings for this sector also showed a modest increase.Although many trends faced declines in investment and hiring in 20

20、23,the long-term outlook remains positive.This optimism is supported by the continued longer-term growth in job postings for the analyzed trends(up 8 percent from 2021 to 2023)and enterprises continued innovation and heightened interest in harnessing these technologies,particularly for future growth

21、.In 2023,technology equity investments fell by 30 to 40 percent to approximately$570 billion due to rising financing costs and a cautious near-term growth outlook,prompting investors to favor technologies with strong revenue and margin potential.This approach aligns with the strategic perspective le

22、ading companies are adopting,in which they recognize that fully adopting and scaling cutting-edge technologies is a long-term endeavor.This recognition is evident when companies diversify their investments across a portfolio of several technologies,selectively intensifying their focus on areas most

23、likely to push technological boundaries forward.While many technologies have maintained cautious investment profiles over the past year,gen AI saw a sevenfold increase in investments,driven by substantial advancements in text,image,and video generation.Despite an overall downturn in private equity i

24、nvestment,the pace of innovation has not slowed.Innovation has accelerated in the three trends that are part of the“AI revolution”group:generative AI,applied AI,and industrializing machine learning.Gen AI creates new content from unstructured data(such as text and images),applied AI leverages machin

25、e learning models for analytical and predictive tasks,and industrializing machine learning accelerates and derisks the development of machine learning solutions.Applied AI and industrializing machine learning,boosted by the widening interest in gen AI,have seen the most significant uptick in innovat

26、ion,reflected in the surge in publications and patents from 2022 to 2023.Meanwhile,electrification and renewable-energy technologies continue to capture high interest,reflected in news mentions and web searches.Their popularity is fueled by a surge in global renewable capacity,their crucial roles in

27、 global decarbonization efforts,and heightened energy security needs amid geopolitical tensions and energy crises.The talent environment largely echoed the investment picture in tech trends in 2023.The technology sector faced significant layoffs,particularly among large technology companies,with job

28、 postings related to the tech trends we studied declining by 26 percenta steeper drop than the 17 percent decrease in global job postings overall.The greater decline in demand for tech-trends-related talent may have been fueled by technology companies cost reduction efforts amid decreasing revenue g

29、rowth projections.Despite this reduction,the trends with robust investment and innovation,such as generative AI,not only maintained but also increased their job postings,reflecting a strong demand for new and advanced skills.Electrification and renewables was the other trend that saw positive job gr

30、owth,partially due to public sector support for infrastructure spending.Even with the short-term vicissitudes in talent demand,our analysis of 4.3 million job postings across our 15 tech trends underscored a wide skills gap.Compared with the global average,fewer than half of potential candidates hav

31、e the high-demand tech skills specified in job postings.Despite the year-on-year decreases for job postings in many trends from 2022 to 2023,the number of tech-related job postings in 2023 still represented an 8 percent increase from 2021,suggesting the potential for longer-term growth(Exhibit 1).+8

32、%17%26%tech trends job postings from 2021 to 2023global job postings from 2022 to 2023tech trends job postings from 2022 to 20235Technology Trends Outlook 2024Climate technologies beyond electrifcation and renewablesIndustrializingmachine learningImmersive-reality technologiesFuture of mobilityAppli

33、ed AINext-generationsoftwaredevelopmentFuture ofbioengineeringAdvancedconnectivityFuture ofroboticsQuantumtechnologiesFuture of space technologiesGenerativeAIDigital trust and cybersecurityElectrifcationand renewablesCloud and edgecomputing2021202320212023Annual change in tech trend job postings,202

34、123,millions of postingsDespite a one-year drop in job postings,demand for jobs in many technology trends has increased over two years.Climate technologies beyond electrifcation and renewablesIndustrializingmachine learningImmersive-reality technologiesFuture of mobilityApplied AINext-generationsoft

35、waredevelopmentFuture ofbioengineeringAdvancedconnectivityFuture ofroboticsQuantumtechnologiesFuture of space technologiesGenerativeAIDigital trust and cybersecurityElectrifcationand renewables 1Out of 130 million surveyed job postings(extrapolated JanOct 2023).Job postings are not directly equivale

36、nt to numbers of new or existing jobs.Source:McKinseys proprietary Organizational Data Platform,which draws on licensed,de-identifed public professional profle dataMcKinsey&CompanyCloud and edgecomputing00.20.40.600.20.400.20.60.81.01.21.400.20.40.600.20.400.20.60.81.01.21.420212023Building the digi

37、tal futureAI revolutionCompute and connectivityCutting-edge engineeringA sustainable world+52%change37%+49%34%+39%38%+77%36%202120222023Cumulative change in tech trend job postings,202123,millions of postings+6%23%+29%9%+110%+111%+29%20%+44%17%18%+18%+341%+3%+19%6%+20%+48%+73%5%change1%14%+14%+0%1%+

38、33%29%+34%11%+55%5%+72%+1%+49%34%+39%38%+77%36%+32%24%+55%36%00.20.40.600.20.400.20.81.000.20.40.600.20.400.20.81.0Exhibit 1 6Technology Trends Outlook 2024Adoption curve of technology trends,adoption scoreTechnologies progress through diferent stages,with some at the leading edge of innovation and

39、others approaching large-scale adoption.McKinsey&CompanyHigher adoptionLower adoptionWeb Exhibit of 1 Frontier innovation2 Experimenting3 Piloting4 Scaling5 Fully scaledAdoptionAdvanced connectivityApplied AICloud and edge computingGenerative AIDigital trust and cybersecurityElectrifcation and renew

40、ablesIndustrializing machine learningNext-gen software developmentClimate technologies beyond electrifcation and renewablesFuture of bioengineeringFuture of mobilityFuture of roboticsImmersive-reality technologiesFuture of space technologiesQuantum technologies4321Trend is more relevant to certain i

41、ndustries,resulting in lower overall adoption across industries compared with adoption within relevant industries.Source:McKinsey technology adoption survey data;McKinsey analysisExhibit 2Enterprise technology adoption momentum The trajectory of enterprise technology adoption is often described as a

42、n S-curve that traces the following pattern:technical innovation and exploration,experimenting with the technology,initial pilots in the business,scaling the impact throughout the business,and eventual fully scaled adoption(Exhibit 2).This pattern is evident in this years survey analysis of enterpri

43、se adoption conducted across our 15 technologies.Adoption levels vary across different industries and company sizes,as does the perceived progress toward adoption.We see that the technologies in the S-curves early stages of innovation and experimenting are either on the leading edge of progress,such

44、 as quantum technologies and robotics,or are more relevant to a specific set of industries,such as bioengineering and space.Factors that could affect the adoption of these technologies include high costs,specialized applications,and balancing the breadth of technology investments against focusing on

45、 a select few that may offer substantial first-mover advantages.As technologies gain traction and move beyond experimenting,adoption rates start accelerating,and companies invest more in piloting and scaling.We see this shift in a number of trends,such as next-generation software development and ele

46、ctrification.Gen AIs rapid advancement leads among trends analyzed,with about a quarter of respondents self-reporting that they are scaling its use.More mature technologies,like cloud and edge computing and advanced connectivity,continued their rapid pace of adoption,serving as enablers for the adop

47、tion of other emerging technologies as well(Exhibit 3).7Technology Trends Outlook 2024Self-reported adoption level by tech trend,2023,1%of respondents1Respondents may interpret these categories diferently based on their organizations.As such,the results should be considered as indicative of organiza

48、tions self-assessments,rather than precise measurements.2For a deeper look at our AI-related trends,see“The state of AI in early 2024:Gen AI adoption spikes and starts to generate value,”McKinsey,May 30,2024.Source:McKinsey technology adoption survey dataMore-mature technologies are more widely adop

49、ted,often serving as enablers for more-nascent technologies.McKinsey&CompanyExperimentingNot investingPilotingScalingFully scaledWeb Exhibit of 2217101181078554353141418181418171618161817221815253326263737373745464350414757131620211815192016182015192013262026242320201916151515131512Digital trust and

50、 cybersecurityNext-generation software development Future of space technologiesQuantum technologiesFuture of roboticsFuture of bioengineeringFuture of mobilityClimate technologies beyond electrifcation andrenewablesImmersive-reality technologiesIndustrializing machine learningElectrifcation and rene

51、wablesApplied AIAdvanced connectivityGenerative AI2Cloud and edge computingExhibit 3The process of scaling technology adoption also requires a conducive external ecosystem where user trust and readiness,business model economics,regulatory environments,and talent availability play crucial roles.Since

52、 these ecosystem factors vary by geography and industry,we see different adoption scenarios playing out.For instance,while the leading banks in Latin America are on par with their North American counterparts in deploying gen AI use cases,the adoption of robotics in manufacturing sectors varies signi

53、ficantly due to differing labor costs affecting the business case for automation.As executives navigate these complexities,they should align their long-term technology adoption strategies with both their internal capacities and the external ecosystem conditions to ensure the successful integration o

54、f new technologies into their business models.Executives should monitor ecosystem conditions that can affect their prioritized use cases to make decisions about the appropriate investment levels while navigating uncertainties and budgetary constraints on the way to full adoption(see the“Adoption dev

55、elopments across the globe”sections within each trend that showcase examples of adoption dimensions for the trends or particular use cases therein that executives should monitor).Across the board,leaders who take a long-term viewbuilding up their talent,testing and learning where impact can be found

56、,and reimagining the businesses for the futurecan potentially break out ahead of the pack.8Technology Trends Outlook 20249Technology Trends Outlook 2024The 15 tech trendsThis report lays out considerations for all 15 technology trends.For easier consideration of related trends,we grouped them into f

57、ive broader categories:the AI revolution,building the digital future,compute and connectivity frontiers,cutting-edge engineering,and a sustainable world.Of course,theres significant power and potential in looking across these groupings when considering trend combinations.To describe the state of eac

