1、Conversational AI is reshaping the human-machine interaction November 2020 2 Contents What is CAIWhat is CAI 02 01 Customer Customer e experiencexperience04 03 05 06 CAI CAI d developmentevelopment Implementation Implementation j journeyourney Industry Industry a applicationspplications D.BotD.Bot i

2、 introductionntroduction Copyright 2020 Deloitte Development LLC. All rights reserved.3 What is Conversational AI ? 3 4 Conversational AI (CAI)combines natural language processing , AI, and machine learning to understand and respond to free-form text or voice in an engaging and personalized manner.

3、Healthcare Voice diagnosis Hospital guidance Education Speech teaching Speech assessment Finance Call center Interactive voice response Manufacturing Industrial robot Smart phone Voice assistant APP Smart home Intelligent lighting Intelligent kitchen Wearable device Smart watch Smart band Intelligen

4、t vehicle Speech navigation Voice control Typical scenarios Transform the ways we live and work Consumer application Business application 5 Generate quick responses to reduce waiting time and processing time Reduce operational costs such as labor cost and time cost Answer questions and process tasks

5、 in a uniform format Reduce human errors with more consistent services Use CAI to replace manual operations enabling employees to focus on more creative tasks Interact with customers smoothly and efficiently creating better customer experience More diverse interaction methods can be achieved, such a

6、s text, voice, gesture, etc. CAI can be applied to scenarios where the traditional interaction methods could be inconvenient, such as self- driving CAI will change ways how human and machine interact, optimize task-based working processes, and can tremendously reduce the time for information retriev

7、al Being a productive personal assistant, CAI will augment human capabilities in the future of work Efficiency Standardization Experience Diversity Personalization Benefits of CAI in organizations 6 CAI development 6 7 CAI growth in applications CAI plays a major role in the applications of Artifici

8、al Intelligence Facial Facial r recognitionecognition CallCall- -Center Center virtual virtual c customerustomeragentsagents ChatbotsChatbots Fraud Fraud analysis analysis on on transactional transactional d dataata Process Process optimization optimization Market/Consumer Market/Consumer segmentati

9、onsegmentation Virtual Virtual personal personal a assistantsssistants Smart Smart roboticsrobotics 45% 42% 39% 37% 37% 37% 17% 17% Speech-activated applications have been widely adopted in the field of AI, such as call-center virtual customer agents, chatbots, virtual personal assistants, and smart

10、 robotics. Source: 1. Market Guide: China AI Startups, Gartner, November 2019 2. Global AI Development Whitepaper, Deloitte, September 2019 CAI penetration in leading AI industries Low High High Finance Education Healthcare Government Smart City Self-driving Communications Retail Market size Penetra

11、tion (Percentage of China respondents) CAI has a higher penetration rate in finance, education, government and healthcare among applications across industries. Leading use cases for CAI in AI deployment (Penetration: industry application degree; Market size: marketing opportunity) 1 2 8 Extensive ap

12、plications of CAI have emerged in consumer market. CAI brings value enhancement to various industries. Users hold more positive attitudes toward CAI applications. The development of chips and cloud computing shows the trend of integrating with AI. With the development of edge AI chips, CAI will find

13、 its way into mobile devices. Cloud computing enables enterprises and governments to offer more personalized and intelligent services and products. Deep learning has made great breakthroughs in speech recognition, natural language processing and speech synthesis. In the future, it is possible to rea

14、lize barrier-free human-machine emotional communications. Algorithm User demand Policy Computing power Machine learning, deep learning and other technologies build solid foundation for CAI CAI is transforming the ways we live and work Development of chips and cloud technology has fueled the basic co

15、mputing power to CAI 1 2 3 4 Policy is a catalyst for the future of AI growth Policy has gone through three phases, from single products to data-driven innovation platforms, from individual actions to national strategies, and most importantly, from AI technology development to the integration of AI

16、and the real economy. Four drivers of CAI growth 9 Prior to 2011 when deep learning, big data and cloud computing were not integrated with speech recognition technology, the accuracyaccuracyof speech recognition was 54.61%54.61%. In 2011, Microsoft introduced deep learning into speech recognition, w

