1、CONTENTSExecutive Summary 1Part I: Building the Intelligent Organization 2The CEOs Role in Driving AISupporting Grassroots AI InitiativesAI Demands Leadership Focus on DataHow AI Drives Organizational ChangeBuilding Workforce Skills for AI 8Part II: AIs Implications for IT Leadership 9Sharpening Foc

2、us on Critical Enabling Technology CompetenciesAI Changes How Software Is Developed and DeployedHow Data Is Influencing Software Development PracticesSharing Responsibility for the Organizations DataPart III: How AI Raises the Stakes for Trust and Risk 14New Risks Call For New Risk Management Struct

3、uresThe Quest for ExplainabilityAI Puts Ethics Conversations on the TableWorking to Mitigate BiasFrameworks for AI and Data Ethics 19About the Research 20Acknowledgments 20Sponsors Viewpoint 21MIT SMR Connections develops content in collaboration with our sponsors. It operates independently of the M

4、IT Sloan Management Review editorial group.Copyright Massachusetts Institute of Technology, 2020. All rights reserved.MIT SMR CONNECTIONSExecutive SummaryMany leaders are excited about AIs potential to profoundly transform organizations by making them more innovative and productive. But implementing

5、 AI will also lead to significant changes in how organizations are managed, according to our recent survey of more than 2,200 business leaders, managers, and key contributors. Those survey respondents, rep- resenting organizations across the globe, expect that reaping the benefits37%14%20%22%7%of AI

6、 will require changes in workplace structures, technology strategies, and technology governance.Survey respondents represent 110 countries.Findings from each of the five global regions were consistent with the overall results.The energy for exploring AI is widespread: Nearly two-thirds of survey res

7、pondents reported increased spending on AI technolo- gies in the past year. However, for most, its still too early to realize benefits at scale. Less than half of respondents reported active adoption, with just 1 in 20 indicating that they have implemented AI broadly, while 18% have implemented AI i

8、n a few processes and 19% are running pilot projects.The responses from those who have implemented AI indicate that these initiatives have implications for general management and technology leaders in three significant ways:AI will drive organizational change and ask more of top leaders. The majorit

9、y of respondents to our survey expect that imple- menting AI will require more significant organizational change than other emerging technologies. AI demands more collaboration among people skilled in data management, analytics, IT infrastructure, and systems development, as well as business and ope

10、ra- tional experts. This means that organizational leaders need to ensure that traditional silos dont hinder AI efforts, and they must support the training required to build skills across their workforces.AI will place new demands on the CIO and CTO. AI implementation will influence the choices CIOs

11、 and CTOs make in setting their broad technology agendas. They will need to prioritize developing foundational technology capabilities, from infrastructure and cybersecurity to data management and development processes areas in which those with more advanced AI implemen- tations are already taking t

12、he lead compared with other respondents. CIOs will also need to manage the significant changes to software development and deployment processes that most respondents expect from AI.Many CIOs will also be charged with overseeing or supporting formal data governance efforts: CIOs and CTOs are more lik

13、ely than other executives to be tasked with this, according to our survey. As leading practitioners note, AI requires both quality data and ongoing support to improve the efficacy of its results and to achieve strong ROI.AI will require an increased focus on risk management and ethics. Our survey sh

14、ows a broad awareness of the risks inher- ent in using AI, but few practitioners have taken action to create policies and processes to manage risks, including ethical, legal, reputational, and financial risks. Managing ethical risk is a particular area of opportunity. Those with more advanced AI pra

15、ctices are establishing processes and policies for technology governance and risk management, including providing ways to explain how their algorithms deliver results. They point out that understanding how AI systems reach their conclusions is both an emerging best practice and a necessity, in order

16、 to ensure that the human intelligence that feeds and nurtures AI systems keeps pace with the machines advancements.The report that follows explores these findings in depth. Read on to learn more about the changes that leaders must prepare for to successfully implement trusted AI. lBuilding the Inte

17、lligent OrganizationArtificial intelligence (AI) is recognized as a technology of vital strategic importance to enterprise leaders. The majority of respondents to our global survey have begun the AI journey, reporting that they are in the planning stages or have already piloted or implemented the te

18、chnology. And organizations are stepping up their financial investments, with 62% reportingincreased spending on AI in the past year (see Figure 1, “Spending on Emerging Technologies”).Our survey of more than 2,200 leaders, managers, and contributors across a wide range of industries and geographies

19、 reveals how the drive to implement AI is reshaping organizational culture and processes and creating new mandates for CIOs and other technology leaders. It also provides insight into organizations perceptions of AI risk and ethical issues, and how those attitudes are affecting technology governance

20、.While AI promises to transform business and create value, for most organizations we surveyed, it is still too early to realize those benefits at scale. Just 5% are implementing AI widely across the organization, while 18% have implemented it in a few processes, and 19% are running pilot projects (s