58、h trend,we developed scores for innovation(based on patents and research)and interest(based on news and web searches).We also sized investments in relevant technologies and rated their level of adoption by organizations(Exhibit 4).Whats new in this years analysisThis year,we reflected the shifts in

59、the technology landscape with two changes on the list of trends:digital trust and cybersecurity(integrating what we had previously described as Web3 and trust architectures)and the future of robotics.Robotics technologies synergy with AI is paving the way for groundbreaking innovations and operation

60、al shifts across the economic and workforce landscapes.We also deployed a survey to measure adoption levels across trends.1.00001.00.80.60.40.201.00.80.60.40.212345Interest,2 score(0=lower;1=higher)Innovation,1 score(0=lower;1=higher)Adoption level,score(1=frontier innovation;5=fully scaled)McKinsey

61、&CompanyInnovation,interest,investment,and adoption,by technology trend,2023Each trend is scored based on its level of innovation,interest,investment,and adoption.25015075 20Equity investment,$billionApplied AIIndustrializing machine learningFuture of bioengineeringElectrifcation/renewablesDigital t

62、rust and cybersecurityAdvanced connectivityFuture ofmobilityCloud and edge computingGenerative AINext-generation software developmentImmersive-reality technologiesClimate technologies beyond electrifcation and renewablesFuture of space technologiesFuture of roboticsQuantum technologiesNote:Innovatio

63、n and interest scores for the 15 trends are relative to one another.All 15 trends exhibit high levels of innovation and interest compared with other topics and are also attracting significant investment.1The innovation score combines the 01 scores for patents and research,which are relative to the t

64、rends studied.The patents score is based on a measure of patent filings,and the research score is based on a measure of research publications.2The interest score combines the 01 scores for news and searches,which are relative to the trends studied.The news score is based on a measure of news publica

65、tions,and the searches score is based on a measure of search engine queries.Exhibit 4Research methodologyTo assess the development of each technology trend,our team collected data on five tangible measures of activity:search engine queries,news publications,patents,research publications,and investme

66、nt.For each measure,we used a defined set of data sources to find occurrences of keywords associated with each of the 15 trends,screened those occurrences for valid mentions of activity,and indexed the resulting numbers of mentions on a 01 scoring scale that is relative to the trends studied.The inn

67、ovation score combines the patents and research scores;the interest score combines the news and search scores.(While we recognize that an interest score can be inflated by deliberate efforts to stimulate news and search activity,we believe that each score fairly reflects the extent of discussion and

68、 debate about a given trend.)Investment measures the flows of funding from the capital markets into companies linked with the trend.Data sources for the scores include the following:Patents.Data on patent filings are sourced from Google Patents,where the data highlight the number of granted patents.

69、Research.Data on research publications are sourced from Lens.News.Data on news publications are sourced from Factiva.Searches.Data on search engine queries are sourced from Google Trends.Investment.Data on private-market and public-market capital raises(venture capital and corporate and strategic M&

70、A,including joint ventures),private equity(including buyouts and private investment in public equity),and public investments(including IPOs)are sourced from PitchBook.Talent demand.Number of job postings is sourced from McKinseys proprietary Organizational Data Platform,which stores licensed,de-iden

71、tified data on professional profiles and job postings.Data are drawn primarily from English-speaking countries.In addition,we updated the selection and definition of trends from last years report to reflect the evolution of technology trends:The future of robotics trend was added since last years pu

72、blication.Data sources and keywords were updated.For data on the future of space technologies investments,we used research from McKinseys Aerospace&Defense Practice.10Technology Trends Outlook 2024Research methodology(continued)Finally,we used survey data to calculate the enterprise-wide adoption sc

73、ores for each trend:Survey scope.The survey included approximately 1,000 respondents from 50 countries.Geographical coverage.Survey representation was balanced across Africa,Asia,Europe,Latin America,the Middle East,and North America.Company size.Size categories,based on annual revenue,included smal

74、l companies($10 million to$50 million),medium-size companies($50 million to$1 billion),and large companies(greater than$1 billion).Respondent profile.The survey was targeted to senior-level professionals knowledgeable in technology,who reported their perception of the extent to which their organizat

75、ions were using the technologies.Survey method.The survey was conducted online to enhance reach and accessibility.Question types.The survey employed multiple-choice and open-ended questions for comprehensive insights.Definition of enterprise-wide adoption scores:1:Frontier innovation.This technology

76、 is still nascent,with few organizations investing in or applying it.It is largely untested and unproven in a business context.2:Experimentation.Organizations are testing the functionality and viability of the technology with a small-scale prototype,typically done without a strong focus on a near-te

77、rm ROI.Few companies are scaling or have fully scaled the technology.3:Piloting.Organizations are implementing the technology for the first few business use cases.It may be used in pilot projects or limited deployments to test its feasibility and effectiveness.4:Scaling.Organizations are in the proc

78、ess of scaling the deployment and adoption of the technology across the enterprise.The technology is being scaled by a significant number of companies.5:Fully scaled.Organizations have fully deployed and integrated the technology across the enterprise.It has become the standard and is being used at

79、a large scale as companies have recognized the value and benefits of the technology.11Technology Trends Outlook 2024Aakanksha Srinivasan Ahsan Saeed Alex Arutyunyants Alex Singla Alex Zhang Alizee Acket-Goemaere An Yan Anass Bensrhir Andrea Del Miglio Andreas Breiter Ani Kelkar Anna Massey Anna Orth

80、ofer Arjit Mehta Arjita Bhan Asaf Somekh Begum Ortaoglu Benjamin Braverman Bharat Bahl Bharath Aiyer Bhargs Srivathsan Brian Constantine Brooke Stokes Bryan Richardson Carlo Giovine Celine Crenshaw Daniel Herde Daniel Wallance David Harvey Delphine Zurkiya Diego Hernandez Diaz Douglas Merrill Elisa

81、Becker-Foss Emma Parry Eric Hazan Erika Stanzl Everett Santana Giacomo Gatto Grace W Chen Hamza Khan Harshit Jain Helen Wu Henning Soller Ian de Bode Jackson Pentz Jeffrey Caso Jesse Klempner Jim Boehm Joshua Katz Julia Perry Julian Sevillano Justin Greis Kersten Heineke Kitti Lakner Kristen Jenning

82、s Liz Grennan Luke Thomas Maria Pogosyan Mark Patel Martin Harrysson Martin Wrulich Martina Gschwendtner Massimo Mazza Matej Macak Matt Higginson Matt Linderman Matteo Cutrera Mellen Masea Michiel Nivard Mike Westover Musa Bilal Nicolas Bellemans Noah Furlonge-Walker Obi Ezekoye Paolo Spranzi Pepe C

83、afferata Robin Riedel Ryan Brukardt Samuel Musmanno Santiago Comella-Dorda Sebastian Mayer Shakeel Kalidas Sharmila Bhide Stephen Xu Tanmay Bhatnagar Thomas Hundertmark Tinan Goli Tom Brennan Tom Levin-Reid Tony Hansen Vinayak HV Yaron Haviv Yvonne Ferrier Zina ColeMichael Chui McKinsey Global Insti

84、tute partner,Bay AreaRoger Roberts Partner,Bay AreaMena Issler Associate partner,Bay AreaLareina Yee Senior partner,Bay Area;chair,McKinsey Technology CouncilAbout the authorsThe authors wish to thank the following McKinsey colleagues for their contributions to this research:We appreciate the contri

85、butions of members of QuantumBlack,AI by McKinsey,to the insights on the AI-related trends.They also wish to thank the external members of the McKinsey Technology Council for their insights and perspectives,including Ajay Agrawal,Azeem Azhar,Ben Lorica,Benedict Evans,John Martinis,and Jordan Jacobs.

86、12Technology Trends Outlook 2024The AI revolution13Technology Trends Outlook 2024Generative AIThe trendand why it mattersGenerative AI(gen AI)has been making significant strides,pushing the boundaries of machine capabilities.Gen AI models are trained on vast,diverse data sets.They take unstructured

87、data,such as text,as inputs and produce unique outputsalso in the form of unstructured dataranging from text and code to images,music,and 3D models.Over the past year,weve seen remarkable advancements in this field,with text generation models such as OpenAIs GPT-4,Anthropics Claude,and Googles Gemin

88、i producing content that mimics human-generated responses,as well as with image-generation tools such as DALL-E 3 and Midjourney creating photorealistic images from text descriptions.OpenAIs recent launch of Sora,a text-to-video generator,further showcases the technologys potential.Even music compos

89、ition is being revolutionized,with models such as Suno creating original pieces in various styles.Gen AI has sparked widespread interest,with individuals and organizations across different regions and industries exploring its potential.According to the latest McKinsey Global Survey on the state of A

90、I,65 percent of respondents say their organizations are regularly using gen AI in at least one business function,up from one-third last year,1 and gen AI use cases have the potential to generate an annual value of$2.6 trillion to$4.4 trillion.2 However,its important to recognize the risks that accom

91、pany the use of this powerful technology,including bias,misinformation,and deepfakes.As we progress through 2024 and beyond,we anticipate organizations investing in the risk mitigation,operating model,talent,and technological capabilities required to scale gen AI.Talent demand Ratioof skilled people

92、to job vacanciesEquity investment Private-and public-market capital raises for relevant technologiesPatents Patentflings for technologies related to trendNews Press reports featuring trend-related phrasesSearches Search engine queries for terms related totrendResearch Scientifc publications on topic

93、s associated with trend0.20.40.60.8NewsTalent demandResearchSearches1.0PatentsEquityinvestmentScoring the trendTHE AI REVOLUTIONScore by vector(0=lower;1=higher)Generative AIGen AI saw a surge in 2023,driven by ChatGPTs late-2022 launch,alongside earlier models such as DALL-E 2 and Stable Difusion.G