17、hich improved the accuracyaccuracyof speech recognition to 81.55%81.55%. Before 2011, voice applications developed by Microsoft and Google were only based on basic grammar basic grammar analysisanalysisand machine translationmachine translation, mainly in information retrieval and extraction. After

18、2011, NLP text analysis was advanced towards deep deep understandingunderstanding,making dialogue robots more practicalpracticaland scene scene orientedoriented. With the application of pre-training language model in NLP task, NLP technology started to focus on emotional emotional text analysistext

19、analysisand text text reasoningreasoning. Before 2011, first-generation voice assistants were mainly used for navigating on PCs and information retrieval in a monotone voicemonotone voice. After artificial neural network had been used in speech synthetics technology, voice assistants started to imit

20、ate human human intonationintonation. In 2017, Tacotron 2, a speech synthesis system released by Google, was developed as close as human voicehuman voiceand became a benchmark system. In 2017, the accuracyaccuracy of Microsofts switchboard reached 94.9%, 94.9%, surpassing human for the first time. D

21、eep Deep learning learning o outbreak utbreak p perioderiod Deep Deep learning learning e exploration xploration p period eriod Traditional Traditional machine machine l learning periodearning period Speech Speech recognitionrecognitionNatural language Natural language p processingrocessingSpeech Sp

22、eech synthesissynthesis AccuracyAccuracy Text Text a analysis nalysis Emotion Emotion Algorithm 10 200620072008201020132015 20162017 I a a S In 2006, Amazon created AWS, providing a variety of cloud-based services on IaaS including storage and computation. Before 2007 when AI was in the early stage,

23、 CPU chips were sufficient to provide enough computing power. C P U P a a S Salesforce pioneered the enterprise PaaS market when it launched the Force platform in 2008. S a a S After 2010, SAP, Oracle and other traditional software companies began to launch cloud services, and enterprise SaaS develo

24、ped rapidly. G P U After 2013, GPU were widely used for AI. A S I C In 2015, Google first released the ASIC chip tpu1.0, and the industry started to develop special chips for AI. FPGA came out after Intel acquired Altera in 2015. F P G A In 2017， Huawei Qilin 970 became the first mobile phone AI chi

25、p, introducing AI into mobile devices. In 2016, AWS officially launched its own AI product line, and cloud computing started to show the trend of integration with AI. The continuous chip evolution integrated with cloud computing Bringing AI to the device: Edge AI chips come into its own. Edge AI chi

26、ps will find the way into consumer devices and enterprise markets, such as smartphones, smart speakers, wearables and robots, cameras, sensors, and other IoT devices in general. Services become more intelligent: virtual computing platform greatly improves the data processing and reduces the cost of

27、using data. Enter the era of customization: based on the extensive collection of information, user behavior and needs are accurately pinpointed. Chips Cloud AI + Edge Computing AI + Edge Computing AI + Cloud AI + Cloud Computing power The increasing computing power releases the potential of AI algor

28、ithms 11 Single product Innovation platform Individual action National strategy Real Economy 2015.05 Made in China 2025 Developintelligent equipments and products, and promote intelligent production processes 2017.03 Government work report MentionAI for the first time 2016.05 “Internet+” AI three-ye

29、ar action plan By 2018, create basic artificial intelligence resources and innovation platforms 2017.07 New generation of artificial intelligence development plan AI become the national strategy 2017.10 The report of the 19th national congress Promote the close integration of Internet, big data, AI

30、and the real economy 2017.12 Three-year action plan to promote the development of a new generation of Artificial Intelligence industry (2018-2020) Push further integration of AI with the real economy 2016.07 The 13th five-year plan for national technology innovation Focus on the development of AI te

31、chnology driven by big data 2016.09 Special action for innovation and development of intelligent hardware industry Focus on developing intelligent equipmentsand service robots 2018.09 List of innovation projects of deep integration of AI and the real economy Promote the integration of AI and the rea

32、l economy 2019.03 2019 Government work report Transform R&D results into applications and establish a data-driven smart economy 2019.03 Guiding opinions on promoting the deep integration of AI and real economy Establish data-driven smart economy and boost innovation vitality Real economy integration