21、ee Figure 2, “Most Are in the Early Stages of the AI Journey”). Although another 13% are planning AI adoption, the larg- est group, 27%, are still investigating it.The CEOs Role in Driving AITheres an opportunity to learn from the or- ganizations now implementing AI, and this report will look at how

22、 they differ from the rest. One distinction we noted: A commitment from the top appears to be key in driving AI forward. We found that those organizations broadly implementing AI are significantly more likely than others to have top management involved in identifying use cases. They are also more li

23、kely to identify the CEO and board as the primary champions for the introduction of new strategic technologies.Figure 1: Spending on Emerging Technologies4% 3% 11%82%19%1% 18%62%24%2%25%49%43%2%34%21%Cloud ServicesAI/machine learningInternet ofthingsBlockchainNot budgeted Decreased spending No chang

24、eIncreased spendingSpending on cloud services has increased for the largest number of survey respondents, followed by spending on AI/machine learning.Figure 2: Most Are in the Early Stages of the AI Journey5%18%19%1314%4%Implementing AI widely across the organizationImplemented AI in a few processes

25、Piloting AI projectsPlanning AI adoptionInvestigatingLeadership requires both taking action and set- ting an example. At the Toronto headquarters of Sun Life, an insurance and asset management company, President and CEO Dean Connor has set the expectation that leaders in the business should be able

26、to discuss architecture and tech- nology, and he devoted half a day of an annual executive summit to a data analytics boot camp.AI adoptionNot adopting and not consideringDont knowWhile 42% of respondents are actively engaging with AI in their organizations, only 5% of all respondents reported imple

27、menting it widely to date. Meanwhile, 40% are investigating or planning adoption.27%“AI requires a long-term philosophy. Companies that are going tosucceed in AI are the ones that have a long-term view, that understand that this is a significant competitive edge and this is a long-termRAY WANG, CONS

28、TELLATION RESEARCHinvestment. Everything else is just lipstick on a pig.”“It sent the message, If the CEO is discussing topics like our API strategy, you should be thinking about them too,” says Eric Monteiro, senior vice president and chief client experience officer at Sun Life.Peter Guerra, North

29、America chief data scientist at Accenture, cautions leaders not to see AI simply as the latest hot technology offering a quick payback. Instead, they should be thinking about the direction of their businesses and how AI can help solve prob- lems. “It should be something that drives everything you do

30、, meaning it underpins everything that you do,” Guerra says. “You dont focus on Im going to go do AI. You focus on Im going to do supply chain better, and Im going to leverage AI to do that.”AI also requires a long-term mindset that can be difficult to square with the pressures CEOs face, says Ray W

31、ang, principal analyst, founder, and chairman at Constellation Research. “AI requires a long-term philosophy. You have to understand that youre going to make this long-term investment with an exponential payoff toward the end. Leaders often dont understand how to do that, so they keep making short-t

32、erm decisions for earnings per share instead of thinking about the long-term health of the company,” he says. “Companies that are going to succeed in AI are the ones that have a long-term view, that understand that this is a significant competitive edge and this is a long-term investment. Everything

33、 else is just lipstick on a pig.”Supporting Grassroots AI InitiativesSucceeding with AI is not only about decisions at the top. Guerra says that the best-case scenario for adopting AI and delivering a positive long-term impact involves balancing leadership support and grassroots enthusiasm.AI Create

34、s New Opportunities for the MidmarketWe compared survey responses from organizations whose annual revenues range from $50 million to $499 million with those from larger enterprises and found that midmarket companies are less likely to be piloting or implementing AI, less likely to have increased the

35、ir spending on AI and cloud, and less likely to rate their foundational technologycapabilities as mature. What are the implications of these findings for midmarket organizations? One possibility is that midmarket playersthat have the commitment to innovate with AI have the opportunity to gain a comp

36、etitive advantage over their peers. By moving sooner, they can stay ahead of the AI maturity curve.“The best examples Ive seen have been grassroots, bottom-up efforts combined with a supportive, top-down drive,” Guerra says. “Where I havent seen AI be successful is when its been only top-down driven

37、 or when its only grassroots, bottom-up. That leads to complete chaos, with nobody setting priorities.”The kind of effort that combines top-down and grass- roots approaches is working at American Fidelity Assur- ance Company, says Shane Jason Mock, vice president for research and development at the

38、financial services and insurance firm. One of its successful projects used machine learning and robotic process automation to interpret and route customer emails for faster, more effective service.The companys R&D and innovation teams meet reg- ularly with its executive team to discuss business need

39、s that AI and machine learning can address. Its a two-way channel. Sometimes the executive team willWhats Working: Top Leadership Sets the Tone at Sun LifeAt Sun Life, President and CEO Dean Connor sets the expectation that executives in the business must understand informationarchitecture and techn