94、en AI saw signifcant growth from 2022 to 2023 across each quantitative dimension,such as a sevenfold increase in the number of searches and investments,refecting a strong sense of excitement about the trend.Industries afected:Aerospace and defense;Agriculture;Automotive and assembly;Aviation,travel,

95、and logistics;Business,legal,and profes-sional services;Chemicals;Construction and building materials;Consumer packaged goods;Education;Electric power,natural gas,and utilities;Financial services;Healthcare systems and services;Information technology and electronics;Media and entertainment;Metals an

96、d mining;Oil and gas;Pharmaceuticals and medical products;Public and social sectors;Real estate;Retail;Semiconductors;TelecommunicationsAdoption score,2023$36+111%12345FullyscaledFrontierinnovationEquity investment,2023,$billionJob postings,202223,%diference20191.0020231“The state of AI in early 202

97、4:Gen AI adoption spikes and starts to generate value,”McKinsey,May 30,2024.2 The economic potential of generative AI:The next productivity frontier,McKinsey,June 14,2023.14Technology Trends Outlook 2024Latest developmentsGen AI is a fast-growing and constantly innovating trend,with recent developme

98、nts including the following:Multimodal generative models are on the rise.As gen AI continues to evolve and gain more attention in various industries,its becoming increasingly clear that multimodality will play a pivotal role.By combining text,images,sounds,and videos,AI models can generate outputs a

99、pplicable across a wide range of industries and business functions.This pursuit of multimodality is intensifying across leading players such as OpenAI and Google(with its Lumiere AI web app).For example,Googles Gemini showcases a powerful multimodal system capable of processing information in variou

100、s formats,including text,code,tables,images,and even audio.Powerful open-source models are challenging their closed-source counterparts in performance and developer adoption.While significant investments are encouraging the development of proprietary large language models(LLMs),such as GPT-4 with vi

101、sion(GPT-4V),the AI community is also witnessing a surge in open-source models,such as Llama 3.This momentum is fueled by the enthusiasm of developers and users who welcome the unprecedented access to build innovative tools and study complex systems.The accessibility of open-source models is attract

102、ing a growing developer base.The context window in natural-language processing(NLP)is expanding.This expansion allows for longer and smarter prompts.For instance,in early 2024,Google released the largest context window in the market with Gemini 1.5 Pro,which has a standard context window of 128,000

103、tokens,with the potential to reach two million tokens.3 This larger context window enables the model to generate more coherent and contextually relevant responses by considering a larger amount of text.However,expanding prompt size can paradoxically lead to models getting“lost in the middle,”as they

104、 tend to focus on specific parts of the text while avoiding the rest.LLMs are increasingly being embedded into various enterprise tools.We are witnessing a significant uptick in the integration of LLMs into various enterprise tools.This surge is fueled by the growing demand for automation,efficiency

105、,personalized user experiences,and the capacity to decipher complex patterns that can lead to actionable insights.Consequently,a rising number of vendors are choosing to integrate LLMs into their applications and tools.This trend is especially prominent in the marketing and customer care domains,wit

106、h Salesforce Einstein and ServiceNow serving as prime examples.The multiagent approach has gained significant traction with the rapid development of LLMs and continued innovation.Companies now recognize the benefits of employing multiple language models that work in harmony rather than relying on a

107、single model.This approach offers a fresh perspective on tackling complex challenges by leveraging the capabilities of multiple AI agents,each specializing in different domains,to solve a single problem collaboratively.By working together,these agents can not only accelerate problem-solving but also

108、 leverage varied perspectives and expertise to deliver more effective and efficient solutions.Some of the tools using this approach tend to be unstable,but as models improve,their throughput should significantly increase,making them increasingly relevant for the future.3 The Keyword Googles official

109、 blog,“Gemini breaks new ground with a faster model,longer context,AI agents and more,”blog entry by Demis Hassabis,May 14,2024.15Technology Trends Outlook 2024Job postings by title,201923,thousandsDemandGenerative AITalent and labor marketsRoles related to gen AI have experienced signifcant and rap

110、id growth in talent demand since 2019,with a 111 percent increase in job postings compared with 2023.This growth is driven by increased interest and invest-ment in the feld.Almost all roles in gen AI,except for regulation afairs directors,have seen a notable rise in demand,particularly for individua

111、l contributor roles.Organizations are now focusing on scaling and expanding their internal capabilities to harness the potential of gen AI,leading to a sharp increase in demand for data scientists,software engineers,and data engineers.Talent availability,%share of postings requiring skillTalent avai

112、lability,ratio of talent to demandSkills availabilityProfciency in gen AI necessitates expertise in AI,machine learning,and programming languages such as Python.The availability of high-level gen AI skills is notable,with individuals citing profciency in this area as necessary to capture employers a

113、ttention.There are signifcant skills overlaps with the“applied AI”and“industrializing machine learning”trends(please refer to those trends for more details).Data scientistSoftware engineerData engineerSoftware developerProject managerRegulation afairs directorProduct managerMachine learning engineer

114、0123465201920230.34.12.93.70.412.10.23422201312126ArtifcialintelligenceMachinelearningPythonDataanalysisSoftwareengineeringGenerativeAIRegulatorycomplianceArtifcialintelligenceMachinelearningPythonDataanalysisSoftwareengineeringGenerativeAIRegulatorycompliance16Technology Trends Outlook 2024Adoption

115、 developments across the globeGen AI emerges as a front-runner in the trends landscape,sharing the top spot with electrification and renewables for the highest percentage of respondents scaling its implementation.This underscores its significance as a pivotal,high-growth trend to closely monitor thr

116、oughout the year.Many companies have made progress throughout the year on adopting gen AI and are currently working on scaling it across their businesses.While gen AI adoption has surged across various sectors,the technology,media,and telecommunications sector has notably emerged as a leader in the

117、deployment of the technology.The lack of availability of local language support poses challenges to adoption globally.Some countries,including India,Japan,and countries in the Middle East,have pushed to develop their own LLMs.In Africa,the prioritization of data locality and proximity hampers the bu

118、ilding of LLMs.Significant progress has been made recently with the emergence of multilingual models.Multilingual capabilities could become essential for any LLM,with the primary focus shifting to the degree of localization,including the use of slang,technical terms,and other nuances.Adoption dimens

119、ionsThe adoption trajectory of advanced technologies varies for each technology and each use case within that technology.Advancements along the following dimensions could enable reaching the next level of adoption for gen AI:a clearly defined ROI for widespread horizontal and vertical use cases by s

120、ector,along with a demonstrated ability to control risks and ensure safety with the development and deployment of new AI solutions decreased computational costs,alongside improvement in overall AI efficiencies(for example,improving latency)Since gen AI captured public attention at the end of 2022,a

121、significant amount of focus has been placed on delivering value through foundation models.Many are already demonstrating cross-industry value,such as coding acceleration or sales and marketing use cases,as well as domain-specific models,such as protein engineering or chemistry discovery foundation m

122、odels.The field continues to improve quickly with new toolsfor example,multimodal,agent-based models.Companies should concentrate on building capabilities in this domain and prioritize areas of focus to ensure they capture early value and arent left behind.Matej Macak,partner,London 17Technology Tre

123、nds Outlook 2024In real life Real-world examples involving the use of gen AI include the following:ING,a global financial institution,leveraged gen AI to enhance customer service in the Netherlands,one of its key markets.4 While the current classic chatbot usually resolves 40 to 45 percent of those

124、chats,that leaves another 16,500 customers a week who need to speak with a live agent for help.Recognizing gen AIs potential,ING developed a bespoke customer-facing chatbot to provide immediate,tailored assistance.This resulted in helping 20 percent more customers avoid long wait times and offering

125、instant gratification in the first seven weeks of use compared with the previous solution.The chatbot is expected to reach 37 million customers as it expands across ten markets.Recursion,a biotech company,has developed a new gen AI platform and trained an LLM to accelerate drug discovery.This platfo

126、rm enables scientists to simultaneously access multiple machine learning models that can process large amounts of proprietary biological and chemical data sets to save time during drug development.Ita Unibanco,Latin Americas largest private sector bank,created ad campaigns dedicated to women footbal

127、l athletes using AI.5 The campaign highlighted a 1941 Brazilian law that banned women from playing football.It used AI-generated images from conversations with real players and historians,among others,to create the“Brazilian teams that have never existed”campaign,paving the way for a new generation

128、of AI-based media advertising.Nubank is piloting a gen AI virtual assistant to boost customer service.6 The virtual assistant focuses on delivering personalized credit-related options within the Nubank ecosystem.It helps customers understand their credit card usage,explore collateralized credit oppo

129、rtunities,and use its online payment service NuPay without affecting their credit limits.Additionally,it may offer personal loans based on the customers profile and eligibility.For the initial phase,this innovative solution will be available exclusively to members of NuCommunity,Nubanks customer eng

130、agement platform.The assistant,developed using GPT-4 and Nubanks proprietary tools,was designed to evolve and improve through continuous customer interactions,ensuring a dynamic,customer-focused service.Underlying technologiesMultiple types of software and hardware power gen AI across the entire tec

131、h stack.These include the following:Application layer.Typically,this is the interface that the end user interacts with(for example,chat).Integration/tooling layer.Sitting between the application layer and foundation model,this layer integrates with other systems to retrieve information,filter respon

132、ses,save inputs and outputs,distribute work,and enable new features.Examples include the large-language-programming framework LangChain and vector databases such as Pinecone and Weaviate.Foundation models.These are deep learning models trained on vast quantities of unstructured,unlabeled data that c

133、an be used for a wide range of tasks out of the box or adapted to specific tasks through fine-tuning.Digital infrastructure.This involves using the digital abstraction of physical infrastructure to support data storage,processing,and computation.Digital abstraction includes databases(for example,SQL

