1、The art of customer-centric artificialintelligence How organizations can unleash the full potential of AI in the customer experience Introduction As the COVID-19 crisis grips the world, the impact of AI on the customer experience is accelerating. As customers seek contactless or non-touch interfaces

2、, numerous sectors have stepped up their digital and AI game: Consumer products and retail has seen the use of AI-powered delivery robots to get groceries and medicines to people subject to shelter-in-place orders.1 In automotive, dealerships and OEMs have moved to digitize showrooms offering a full

3、y digital customer experience powered by augmented reality and AI.2 In the public sector, AI is a driving force in health and safety, from using augmented reality glasses to check the temperature of hundreds of people in a matter of minutes,3 or recognizing whether commuters on public transport are

4、wearing face masks or not.4 These innovations reflect how much progress has been made in AI as a technology over recent years. In late 2018, for example, Nvidia demonstrated that a class of AI algorithms “Generative Adversarial Networks” had not only become better at generating realistic human faces

5、, but that similar algorithms had also improved at discerning real images from any fake, AI-generated ones.6 In November 2019, Google made its contact center AI technology open and freely available.7 In natural language understanding, Google Translate now translates more than 100 languages into Engl

6、ish, with accuracy of translation improving remarkably across all languages in the last four years.8 AI capabilities have been evolving, but COVID-19 has accelerated adoption of these tools and made intelligent machines part of our new normal lives, both now and in the future. As our research shows:

7、 More than half of customers (54%) use AI daily compared to just 21% in 2018. Over three-quarters (77%) expect to increase the use of touchless interfaces such as voice assistants and facial recognition to avoid direct interactions with humans or touchscreens during COVID-19, and 62% will continue t

8、o do so post-COVID. This latest research builds on an earlier study we conducted in 2018. Back then, while we found significant customer appetite for AI interactions, we also unearthed two significant challenges. First, customers want and expect “humanized” experiences from their AI interactions wit

9、h organizations. Second, many organizations were not focusing on enhancing customer experience with their AI efforts instead, they were focused on cost optimization. To understand what progress has been made in tackling these challenges, and the progress that organizations have made with the AI-powe

10、red customer experience, we launched a wide-ranging research: A survey of more than 5,000 customers across twelve countries: Australia, Brazil, China, France, Germany, India, Italy, Netherlands, Spain, Sweden, the UK, and the US. We also undertook focus group discussions with customers in the US and

11、 Germany. A survey of more than 1,000 industry executives across eight major industries banking, insurance, consumer products, retail, automotive, utilities, government/public services, and public sector enterprises. We also conducted in-depth interviews with industry executives. More details on the

12、 research methodology are in the appendix. Drawing on that research, this report examines several themes: How have customers wants and needs evolved and have AI interactions lived up to their expectations? To what extent have the potential gains from the use of AI materialized? How have organization

13、s adapted AI applications given changing customer needs and increasing concerns about the need to avoid issues such as bias and discrimination in AI systems? How have customers been using AI in a COVID-19 world and how can organizations meet their needs? 2The art of customer-centric artificial intel

14、ligence What is artificial intelligence? Artificial intelligence (AI) is a collective term for the capabilities shown by learning systems that are perceived by humans as representing intelligence. These intelligent capabilities typically can be categorized into machine vision and sensing, natural la

15、nguage processing, predicting and decision making, and acting and automating. Various applications of AI include speech, image, audio and video recognition, autonomous vehicles, natural language understanding and generation, conversational agents, prescriptive modelling, augmented creativity, smart

16、automation, advanced simulation, as well as complex analytics and predictions. Technologies that enable these applications include automation, big data systems, deep learning, reinforcement learning and AI acceleration hardware. this drops to 39% for post- purchase interactions such as returns. Figu

17、re3: More customers prefer AI-only engagements in the early stages of their customer journey researching, browsing, and purchasing Source:Capgemini Research Institute, AI in Customer Experience Customer Survey, AprilMay 2020, N=5,300. 41% 38% 25% 29% 32% 21% 18% 27% 16% 26% 48% 25% 26% 14% 18% 14% 1