40、ology in order to engage as collaborators in the development of AI applications. He has even devoted a seg- ment of an executive summit to a data analytics boot camp.discuss problems, and Mocks group will suggest solutions; at other times, the R&D team will bring an idea for a new AI or machine lear

41、ning use case to the attention of executives, Mock explains.Creating an environment where the power of AI can be tapped throughout the organization is a cultural shift the U.S. Air Force is pursuing as it aims to democratize access to AI technologies, according to Capt. Michael J. Kanaan, cochair fo

42、r AI at U.S. Air Force headquarters. Applying AI across a global enterprise employing 450,000 people means looking for ways to use it across typical business functions warehousing, supply chain, and human resources as well as military-specific missions like aircraft opera- tions, intelligence, surve

43、illance, reconnaissance, and cyber operations.“Empowering everyone to execute their organizations missions with AI technologies,” as Kanaan describes the goal, demands a leadership commitment to creating new, cross-functional structures. The Air Force has done this by bringing together people from 2

44、6 support and mission operations. “We have in one room people who represent all of these things so that they have a say in the matter and can represent what they know best,” Kanaan says.AI Demands Leadership Focus on DataLeveraging AI for business advantage requires top leaderships commitment to man

45、aging data as a key asset, according to industry experts and AI practitioners we interviewed.“Being able to treat data as an enterprise asset means that there is somebody whose primary job is to understand where the data is and how it can be used to further the enterprise mission or goals,” says Mel

46、vin Greer, chief data scientist for the Americas at Intel.The experience of working with AI and machine learning had led to a profound change in the way Sun Life works with data, says Monteiro.“Its raised awareness of the importance of data in all our applications,” he says. When a business unit tea

47、m works to develop a new application, its members now think about the fact that other teams at Sun Life may be using the same data in the future. They know that the data has to be validated and its definitions must be clear. “Theyre worried now about the quality of data creation, even in the process

48、 of designing applications or changing applications, which just wasnt true in the past,” Monteiro adds.Treating data as an asset, and devoting resources to managing it, can represent a challenging shift in thinking for corporate lead- ers, says Astrid Undheim, vice president of analytics and AI at T

49、elenor Group, an Oslo-based telecommunications company serving the Nordic countries and Asia. “If we think about the big barrier to really succeeding with AI, the big job that needs to be done that executives may not truly understand is the need for good quality and high volumes of data. AI and pred

50、iction models often need completely different sets of data than what we have had before,” Undheim says.“And thats a huge task,” she adds. “Its not about the data that we have in our systems and how can we use AI to create value from that. The thinking is really, how can we change our processes? How

51、can we change the way we work, and how can we collect data that we“It was important to lead with asking, What is the objective, strategically, of our company? What questions do we want solved? If you say, Lets just start gathering data, you might not be going down the best path.”ROB STILLWELL, SEACO

52、AST BANKneed to build models for solving our most important business problems? Its not like you just hire experts, AI people, and they will do magic. Most often, you have to go all the way back to what data are you collecting before these AI experts can actually do their job.”Seacoast Bank recognize

53、d the importance of working on its data early in the machine learning journey, but first it identified its business objective: to gain intelligence that would help it to better serve customers.“It was important to lead with asking, What is the objective, strategically, of our company? What questions

54、 do we want solved? If you say, Lets just start gathering data, you might not be going down the best path. But if you start with, Whats our objective? then you can start getting the ap- propriate data to do it. In our case, only then did we develop the machine learning and everything on top of it,”

55、says Rob Stillwell, senior vice presi- dent and business analytics officer at Seacoast.With its objective set, the 93-year-old financial institution serving southern Florida worked to reinvent the way it collected and analyzed data about its customers so it could serve them better. For example, inst

56、ead of focusing separately on each of the various transactions the bank conducted, Seacoast created what Stillwell calls a holistic data set on the banks customers. The bank now uses machine learning models on this data to analyze the lifetime value of a customer relationship and target services at

57、increasing that value.“We worked on data for about two years before utilizing machine learning, because we knew how important it was to get the data part right,” Stillwell says, adding that the second element was determining how any data analytics or machine learning effort would match the banks str

58、ategy.How AI Drives Organizational ChangeThis expectation that AI requires a commitment to organizational change is supported by our survey respondents, who see AI as driving more change than other emerging technologies. Sixty-three percent of respondents overall said they expect AI to drive dramati

59、c or significant organizational change (see Figure 3, “AI Is Expected to Drive Organizational Change”). Among respondentsFigure 3: AI Is Expected to Drive Organizational ChangeAI/machine learningCloud servicesIoTBlockchainDont know No change Little change4% 4% 8%22%42%21%3%3%11%24%39%20%8%8%13%24%33

60、%15%19%15%18%21%18%9%Moderate change Significant change Dramatic changeSixty-three percent of respondents expect AI/machine learning to drive dramatic or significant change, slightly ahead of cloud and well ahead of IoT or blockchain. Note: Some percentages do not total 100 due to rounding.who are w