134、 and NoSQL)and core tech services(for example,compute,storage,and networking).Physical infrastructure.This encompasses hardware that enables computational,data storage,and networking needs,including data centers,AI accelerator chips,and data center mechanical,electrical,and plumbing technologies.4“B

135、anking on innovation:How ING uses generative AI to put people first,”McKinsey,accessed May 2024.5“Brazilian teams that have never existed,”Ads of the World,Clio Awards,accessed May 2024.6“Nubank begins testing with generative artificial intelligence to enhance customers experience with credit,”Nuban

136、k press release,October 18,2023.18Technology Trends Outlook 2024Key uncertainties The major uncertainties affecting gen AI include the following:Cybersecurity and privacy concerns are prevalent,notably regarding data leakage risks and vulnerabilities(including customer and protected data).Ethical co

137、nsiderations surround the responsible use of gen AI,including data governance,justice and fairness,accountability,and explainability.Regulation and compliance might affect research into gen AI and its potential applications.Copyright ownership and protection of content generated by open-source model

138、s remains an unanswered question.Environmental impact may increase as training models expend exponentially more computational resources.Inaccuracies are the most recognized and experienced risk for gen AI uses,7 and they can affect use cases across the gen AI value chain.Big questions about the futu

139、reCompanies and leaders may want to consider a few questions when moving forward with gen AI:How will the cost of model creation evolve,and how will it affect competitive dynamics?Will enterprise adoption experience the same level of exponential growth and monetization as seen in consumer adoption?H

140、ow will the market develop in terms of open-source solutions versus closed-source?How should companies approach gen-AI-related risks,including data privacy and security,equity,fairness,compliance,and copyright protections?What strategies should policy makers adopt to address the risk of social engin

141、eering from third-party LLM solutions?When will error rates and avoidance of hallucinations get to an acceptable level for large-scale implementations of gen AI in everyday use cases?Which workers will see their roles shift due to gen AI,and to what extent will they be affected?As technological adva

142、ncements such as gen AI models,accelerators,and throughput continue to evolve,what are the primary use cases that companies should prioritize,and how should they position themselves for future relevance in terms of their degree of involvement,whether as shapers,takers,or makers?19Technology Trends O

143、utlook 20247“The state of AI in early 2024,”May 30,2024.Gen AI is currently at the exciting nexus of demonstrated proof of value,rapid innovation,significant public and private investment,and widespread consumer interest.The year 2023 was the year of pilots,and,moving forward,we can expect to see tw

144、o important areas of focus to accelerate adoption and value creation:one is a rapid expansion of modular and secure enterprise platforms that will serve as the foundation for developing gen AI applications,and two,a focus on the reskilling and rewiring of processes required in a business domain to d

145、rive user adoption and capture value.Delphine Nain Zurkiya,senior partner,Boston Applied AIThe trendand why it mattersAs we navigate through 2024,the impact of analytical AI technologies,including applications of machine learning(ML),computer vision,and natural-language processing(NLP),continues to

146、grow across all sectors.Companies are using data to derive insights to automate processes,transform businesses,and make better decisions.McKinsey research estimates that AI applications can potentially unlock an economic value of$11 trillion to$18 trillion annually.1 The excitement around generative

147、 AI(gen AI)has led to increasing awareness of the potential value of applied AI.In our recent global survey on the state of AI in 2024,67 percent of respondents say they expect their organizations to invest more in AI over the next three years.2 The survey highlights that organizations continue to s

148、ee returns from AI efforts across business domains.Regulators and policy makers alike are also taking note of AIs increasing impact,with the European Parliament,for example,passing the unified EU Artificial Intelligence Act.3 However,the journey to AI adoption is filled with challenges and learning

149、opportunities,such as transforming organizational culture to foster collaboration,trust,and adaptation to new ways of working;acquiring,leveraging,and organizing valuable sources of large data sets;and interpreting model outputs to build end-user trust in them.Leaders should anticipate challenges su

150、ch as governance conflicts across the entire businessgiven the cross-domain nature of AIand the rapid evolution of the regulatory and ethical landscape.Despite these challenges,establishing protocols and guardrails,along with effective change management,can help mitigate risks and ensure the success

151、ful incorporation of AI into business operations.Talent demand Ratioof skilled peopleto job vacanciesEquity investment Private-and public-market capital raises for relevant technologiesPatents Patentflings for technologies related to trendNews Press reports featuring trend-related phrasesSearches Se

152、arch engine queries for terms related totrendResearch Scientifc publications on topics associated with trend0.20.40.60.8NewsTalent demandResearchSearches1.0PatentsEquityinvestmentScoring the trendTHE AI REVOLUTIONScore by vector(0=lower;1=higher)Applied AIHigh innovation and investment scores for ap

153、plied AI are commensurate with its large potential impact.Each year from 2019 to 2023,applied AI has had the highest innovation scores of all the trends we studied,and its investment score also ranks in the top fve.While demand for applied AI talent declined 29 percent from 2022 to 2023,perhaps unsu

154、rprisingly,in 2023,demand for talent in applied AI remained among the highest of all the trends we studied.Industries afected:Aerospace and defense;Agriculture;Automotive and assembly;Aviation,travel,and logistics;Business,legal,and profes-sional services;Chemicals;Construction and building material

155、s;Consumer packaged goods;Education;Electric power,natural gas,and utilities;Financial services;Healthcare systems and services;Information technology and electronics;Media and entertainment;Metals and mining;Oil and gas;Pharmaceuticals and medical products;Public and social sectors;Real estate;Reta

156、il;Semiconductors;TelecommunicationsAdoption score,2023$8629%12345FullyscaledFrontierinnovationEquity investment,2023,$billionJob postings,202223,%diference20191.0020231 The economic potential of generative AI:The next productivity frontier,McKinsey,June 14,2023.2“The state of AI in early 2024:Gen A

157、I adoption spikes and starts to generate value,”McKinsey,May 30,2024.3 Melissa Heikkil,“The AI Act is done.Heres what will(and wont)change,”MIT Technology Review,March 19,2024.20Technology Trends Outlook 2024Latest developmentsRecent developments involving applied AI include the following:The emphas

158、is on data-centric AI is growing.Rich,high-quality data sets are essential assets for capturing value from AI.The shift toward data-centric AI represents a significant evolution in the field,as capabilities such as picking the right model or hyperparameter tuning become more automated and easier to

159、use.Data-centric AI use cases are diverse and widespread,but specific examples include financial institutions using it to detect and prevent fraudulent activities,healthcare providers promoting transparency in AI-driven diagnoses,or manufacturers identifying potential biases in quality control syste

160、ms.A companys unique data can be used to train AI models to automate and optimize core processes and unlock new business potential.Having access to unique data sets can provide a distinct competitive advantage,which explains why companies such as OpenAI are actively seeking new data and purchasing e

161、xclusive rights.As companies build their own private AI environments,the scope of data governance will expand beyond privacy to address interconnected threats such as data poisoning(for example,tampering with the training data)and model hijacking(for instance,taking control of an existing model and

162、manipulating it to perform unauthorized tasks).This transition requires robust data practices,including maintaining data quality,tracing data lineage,and employing explainable AI to foster trust and reduce bias.Hardware acceleration has gained significant momentum in applied AI.The continuous increa

163、se in the scale and complexity of deep learning models has surpassed the abilities of conventional central processing units,accelerating hardware development.To train these large models and operate them in real time,organizations are shifting toward specialized hardware such as graphics processing u

164、nits(GPUs),field-programmable gate arrays(FPGAs),application-specific integrated circuits(ASICs),and high-bandwidth memory(HBM)chips.Originally designed for graphics,GPUs now provide the parallel processing power needed for AI tasks.FPGAs offer adaptability for custom solutions at the edge,while ASI

165、Cs offer top-tier performance and efficiency for specific tasks.As the complexity of AI increases,the search for faster,more efficient hardware persists.By leveraging the capabilities of specialized hardware,organizations can spearhead the forthcoming wave of AI innovation.Generative AI(gen AI)opens

166、 the door to more applied AI.Gen AI adoption is not only increasing among curious individuals but also catalyzing increased adoption of applied AI.We see organizations getting the most impact from gen AI when they intertwine it with applied AI use cases.For instance,a digital-marketing company is us

167、ing gen AI to create a variety of unique and engaging content for its customers.However,the real magic happens when its applied AI systems analyze the performance of the generated content,identifying patterns and trends in user engagement.This data is then used to inform the gen AI system,generating

168、 insights to produce more effective content in the future.In this way,gen AI is being informed by real-world data and feedback.The synergy between gen AI and applied AI is what truly unlocks the potential of both technologies.The prominence of generative AI has opened the aperture for business leade

169、rs to explore applied AI,which could have as much or greater business impact.Michael Chui,partner,Bay Area21Technology Trends Outlook 2024Job postings by title,201923,thousandsDemandApplied AITalent and labor marketsBetween 2019 and 2022,applied AI saw rapid growth in demand for talent,with job post

170、ings more than tripling.Then,in line with the overall job market,applied AI saw a 29 percent reduction in total job postings across the most common positions in 2023 compared with 2022.However,applied AI continues to have among the most job postings per trend,with more than 500,000 job postings in 2

171、023.And with highinvestment activity,one could expect the demand for applied AI talent to remain steady.Talent availability,%share of postings requiring skillTalent availability,ratio of talent to demandSkills availabilityThere is a signifcant demand for specialized AI-related skill sets,and more pe

172、ople are striving to acquire these skills,leading to larger numbers of people listing these skills on their profles.As the level of competency achieved can vary,companies will need to assess the skills profciency of potential job applicants.020406080120100Data scientistSoftware engineerData engineer