18、4% 11% Percentage of customers preferring AI-only interactions with organizations (2020 compared to 2018) Percentage of customers preferring AI-only interactions in 2020 - by age % of customers preferring AI-only interactions in 2020 While researching and browsing While making a purchase While using

19、 the product/service For after-sales support 18-30 31-40 41-50 51-60 60+ % of customers preferring AI-only interactions in 2018 % of customers preferring AI-only interactions for researching and browsing (2020) % of customers preferring AI-only interactions for making a purchase (2020) 7 Key factors

20、 that drive customer AI adoption: 1.COVID-19 is increasing customer adoption, driven by the appetite for non-touch-based interactions As a result of COVID-19, customers are increasingly looking for digital, no-touch connections with organizations, given the constraints and concerns about physical in

21、teractions in a new-normal, socially distanced world. Our research shows that even when lockdowns are lifted, customers across the world say they will still be looking to make increased use of touchless interfaces, such as voice interfaces, facial recognition, or apps (see Figure 4). The fact that t

22、ouchless interfaces are becoming integral to the customer experience in a health-and-safety conscious world is also recognized by organizations too. Three-quarters of organizations believe that increasing customer appetite for non-touch practices will persist even in the post-pandemic world. For ins

23、tance, India-based Bajaj Allianz points out to increase in their chat interface usage by customers during the pandemic, for registering claims, renewing policies or checking policy status and for locating hospitals.12 Figure4:Increase in touchless interactions during the pandemic and beyond Source:C

24、apgemini Research Institute, AI in Customer Experience Customer Survey, AprilMay 2020, N=4,818. I expect to increase my use of touchless interactions, through voice assistants, facial recognition, or apps, to avoid human interactions and touchscreens” 77% 90% 86% 84% 83%82% 80% 78% 77% 74% 71% 65% 6

25、1% 62% 68% 47% 55% 63% 48% 66% 63% 73% 63% 60% 69% 71% Overall China UK US Australia India Sweden Netherlands Germany France Spain Italy Brazil During COVIDPost-COVID 8The art of customer-centric artificial intelligence Figure5:In two to three years, 80% of organizations will have most of their cust

26、omer interactions enabled by AI Source:Capgemini Research Institute, AI in Customer Experience Executive Survey, AprilMay 2020, N=1,060. 2.Organizations step up AI deployments and look to transform the customer experience with AI. As Figure 5 shows, two to three years ago, most organizations (93%) h

27、ad less than 30% of interactions AI-enabled: Today, however, only 10% of organizations are at that low bar, with 80% saying that 30% to 50% of customer engagements are AI enabled. In two to three years time, the vast majority (80%) will have more than half of their interactions enabled by AI. Eight

28、in ten organizations will have more than half of their customer interactions enabled by AI in two to three years. Procter accountability because you take responsibility for what you are doing; and empowerment because you put people in charge to tell you if something you did was not right or not good

29、.” LucianoFloridi, Professor of Philosophy and Ethics of Information and director of Digital Ethics Lab, Oxford Internet Institute, University of Oxford.13 10The art of customer-centric artificial intelligence 3.Customers trust in AI is on the rise CustomertrustwithAIengagementshasincreased since201

30、8 Our research found that: 67% of customers trust the personalized recommendations and suggestions provided by AI enabled interactions. Close to half of customers (46%) find AI-enabled interactions to be trustworthy compared to 30% in 2018. The share of customers who say that they do not trust machi

31、nes with the security and privacy of their personal data has dropped to 36% today, down from the 49% who said that trust was lacking in 2018. Organizations also adopt customer trust as a key lever to increase AI adoption: 72% of the organizations we surveyed say that gaining customer trust is the mo

32、st powerful lever for increasing customer adoption of AI systems. Furthermore, with new data and privacy protection in place such as the GDPR people have greater reassurance than their data rights are protected and that companies are following agreed standards and rules when developing their systems

33、. Eric Chaniot, chief digital officer at Michelin says, “Michelin is fully compliant with the GDPR, which is very close to the French government regulations. And sometimes we go beyond even what they are asking for in the GDPR. So, I think our customers trust that we are going to do the right thing