33、 Policy Technology development 12 81.6% Mobile phone 58.0% Robot 73.6% Television 25.2% Speaker 65.1% Car 40.7% Watch 56.8% Air conditioner 47.8% Refrigerator Users expectation of CAI for equipment On smartphones, CAI related applications that users expect to see are information inquiry, weather bro

34、adcast, function setting and chatting. On televisions, voice control switch and timer setting are the top two desired CAI applications. In terms of intelligent vehicles, applications with conversational features expected by users are mainly voice navigation and control system. Users hold a more posi

35、tive attitude toward CAI Why Conversational AI in consumer market ? Hands-free Improve the quality of life Enhance human-machine interaction User demand - consumer Source: Conversational AI Whitepaper, IDC, 2017 13 Why Conversational AI in business market ? Improve operational efficiency Reduce labo

36、r cost Enhance quality of services Transform the ways we work Top AI benefits selected by organizations across the globe Making process more efficient: free up hands, change working paradigms and replace the tedious and high-cost operations with machines. Improving decision-making: find hot spots fo

37、r the market and provide data support for follow-up planning. Enhancing existing products and services: transform human-machine interaction to provide customers with more personalized experience. CAI plays key role in AI benefits 0%10%20%30%40%50% Reducing headcount Lowering costs Enabling new busin

38、ess models Enhancing relationships with client Making employees more productive Enhacing existing products and services Improving decision-making Making process more efficient User demand - business Source: Thriving in the era of pervasive AI, Deloitte, July 2020 14 Industry specific commercial appl

39、ications 14 15 Applications of CAI in financial services focus on front-end customer services Mostly adopted applications of CAI in financial service Application scenarios embedded CAI in different fields of financial industry Customer service Online virtual assistants Online virtual assistants play

40、 a complementary role to human agents. These virtual assistants would chat with customers, understand customers requirements and respond to their inquiries in an interactive way. Outbound call system Outbound virtual agents would take the initiative to call customers, screening out most of the repet

41、itive work for human assistant, such as tele-sales, customer feedback collection, and identity authentication. 1 Extensive analysis Find hot spots in the market With massive amount of speech information, identify most frequently used words and extract information valuable to potential commercial opp

42、ortunities. Provide data support for machine learning The corpus, generated from the speech information, greatly improves the machine learning database. Additionally, such database would provide reference for subsequent conversational interaction. 3 Speech navigation Speech navigation This applicati

43、on replaces manual press-button function with navigation via speech, making it easier for customers to navigate on the phone when calling to service center. 2 Full penetration of CAI into customer relationship management in financial services In the financial industry, hundreds of banks, security co

44、mpanies and insurance companies have launched intelligent customer service systems since 2017. Due to huge total investment and fund size, Robo- advisor can be expected. Industry specific commercial application financial services Labor cost reduction Improvement of customer service efficiency Data a

45、ccumulation for machine learning Data support for subsequent marketing Instantaneity and mobility Multi-channel of service Bank Insurance BankingInvestment Provide counter services inquiries such as deposit, transfer, loan and account settings Recommend appropriate insurance products according to us

46、er profile and experience Provide the latest price information of stocks and funds, and speculate about the future 16 Diagnosis Diagnosis Non-contact intelligent voice interaction In the operating room, doctors can retrieve the medical records, images and other relevant data of patients through spee

47、ch input. Medical robot Compare the patients disease description with the standard medical guide, and provide services such as guidance, medical consulting, self diagnosis, etc. At the same time, the intelligent robot can complete medical history collection work in advance through dialogue. Operatio

48、n Operation Reporting Reporting CAI has penetrated in every process of the hospital care CAI has penetrated in every process of the hospital care system system CAI applications have improved the healthcare services and alleviated the medical pressure caused by the shortage of CAI applications have i

49、mproved the healthcare services and alleviated the medical pressure caused by the shortage of medical resources with more efficient medical resources with more efficient solutionssolutions Speech generation of diagnosis report Integrate with each diagnosis report workstation to realize speech generation of diagnosis report, and transmit diagnosis results to patients and doctors in the form of voice output. Formation of electronic medical record Record the verbal consultation betw