173、Software developerProject managerMachine learning engineerScientistProduct manager20192023584.13.76.10.32.92.24836242011MachinelearningArtifcialintelligencePython DataanalysisDatascienceDeeplearningMachinelearningArtifcialintelligencePython DataanalysisDatascienceDeeplearning22Technology Trends Outl

174、ook 2024Adoption developments across the globeApplied AI tools are widely adopted across industries and regions,driven by advancements in AI capabilities and an increase in use cases.Most companies adopt applied AI technologies to increase revenuefor example,through integration with existing offerin

175、gs or completely new product and revenue streams.The technology,media,and telecommunications and financial-services sectors have emerged as leaders in the adoption of applied AI tools.Some of these companies are also the makers and innovators of the technology itself.Across industries,including fina

176、ncial and professional services,energy and materials,and consumer goods,companies also have made significant investments in applied AI tools.Adoption dimensionsThe adoption trajectory for advanced technologies will look different for each technology and each use case within that technology.Advanceme

177、nts along the following dimensions could enable the next level of adoption for applied AI:improved availability of plug-and-play solutions to allow seamless integration into existing IT and cloud infrastructure,combining standardized and interoperable industrialized ML with gen AI capabilities for a

178、 broader range of industry use cases and clear ROI effective change management to foster continuous learning and knowledge sharing through training,best-practice dissemination,and role modeling to drive effective organizational adoption of AI technologies robust implementation of ML-operations(MLOps

179、)and large-language-model-operations(LLMOps)practices to ensure optimal performance of AI models in production environments(for more,see the“Industrializing machine learning”trend),enabling seamless scalability and sustained performance from minimum-viable products to enterprise-wide deployment impr

180、oved data organization,availability,and governance across organizations to enable AI use casesIn real life Real-world examples involving the use of applied AI include the following:Saudi Aramco has built an AI hub to efficiently analyze more than five billion data points per day from wellheads in th

181、e oil and gas fields,enhancing the understanding of petro-physical properties and expediting decision making in exploration and drilling.The solutions provide real-time alerts to prevent business disruption,improve reservoir performance,and save millions of dollars by optimizing field development pl

182、ans and well trajectories.AI technology is also used to predict and prevent drilling challenges,such as stuck pipes,and to monitor the health of essential equipment,such as steam traps,using infrared images.4 4 Victoria Sayce,“The AI Hub at Aramco:The home of our next-generation of digital innovatio

183、n,”Aramco,October 23,2022.Applied AI has been transforming the way we work for some time now.Gen AI takes this to a new level,allowing organizations to tackle end-to-end workflows that were previously too complex to go after.This is possible with gen AIs powerful off-the-shelf models that are comple

184、mented by data-centric approaches.As we apply these technologies,organizations need to emphasize the integral human part of these workflows,ensuring that these solutions are built for end users,by end users.Stephen Xu,director of product management,QuantumBlack,AI by McKinsey,Toronto23Technology Tre

185、nds Outlook 2024 DigitalOwls AI-powered platform facilitates the efficient processing and analysis of extensive medical records,encompassing both traditional and electronic health records.Tailored for life insurance underwriters,this solution simplifies the navigation of complex and voluminous medic

186、al documents by extracting and organizing critical information.5 Vistra Corp,the largest competitive power producer in the United States,committed to a 60 percent emissions reduction by 2030 and net-zero emissions by 2050.6 Among several emissions reduction initiatives,Vistra wanted to understand ho

187、w AI might help it run its power plants more efficiently.The company used a multilayered neural network model,trained on two years of plant data,to determine optimal plant operationsfor example,set points in the control room to achieve maximum heat-rate efficiency for any combination of external fac

188、tors,such as temperature and humidity.Once power engineering experts validated the models,they began to provide recommendations to operators every 30 minutes to enhance the plants heat-rate efficiency,helping operators meet energy targets and improve plant reliability.This led to a 30 percent decrea

189、se in duct burner usage,annual fuel savings of about$175,000,and reduced carbon emissions,ensuring more efficient,more reliable power.The insights were incorporated into a solution named the Heat Rate Optimizer(HRO),which was implemented across the companys entire fleet,yielding$23 million in saving

190、s.Vistra has since extended the HRO to 67 additional power generation units across 26 plants.7 Underlying technologiesAI comprises several technologies that perform cognitive-like tasks.For further information on underlying technology for gen AI,please refer to the gen AI section of the report.The t

191、echnologies underlying applied AI include the following:Machine learning.This term refers to models that make predictions after being trained with data rather than following programmed rules.Computer vision.This type of ML works with visual data,such as images,videos,and 3D signals.Natural-language

192、processing.This type of ML analyzes and generates language-based data,such as text and speech.Deep reinforcement learning.This type of ML uses artificial neural networks and training through trial and error to make predictions.Additional hardware tools and technologies.These are other tools and tech

193、nologiessuch as cloud computing and domain-specific architectures,including GPUsthat improve access to high-capacity compute for AI and ML workflows.Key uncertainties The major uncertainties affecting applied AI include the following:Cybersecurity and privacy concerns,notably on data risks and vulne

194、rabilities,are prevalent:51 percent of survey respondents cited cybersecurity as a leading risk in 2024.8 Regulation and compliance might affect AI research and applications.Ethical considerationsincluding data governance,equity,fairness,and explainabilitysurround the responsible and trustworthy use

195、 of AI.Operational risks may arise from AI failure modes,as well as potential risks associated with data quality and integrity,model drift,adversarial attacks,and the need for ongoing training and education.Big questions about the futureCompanies and leaders may want to consider a few questions when

196、 moving forward with applied AI:How might companies identify the most beneficial AI applications and strategically use generative and applied AI together?What are the talent and tech stack implications of adopting applied AI?How can companies get ahead of their competitors and capture the value at s

197、cale associated with applied AI(regarding either revenue or cost benefits)?How will companies balance AIs potential cost savings while integrating features to make AI trustworthy and responsible?What checks should companies implement to guard against AI-related risks associated with data privacy and

198、 security,equity,fairness,and compliance?5“DigitalOwl revolutionizes medical record analysis and review with the latest release of version 4.0 of their Digital Medical Abstract(DMA),”Business Wire,January 17,2023;“Nationwide is streamlining life underwriting process with DigitalOwls advanced AI tech

199、nology,”LIFE&Health Advisor,June 3,2024.6“An AI power play:Fueling the next wave of innovation in the energy sector,”McKinsey,May 12,2022.7 Ibid.8“The state of AI in early 2024,”McKinsey,May 30,2024.24Technology Trends Outlook 2024Industrializing machine learningThe trendand why it mattersIndustrial

200、izing machine learning(ML),also known as machine learning operations(MLOps),is the process of scaling and maintaining ML applications within enterprises.As we progress through 2024,MLOps tools are rapidly evolving,improving in both functionality and interoperability.These tools are facilitating the

201、transition from pilot projects to robust business processes,enabling the scaling of analytics solutions,and enhancing team productivity.Successful industrialization of ML can help sustain solutions,reduce the production timeline for ML applications by eight to ten times,and decrease development reso

202、urces by up to 40 percent.1 Initially introduced by a few pioneering companies,MLOps is becoming more widely adopted as more companies use AI for a broader spectrum of applications.The rise of generative AI(gen AI)has reshaped the AI landscape,demanding a corresponding upgrade in MLOps capabilities

203、to service its unique demands.This is the newest field for novel developments in the industrializing ML trend.MLOps and foundation model operations(FMOps)are essential for industrializing and scaling gen AI safely and efficiently.Talent demand Ratioof skilled peopleto job vacanciesEquity investment

204、Private-and public-market capital raises for relevant technologiesPatents Patentflings for technologies related to trendNews Press reports featuring trend-related phrasesSearches Search engine queries for terms related totrendResearch Scientifc publications on topics associated with trend0.10.20.30.

205、4NewsTalent demandResearchSearches0.5PatentsEquityinvestmentScoring the trendTHE AI REVOLUTIONScore by vector(0=lower;1=higher)Industrializingmachine learningScores across news,searches,publications,and patents have more than doubled between 2019 and 2023,while demand for talent has more than triple

206、d in the same time frame.These increases suggest that the use of methods for industrializing ML could expand in the years ahead.Equity investment activity in MLOps has dropped in two consecutive years.Industries afected:Aerospace and defense;Agriculture;Automotive and assembly;Aviation,travel,and lo

207、gistics;Business,legal,and profes-sional services;Chemicals;Consumer packaged goods;Education;Electric power,natural gas,and utilities;Financial services;Healthcare systems and services;Information technology and electronics;Media and entertainment;Metals and mining;Oil and gas;Pharmaceuticals and m

208、edical products;Public and social sectors;Real estate;Retail;Semiconductors;TelecommunicationsAdoption score,2023$336%12345FullyscaledFrontierinnovationEquity investment,2023,$billionJob postings,202223,%diference20191.00.5020231 Based on observations from MLOps deployment in a series of large-scale

209、 analytics transformations supported by McKinsey.25Technology Trends Outlook 2024Latest developmentsRecent developments involving industrializing ML include the following:Monitoring and orchestration are becoming crucial components of MLOps.This is particularly evident in the complex task of upgradi

210、ng enterprise technology architecture to integrate and manage models and orchestrate interaction between ML models and other applications and data sources.Several integration patterns are emerging,including those that allow models to call APIs in response to user queries.Recent advancements in integ

211、ration and orchestration frameworks,such as LangChain and LlamaIndex,have greatly facilitated these developments.To effectively integrate these models,its essential for MLOps pipelines to incorporate specific performance measurement tools.For instance,they need to assess a models ability to retrieve

212、 the correct information.Companies such as Fiddler and Databricks are actively investing in this field.They offer performance tracking,validation,and orchestration,enabling companies to monitor their live operations effectively.This ensures the smooth facilitation of their ML applications,further em