34、with their data.” Customers increased trust appears to be driven by increasing fairness and transparency. Our research shows that 69% of customers trust the fairness of the decisions made by AI. Our research also indicates that organizations are increasingly making AI interactions more transparent f

35、or customers. For instance, while in 2018 only 13% of organizations informed customers about the presence of AI before the interaction begins, this increased to 66% in 2020. Along with consent, comes the importance of explainability, transparency, and driving a bias-free AI engagement. We look at it

36、 in detail later in the report. 4.Customers finding AI-enabled interactions to be more human-like. In our 2018 research,14 we found that customers want AI to display human-like capabilities such as such as human-like voice or personality or understanding. And, if interactions were more human-like, t

37、hey would be keener to use more of these AI applications and have greater trust in the company. And indeed, as AI exhibits more human-like features and capabilities, customers are adopting it more. Overall, 64% of customers believe that their AI interactions are more human-like (compared to 48% in 2

38、018). China (74%), Australia (72%) and US (70%) lead in the percentage of customers who believe that their AI interactions are more human-like. Organizations have been consciously trying to build in human-like features in AI applications: 72% of organizations agreed that they are actively trying to

39、make their AI interactions more human-like (compared to less than half in 2018). ”Mia” a virtual assistant developed by the National Australia Banks digital arm, UBank communicates with customers face-to-face and gives on-the-spot answers to more than 300 home loan application questions. Examples in

40、clude “whats a variable rate?” or “what classifies as an expense?” To support UBanks branding, Mia has been positioned as a “cheeky” personality, fond of gifs, animations, and jokes, and provides query resolution in customer-friendly language.15 Customers AI interactions have gone mainstream with or

41、ganizations increasing deployments and using levers of trust and human likeness. But as adoption increases, customers expect more from their AI engagements. Are organizations meeting these expectations? We delve deeper in this in the next section. 11 them to offer recommendations on products that Iv

42、e already purchased.” The participant adds, “I appreciate that it gives me recommendations and links. But at some point, I want to say, Im fine with that Ive moved on now, Im looking at new things. Our research also found that 51% of customers say they will consider an AI experience to be positive i

43、f it provides a unique experience beyond their expectations. However, organizations recognize the challenge of keeping up with evolving customer expectations as AI-enabled interactions increase. Sebastin Fuenzalida Garcs, program manager of artificial intelligence at Falabella Retail S.A. a leading

44、retailer in Latin America says, “When we deployed our chatbot, we started with some options. Then we were adding more features like add to or cancel your sales, store opening hours, etc. Now the chatbot displays all the information, but consumers need more focused resolutions. Currently, we are focu

45、sing on how to help consumers find all the available options and get to the one that they really need. From a country point of view, the share of satisfied customers has reduced across countries, more prominently in India, Italy, Australia, and France as compared to 2018 (see Figure 7). Figure6:Cust

46、omer satisfaction from AI interactions has fallen for all industries Source:Capgemini Research Institute, AI in Customer Experience Customer Survey, AprilMay 2020, N=5,300. Customer satisfaction levels have reduced since 2018 69% 57% 61% 58%58 % 54% 53% % of satisfi ed customers All Sectors (2018) A

47、ll sectors (2020) Banking and Insurance (2020) Automotive (2020) Public sector (2020) Consumer Products and retail (2020) Utilities (2020) Customersatisfactionlevelshavefallensince2018 Overall, 57% of customers are satisfied with AI interactions, compared to the more than two-thirds (69%) who were s

48、atisfied in 2018 (see Figure 6). As a US focus group participant says: “I have very high expectations of these AI systems once they put themselves out there. I dont expect 12The art of customer-centric artificial intelligence There are several issues that organizations need to confront and resolve i

49、f they are to push up satisfaction levels today: Customersmissthe“wow”factorinAIengagements: A significant number of people are unimpressed with the value they receive, with close to half (45%) feeling value is below what they had expected and 42% only marginally pleased (see Figure 8). Only 13% of customers said AI interactions offered much more value than what they expected. Clearly, AI is not yet exceeding customer expectations. By sector, consumer products and retail perform lags, with 55% feeling short-changed. The experienc