213、phasizing the importance of monitoring and orchestration in the successful implementation of MLOps.The use of prebuilt solutions and APIs is on the rise.In recent years,there has been a significant surge in the availability of machine learning APIs and preconfigured solutions,partly due to the explo

214、sive growth of gen AI.Accessed predominantly through APIs,gen AI technologiesencompassing advanced tools such as computer vision libraries and pretrained image recognition modelshave profoundly reshaped the ML development landscape.As these APIs gain popularity and continue to evolve,they are progre

215、ssively assuming responsibilities that were once the purview of ML engineers,such as data preprocessing and model training on predefined data sets.As gen AI continues to evolve,its impact on the field is expected to expand,making it a pivotal driver of industrializing ML technologies.MLOps is import

216、ant to gen AI from the get-go.It is increasingly recognized that gen AI,including large language models,should not be viewed as separate from the ML ecosystem.Instead,there is a call to broaden the scope of MLOps to incorporate gen AI.MLOps is crucial in developing,deploying,and maintaining gen AI s

217、olutions,allowing ML algorithms to be dispatched quickly and effectively.By standardizing processes,enabling version control and tracking,and integrating shared metrics and monitoring,MLOps breaks down organizational silos and fosters close collaboration between data scientists,ML engineers,and oper

218、ations teams and thus is pivotal in the end-to-end life cycle of gen AI.The past several years have yielded huge advances in the mathematics of machine learning,but the tasks of making that math really useful have lagged.MLOpsthe way that the math gets made usefulis finally catching up.The tools and

219、 processes are beginning to mature,but we still need additional talent and skills to reap the benefits of machine learning.Douglas Merrill,partner,Southern California26Technology Trends Outlook 2024Job postings by title,201923,thousandsDemandIndustrializing machine learningTalent and labor marketsJo

220、b postings for roles related to industrializing ML decreased by 36 percent compared with 2022 levels,signifcantly greater than the 26 percent average decrease across all technology trends.As software evolves,many tasks will be automated,and many MLOps tasks are expected to become the responsibility

221、of frontline ML developers.Companies investing in industrializing ML are shifting their focus from rapid application development to efective scaling and implementation.Monitoring is becoming a crucial component,encom-passing performance tracking,validation,and orchestration.MLOps oferings are also g

222、rowing to streamline industrialization(for example,Databricks).Talent availability,%share of postings requiring skillTalent availability,ratio of talent to demandSkills availabilityCompanies that are industrializing their ML initiatives require advanced technological skills,and there is talent avail

223、able to meet this demand.These skills include expertise in containerization with Docker,orchestration with Kubernetes,and profciency in programming languages such as Python.Software engineerSoftware developerData scientistData engineerWeb developerSystems administratorTechnical architectMachine lear

224、ning engineer020406080120100201920231.42.12.91.51.1KubernetesDockerPythonCloudcomputingDevOpsKubernetesDockerPythonCloudcomputingDevOps6845444232MachinelearningMachinelearning184.127Technology Trends Outlook 2024Adoption developments across the globeAdoption levels of industrialized ML fall in the m

225、iddle relative to other trends,as other advancements have generated more buzz in recent years.Some of the leading sectors in adopting industrialized-ML practices include energy and materials and technology,media,and telecommunications.Additionally,financial-services companies made significant invest

226、ments in ML tools,driven by a focus on enhancing customer satisfaction and improving decision making.Adoption dimensionsThe adoption trajectory for advanced technologies will look different for each technology and each use case within that technology.Advancements along the following dimensions could

227、 enable the next level of adoption for industrializing ML:greater availability of simplified tools for data management and an increase in data source availability and robustness in terms of data quality and volume of data points,potentially through improved data services continued standardization an

228、d improvements in underlying technologies across the ML/AI software development life cycle(for example,model development,deployment,and monitoring)organizational adoption and awareness to improvemaking the technology more broadly accessible and understood by nontechnical employees In real lifeReal-w

229、orld examples involving industrializing ML include the following:Meta uses HawkEye internally to gain a comprehensive understanding of its ML workflows.2 HawkEye functions as a real-time monitor,anomaly detector,and analyst for potential issues,from data quality to model performance.It also ensures

230、end-to-end observability through tracing of ML pipelines,integration with explainable AI,and the provision of debugging tools.MLflow,an open-source platform aimed at streamlining ML development,is adding generative AIcentered capabilities.For example,its prompt engineering user interface provides an

231、 opportunity to try out multiple large language models(LLMs),parameter configurations,and prompts.3 2 Partha Kanuparthy,Animesh Dalakoti,and Srikanth Kamath,“AI debugging at Meta with HawkEye,”Engineering at Meta,December 19,2023.3 MLflow Blog,“2023 Year in Review,”blog entry by Carly Akerly,January

232、 26,2024.Solving for gaps in automated monitoring and life cycle management of deployed AI solutions will ensure the lasting and scalable impact of AI.That includes continued focus on gen AI:industrializing bespoke gen AI solutions will require robust gen AI operational ecosystems,and we see more op

233、tions emerging for processing unstructured data,engineering and operating LLM flows,and automating the gen AI solutions life cycle.Continued progress on enablement of regulatory and ethical alignment and explainability will help unlock new areas of AI impact.Alex Arutyunyants,senior principal data e

234、ngineer,QuantumBlack,AI by McKinsey,Boston28Technology Trends Outlook 2024Underlying technologiesSoftware solutions enable the various stages of the ML workflow,which are as follows:Data management.Automated data management software improves data quality,availability,and control in feeding the ML sy

235、stem.Model development.Tooling is used to build and optimize ML models,engineer features,and standardize processes.Model deployment.Provision tooling helps to test and validate ML models,brings them into production,and standardizes processes.Live-model operations.With this process,software maintains

236、 or improves the performance of models in production.Model observability.These tools go beyond basic monitoring and delve deeper into understanding a models behavior.They provide insights into model performance,identify potential biases,explain model decisions,and help diagnose issues such as data d

237、rift or concept drift.Key uncertaintiesThe major uncertainties affecting industrializing ML include the following:Up-front investment and resources will be required to establish industrialized ML in organizations.Processes and accountability will be crucial for maintaining ML solutions at an industr

238、ial scale.A fast-evolving market will require organizations to balance the efficiency of using their existing vendors offerings with realizing value from newer offerings provided by players outside their existing vendor ecosystem.The potential for misaligned capabilities will need to be avoided by e

239、nsuring that organizations are investing in the right solutions and at the right levels for their specific use case needs.Continuous monitoring and evaluation will be crucial for identifying and addressing unwanted bias throughout the ML life cycle,from initial data selection to ongoing model perfor

240、mance assessments.Technology and talent evolution will be essential,due to increasing automation of certain roles and the need for workers who are skilled in building and maintaining productionized ML systems at scale.Big questions about the futureCompanies and leaders may want to consider a few que

241、stions when moving forward with industrializing ML:With the emergence and acceleration of gen AI,how will MLOps practices and the technology ecosystem evolve?With the proliferation of new technologies in ML,how should organizations prioritize those along the ML workflow that are most relevant to the

242、ir needs?How will industrialized ML change organizations,their operating models,and their engineering roles?As industrialized ML proliferates,how can organizations define accountability roles to ensure the trustworthy and responsible use of AI/ML?How can organizations best integrate their MLOps effo

243、rts across machine learning,deep learning,and gen AI models?29Technology Trends Outlook 2024Building the digital future30Technology Trends Outlook 2024Next-generation software developmentThe trendand why it mattersThe landscape of software development is currently experiencing a transformative shift

244、,driven by an influx of cutting-edge technologies such as generative AI(gen AI)and cloud-native architectures.The year 2023 saw a significant rise in AI-powered tools,building on previous years advancements in software development and DevOps automation(for example,continuous integration,continuous d

245、elivery,infrastructure as code,and improved integrated development environments).These innovations are revolutionizing how engineers operate throughout the entire software development life cycle(SDLC),from planning and testing to deployment and maintenance.These technological breakthroughs are not o

246、nly enhancing the capabilities of engineers but also opening doors for less technical professionals to participate in application development as complex tasks are simplified and accelerated.While the path to wide-scale adoption may take more timebecause of obstacles such as integration challenges,la

247、ck of clear measurement metrics for developer productivity,and need for large-scale retraining of developers and test engineersan increase in the adoption of AI-powered software development tools is promising.Early adopters are already experiencing productivity boosts,laying the groundwork for more

248、widespread adoption in the near future.This year promises even more groundbreaking possibilities as maturing technologies like user-friendly low-code platforms,AI assistants throughout the SDLC,integration with gen-AI-enabled product management tools,and scalable cloud architectures converge,leading

249、 to democratized development,hyperefficiency,and exceptional adaptability.Talent demand Ratioof skilled peopleto job vacanciesEquity investment Private-and public-market capital raises for relevant technologiesPatents Patentflings for technologies related to trendNews Press reports featuring trend-r

250、elated phrasesSearches Search engine queries for terms related totrendResearch Scientifc publications on topics associated with trend0.20.40.60.8NewsTalent demandResearchSearches1.0PatentsEquityinvestmentScoring the trendBUILDING THE DIGITAL FUTUREScore by vector(0=lower;1=higher)Next-generationsoft

251、ware developmentA large uptick in searches,publications,patents,and talent demand between 2019 and 2023 clearly signals that both institutions and enterprises are seeing long-term potential in the evolution of next-generation software development tools.The investment climate for this technology has

252、seen peaks and valleys over the past fve years(the peaks refecting a few mega deals in some years).Industries afected:Advanced industries;Business,legal,and professional services;Consumer packaged goods;Financial services;Healthcare systems and services;Information technology and electronics;Manufac

253、turing;Media and entertainment;TelecommunicationsAdoption score,2023$1737%12345FullyscaledFrontierinnovationEquity investment,2023,$billionJob postings,202223,%diference201920231.0031Technology Trends Outlook 2024Latest developmentsRecent developments involving next-generation software development i

254、nclude the following:New versions of AI-powered development tools are transitioning from proof of concept to wide-scale application.The software development industry has witnessed a significant turning point in the past year with the release of new versions of advanced AI-powered tools that are tran

255、sforming the landscape.Unlike their static and off-the-shelf predecessors,these new versions have moved beyond the proof-of-concept phase and now offer a higher degree of adaptability and customization,catering to the specific needs of individual projects.This shift is resulting in a wider applicati

256、on of these tools.For instance,Tabnine,an AI-powered auto-completion tool,has improved its ability to understand the context of developers code,leading to more accurate and relevant code completions.Developers can now create and share custom code templates within Tabnine,allowing them to automate re

257、petitive tasks specific to their projects or coding style,thereby increasing the tools applicability on a larger scale.1 There is a growing trend toward more integrated development platforms.Companies are moving away from a multitude of disparate tools and,instead,adopting a smaller number of robust

258、 or better-integrated solutions that offer a wide range of functionalities throughout the development life cycle.This shift provides several advantages,including streamlined workflows that lead to improved collaboration,reduced context switching,and enhanced data visibility.However,catering to diver

259、se use cases within the organization requires a careful selection of tools with robust capabilities and flexibility.The talent landscape will undergo changes.The availability of advanced underlying technologies,such as gen AI,is enabling software engineers to reallocate their time from tasks such as

260、 pure code generation to tasks such as architecture design and problem solving.This change is not only causing a strong mindset shift among engineers but also influencing how companies approach talent selection,upskilling,and onboarding.The focus is no longer solely on the coding skills of potential

261、 candidates.Instead,companies are investing in defining a differentiated upskilling strategy to retain and develop talent and are now also assessing how effectively candidates can utilize and adapt to these advanced tools in their day-to-day tasks.The focus on compliance and trust is increasing.The

262、software development industry is experiencing a significant shift toward compliance and trust in response to growing concerns about legal and security risks associated with software tools.This past year has seen growing attention to compliance-focused tools like SonarQube,which provide features like

263、 code tagging,labeling,and detection to improve transparency and accountability.Developers are also choosing tools with guaranteed indemnity to mitigate potential legal risks associated with code generated or analyzed by the tool.By prioritizing compliance and safety,the industry can improve the qua

264、lity and reliability of their software while also reducing the risk of legal and security issues.1 “Tabnine+Pieces for Developers is a win-win for your workflow,”Pieces for Developers,May 3,2023.These new-generation tools are now guaranteeing indemnity for use,with the ability to detect,tag,and labe

265、l code.This metadata will make generated code easier to track and manage.The future of tooling is likely to see consolidation over time,with companies opting for several comprehensive tools or tool chains instead of numerous specialized ones.Martin Harrysson,senior partner,Bay Area32Technology Trend

266、s Outlook 2024Job postings by title,201923,thousandsDemandNext-generation software developmentTalent and labor marketsThe number of job postings for next-generation software development peaked in 2022 and showed the most job demand of all the technology trends in that year.Unsurprisingly,2023 saw a

267、decline,but even with a 37 percent decrease in job postings,next-generation software development still scores the highest in job demand among the tech trends,with over 800,000 job postings.Postings across the board have declined compared with 2022,which is in line with the layofs seen predomi-nantly

268、 in the technology industry.In the near future,it will be interesting to see what the impact of gen AI will be on both the types of roles and demand for roles in next-generation software development.Talent availability,%share of postings requiring skillTalent availability,ratio of talent to demandSk

269、ills availabilityKey talent areas for next-generation software development are focused on DevOps,continuous integration,and cloud computing.Some skills are more plentiful(for example,DevOps and cloud computing),while others are harder to fnd(for example,continuous integration).1.50.31.10.4DevOpsCont

270、inuousintegrationCloudcomputingSoftwareengineeringInformationtechnologyDevOpsContinuousintegrationCloudcomputingSoftwareengineeringInformationtechnologyPythonPython0.459434137342.930Software engineerSoftware developerData engineerWeb developerProject managerSystems administratorSolution architectTec

271、hnical architect0250200150100502019202333Technology Trends Outlook 2024Adoption developments across the globeThe financial-services and technology,media,and telecommunications sectors have emerged as leaders in the adoption of next-generation software development.Investments are driven by the changi

272、ng compliance landscape and the availability of more customizable tools.Adoption dimensionsFrom nascent to mainstream,the adoption trajectory will look different for each technology and even each use case within that technology.As AI-generated codethe most recent innovation within next-generation so

273、ftware developmentbecomes a standard way of working in the software development cycle,it provides an interesting example of how the adoption trajectory could develop.Advancements along the following dimensions could enable the next level of adoption:clear and measurable ROI of AI-generated code tool

274、s(for example,an increase of more than 25 percent in developer productivity for high-complexity tasks)2 a legal framework for liability of AI-generated code outcomes to create transparency on who bears the risk in case of malfunctions an increase in the applicability of AI-generated tools that can p

275、rovide sufficient performance for most common software development use cases implementation of AI-generated code as part of the core curriculum and upskilling programs for software developersIn real life Real-world examples involving the use of next-generation software development include the follow

276、ing:Citi leverages the Harness Continuous Delivery platform to provide an integrated experience across all stages of software delivery,with a user base of over 20,000 engineers.The platform brings together all the tools and services involved in software delivery with the aim of improving performance

277、,consistency,and maintenance across the enterprise while operating in the unique regulatory and risk management environment that comes with financial services.The platform helps streamline software delivery,automating deployment,testing,and change management after code approval.It facilitates faster

278、 rollbacks and integrates with observability tools for proactive issue detection and auto-rollback if needed.This translates into increased developer and operations control,reduced manual effort,and enhanced security.3 Goldman Sachs is exploring the use of gen AI tools to assist its software develop

279、ers in writing and testing code.The tools can automatically generate lines of code,freeing developers from repetitive tasks and allowing them to focus on core functionalities and client needs.2 “Unleashing developer productivity with generative AI,”McKinsey,June 27,2023.3 “Citi improves software del

280、ivery performance,reduces toil with Harness CD,”Harness,accessed on April 22,2024.Next-generation software development tools are fundamentally changing the role of developers,freeing up capacity for improved experiences and architectures,and ultimately,greater value creation.Santiago Comella-Dorda,p

281、artner,Boston34Technology Trends Outlook 2024Underlying technologiesThe technologies that power next-generation software development include the following:AI-generated code.AI applications can go beyond code suggestions and recommendations and also enable users to generate entire functions,optimize

282、existing code,create boilerplate code,and adapt to different programming languages.Low-and no-code platforms.Software development systems,such as Microsoft Power Apps and Google AppSheet,make it easier for nondevelopers to build applications more quickly.Infrastructure as code.This is the process of

283、 configuring infrastructure,such as a data center,with machine-readable code,which enables rapid reconfiguration and version control.The cloud,for example,is based on the concept of infrastructure that is fully abstracted as code.Microservices and APIs.These are self-contained,independently deployab

284、le pieces of code that can be coupled to form larger applications.AI-based testing.Next-generation software can use AI to automate unit and performance testing to reduce the amount of time developers spend on this task.Automated code review.These applications use AI or predefined rules that enable u

285、sers to check source code.Key uncertaintiesThe major uncertainties affecting next-generation software development include the following:Relying on automated testing and reviews without having humans check the work can lead to increased errors in software and erosion of user trust.The growth in the u

286、se of low-and no-code tools by nondevelopers could be limited because experienced developers are needed to monitor and debug applications.Comprehensive monitoring and version control could become more difficult due to uncoordinated changes and upgrades from multiple vendors.Quality and security rema

287、in concerns with code generated by AI pair programmers,particularly if they are not regularly updated with the latest standards or are not trained on clean,fast code.Addressing intellectual property,legal liability,and potential regulations surrounding gen-AI-generated code is essential for responsi

288、ble development and deployment.APIs add an extra layer of potential security vulnerabilities that can be exploited,and their customization can be a challenging task requiring substantial time and effort.Big questions about the futureCompanies and leaders may want to consider a few questions when mov

289、ing forward with next-generation software development:To what extent will the development of AI-generated code affect the day-to-day tasks and responsibilities,as well as number,of software engineers?To what extent might no-code tech used by amateur developers reduce the demand for fully trained sof

290、tware development professionals?From a cultural standpoint,will teamsboth developers and nondevelopersembrace or resist changes in ways of working?What intellectual property issues might affect AI-generated code?To what extent will business units take responsibility for the health of applications,or

291、 will accountability continue to rest with a shared IT function?Will organizations invest in the retraining needed to enable their software teams to adapt to the fast-changing domain?How do organizations upskill engineers to know what good outputs from AI-enabled tools look like?35Technology Trends

292、Outlook 2024Digital trust and cybersecurityThe trendand why it mattersDigital trust and cybersecurity enable organizations to manage technology and data risks,accelerate innovation,and protect assets.Moreover,building trust in data and technology governance can enhance organizational performance and

293、 improve customer relationships.In this trend,we include technologies that enhance trust(for instance,digital identity and privacy-enhancing technologies),cybersecurity capabilities(such as identity and access management),and Web3(such as blockchain).The importance of digital trust and cybersecurity

294、 is increasing as organizations adopt emerging and maturing technologies within their enterprises(for example,cloud and edge computing,applied AI,and next-generation software development).1 While the adoption of these emerging technologies comes with exciting new benefits,it also exposes organizatio

295、ns to cybersecurity and other risks,increasing the need for digital-trust technologies.The adoption of digital trust and cybersecurity,however,has been affected by a range of factors,including integration challenges,organizational silos,talent shortages,and its limited consideration as a critical co

296、mponent of value propositions.Capturing the full benefit of digital trust and cybersecurity will require top-down leadership and deliberate changes to multiple spheres of activity,from strategy and technology to enterprise capabilities.Talent demand Ratioof skilled peopleto job vacanciesEquity inves

297、tment Private-and public-market capital raises for relevant technologiesPatents Patentflings for technologies related to trendNews Press reports featuring trend-related phrasesSearches Search engine queries for terms related totrendResearch Scientifc publications on topics associated with trend0.10.

298、20.30.4NewsTalent demandResearchSearches0.5PatentsEquityinvestmentScoring the trendBUILDING THE DIGITAL FUTUREScore by vector(0=lower;1=higher)Digital trustand cybersecurityThe digital trust and cybersecurity market has experienced high growth over the recent years:the cybersecurity market growth ra

299、te in 2021 was 12.4 percent.But,as with other trends afected by the macroeconomic slowdown,the digital trust and cybersecurity trend took a hit in 2023,compared with 2022,across dimensions such as investment and talent demand.That said,the fve-year view(201923)shows robust growth across all dimensio

300、ns,and as the digitization of enterprises continues,this trend is likely to keep gaining traction.Industries afected:Aerospace and defense;Aviation,travel,and logistics;Consumer packaged goods;Education;Electric power,natural gas,and utilities;Financial services;Healthcare systems and services;Infor

301、mation technology and electronics;Media and entertainment;Pharmaceutical and medical products;Public and social sectors;Retail;TelecommunicationsAdoption score,2023$3434%12345FullyscaledFrontierinnovationEquity investment,2023,$billionJob postings,202223,%diference201920231.00.50Bharath Aiyer,Jefrey

302、 Caso,Peter Russell,and Marc Sorel,“New survey reveals$2 trillion market opportunity for cybersecurity technology and service providers,”McKinsey,October 27,2022.1 “The cyber clock is ticking:Derisking emerging technologies in financial services,”McKinsey,March 11,2024.36Technology Trends Outlook 20

303、24Latest developmentsRecent developments involving digital trust and cybersecurity include the following:Managing generative AI risk and readiness has become a key focus.The rise of generative AI(gen AI)has sparked innovation across industries while also heightening the focus on managing its associa

304、ted risks.Key concerns include fairness and bias,as gen AI can perpetuate existing biases embedded in training data.To counter these concerns,companies like IBM are creating fairness tool kits to identify and remove bias within AI models during the development process.Privacy issues arise as a resul

305、t of gen AIs ability to create deepfakes,prompting research into watermarking AI-generated content.The potential misuse of gen AI for cyberattacks underscores the importance of robust AI security frameworks.Intellectual property(IP)rights over gen AIs creative output remain unclear,and challenges ar

306、ound the explainability of gen AIs outputs hinder trust.President Bidens executive order on gen AI,calling for research into these risks and the development of trustworthy AI standards,and the recently adopted EU AI Act create pressure for responsible deployment and will likely lead to the adoption

307、of new tools,such as those coming from emerging players like Credo AI and Holistic AI.Cybercriminals and threats are evolving at a rapid rate.Threat actors,including cybercriminals and state-sponsored groups,are becoming more sophisticated.Their attacks exploit new vulnerabilities(for example,intric

308、ate ransomware that is debilitating power grids)and aim for maximum disruption(for example,targeting industrial control systems).Unfortunately,current security systems and company readiness are often not at the level needed to deal with this increased cybersecurity risk.New buyers are emerging outsi

309、de of the CISO role.Responsibility for cybersecurity is expanding beyond the office of the chief information security officer(CISO),with cyber spend now increasingly coming from nonsecurity business functions such as product and engineering.2 Consequently,cybersecurity providers must adapt their str

310、ategies to use cases with a wide range of stakeholders,including and stretching beyond the CISO office.Improving cybersecurity maturity,increasing efficiency,and possibly increasing the use of AI-enabled automation remain key growth drivers.The ongoing debate between cybersecurity platforms and best

311、-of-breed solutions is evolving.Cybersecurity platforms offer a unified environment,simplifying management but potentially compromising on functionality.Conversely,best-of-breed solutions offer specialized tools but can struggle with data integration and user experience.We see a shift as the lines b

312、etween platforms and best-of-breed solutions are blurring,with platforms becoming more modular and integrating best-in-class security tools.The market is at an inflection point in the“best of breed”versus“best of suite”debate:customers have not reached a consensus on a preference in any segment.Smal

313、ler companies might favor the simplicity of platforms,while larger ones may value the customization offered by best-of-breed solutions.The best path lies in balancing comprehensive security with manageable complexity while considering the companys security maturity,IT staff skills,and growth prospec

314、ts.Bitcoin and Ethereum ETFs are sparking mainstream interest.After a period marked by regulatory challenges for crypto exchanges,multiple Bitcoin exchange-traded funds(ETFs)have been approved.This has effectively lowered the entry barrier,opening up the world of cryptocurrencies to a wider audience

315、.In addition to Bitcoin,Ethereum ETFs are also gaining traction.Several Ethereum ETFs are currently awaiting approval,indicating a growing interest in diversifying cryptocurrency investments.While these developments have significantly influenced the digital-asset market,it remains volatile.Blockchai

316、n companies are moving from piloting to at-scale deployment of tokenized financial assets.Tokenization,the process of creating a unique digital representation of an asset on a blockchain network,has started to scale after many years of promise and experimentation.The benefitsincluding programmabilit

317、y,composability,and enhanced transparencycan empower financial institutions to capture operational efficiencies,increase liquidity,and create new revenue opportunities through innovative use cases.However,as infrastructure players pivot away from proofs of concept to robust at-scale solutions,many o

318、pportunities and challenges remain to reimagine the future of financial services.2 A recent McKinsey survey on cyber market customers(n=200)asked respondents,“In your best estimation,how much of your cybersecurity spend comes from outside of your CISO organization?Where does that non-CISO cyber spen

319、d come from?”37Technology Trends Outlook 2024Security analystSoftware engineerSecurity engineerSoftware developerData engineerProject managerNetwork engineer01020304051525354520192023Job postings by title,201923,thousandsDemandDigital trust and cybersecurityTalent and labor marketsJob postings for d

320、igital trust and cybersecurity decreased by 34 percent between 2022 and 2023.But in the longer-term view,we saw an increase of 123 percent between 2019 and 2023.Security analyst remains the highest-demand job for digital trust and cybersecurity,followed by software and security engineers.Talent avai

321、lability,%share of postings requiring skillTalent availability,ratio of talent to demandSkills availabilityCompanies expanding their digital trust and cybersecurity initiatives have a strong demand for skills associated with security,compliance,and risk analysis.Despite the short-term decrease,the d

322、emand for relevant skills still generally outpaces supply(except for blockchain),and the talent gap is signifcant.0.10.40.40.40.24.8InformationtechnologyComputersecurityRiskanalysisStakeholdermanagementBlockchainIdentitytheftRegulatorycomplianceInformationtechnologyComputersecurityRiskanalysisStakeh

323、oldermanagementBlockchainIdentitytheftRegulatorycompliance0.13831231917151438Technology Trends Outlook 2024Adoption developments across the globeThe digital trust and cybersecurity trend has seen high adoption levels among our trends,with some subcomponents achieving widespread use,while others rema

324、in at the forefront of innovation,such as emerging Web3 applications.About 30 percent of survey respondents reported that they had either fully scaled or were scaling digital trust and cybersecurity,and more than 60 percent mentioned they had invested in it.Financial-services companies,in particular

325、,have adopted this trend,driven by a need to combat an increasing range of threats and meet regulatory requirements.Telecommunications,media,and technology companies are also at the forefront of adopting digital trust and cybersecurity.This is likely because they are leading the way in enhancing sec

326、urity measures,particularly in the realm of AI,and developing effective tools to address the constantly evolving threat landscape.Companies of any size need to consider how to optimize their defenses as cyberthreats and regulatory and customer pressures increase.Adoption dimensionsThe adoption traje

327、ctory for advanced technologies varies for each technology and each use case within that technology.Advancements along the following dimensions could enable the next level of adoption for digital-trust and cybersecurity technologies:new digital identity systems integrated and scaled into existing pe

328、rsonal-identification processes enhanced integration of advanced technologies into existing cybersecurity frameworks,including upgrading midmarket companies defenses strong protection mechanisms to ensure user privacy and control of personal data improved government and public perception of the bene

329、fits and risks of digital identities security capabilities to meet varying regulatory requirements,ensuring compliance and fostering trust innovative applications with tangible real-world implications for Web3 to continue expansion beyond decentralized finance as practical applications emerge across

330、 various sectorsfor example,the decentralized physical infrastructure network(DePIN),still in its early stages,which aims to enable cell phones to function on a decentralized networkIn real life Real-world examples involving the use of digital trust and cybersecurity include the following:Salesforce

331、 built its Einstein Trust Layer specifically to address security concerns about using large language models(LLMs)within the Salesforce platform.This innovative system acts as a secure intermediary for Salesforce users interacting with LLMs.The Einstein Trust Layer ensures data confidentiality and pr

332、ivacy by masking personally identifiable information(PII)before it is used as input for the LLM and by adhering to a zero-retention architecture,meaning none of the Salesforce data is stored outside the platform or used to train the LLM itself.Additionally,the Trust Layer monitors outputs for inappr

333、opriate content and streamlines communication between the user and the LLM.Cisco created a customer-facing trust portal called the Cisco Trust Portal.This self-service tool provides customers with on-demand access to a wide range of documents related to security,trust,data protection,and privacy compliance.The purpose of the Trust Portal is to assist customers in gaining a deeper understanding of