Artificial Intelligence in Europe Denmark Outlook for 2019 and Beyond

How 277 Major Companies Benefit from AI

REPORT COMMISSIONED BY MICROSOFT AND CONDUCTED BY EY

1 Contents

Preface Foreword 06 Executive Summary - ‘At a Glance’ ...... 08

Setting the Scene About this Report ...... 10 Rich Data ...... 13 Executive Perspective ...... 14 Participating Companies ...... 16 Bits & Bytes ...... 18 Follow the Money ...... 20 Case Study ...... 22 Experts ...... 23

Role of AI in European Business A Strategic Agenda ...... 27 Among Friends ...... 28 Push or Pull ...... 29 Ready, Set...... 30 AI Maturity Curve ...... 32 State Your Business 34 Case Study 36

Business Benefits and Risks Another World ...... 38 AI Here, There, Everywhere 40 Case Study ...... 41 Use it or Lose it ...... 42 Making AI simple ...... 44 Case Study ...... 47 Sector Benefits Landscape ...... 48 Risky Business ...... 51

Learn from the Leaders Capabilities ...... 53 AI Competency Model ...... 55 Advanced Analytics ...... 56 Data Management 58 AI Leadership 60 Open Culture ...... 62 Emerging Technology 64 Agile Development ...... 66 External Alliances ...... 68 Disclaimer Emotional Intelligence ...... 70 This report has been prepared by Ernst & Young LLP in accordance with an This report does not constitute a recommendation or endorsement by Case Study 73 engagement agreement for professional services with Microsoft. Ernst & Ernst & Young LLP or Microsoft to invest in, sell, or otherwise use any of Young LLP’s obl igations to Microsoft are governed by that engagement the markets or companies referred to in it. To the fullest extent permitted agreement. This disclaimer applies to all other parties. by law, Microsoft and Ernst & Young LLP and its members, employees and What’s next for you? agents, do not accept or assume any responsibility or liability in respect of This report has been prepared for general informational purposes only and this report, or decisions based on it, to any reader of the report. Should such How to Get Started 74 is not intended to be relied upon as accounting, tax, or other professional readers choose to rely on this report, then they do so at their own risk. Who to Contact from Microsoft ...... 77 advice. Refer to your advisors for specific advice. Ernst & Young LLP and Mi- crosoft accept no responsibility to update this report in light of subsequent ©2018 EY LLP Limited All Rights Reserved. Contributors from EY ...... 79 events or for any other reason.

2 3 As this report shows, companies in Denmark want more clarity and certainty about future laws and regulations before they bet on AI. We appreciate that new technology brings both opportunities and challenges. Microsoft is committed to working with governments, industries, and civil society to ensure that AI is developed, deployed and used in a responsible way.

—Brad Smith, President and Chief Legal Officer for Microsoft

4 5 Preface Preface

Foreword

Human Ingenuity Icon of Digital Success

Throughout history, embracing new technology has been crucial to in- The printing press, the automobile & the Internet are just a few technolog- novations made by some of the most visionary Danish companies – from ical achievements that have advanced our world. All were driven by human turning wind and solar power into energy, to building super long-haul ingenuity: our innate creativity that inspires us to learn, imagine & explore. cargo ships, to inventing new types of medical treatments, and so much This spirit is what pushes us to challenge the boundaries of the possible to more. go ever forward. Today, the beating heart of technology is data, offering us new possibilities Today, AI is helping to amplify our human ingenuity, opening up exciting to create smarter and more intelligent solutions. But raw data is useless new possibilities for how intelligent technology can shape our world. At if it’s not analyzed to help us discover meaningful patterns. This is where Microsoft, our goal is to democratize access to AI for everyone through in- Artificial Intelligence comes in: understanding and extracting value from novative & powerful platforms, & above all, we’re focused on ensuring that data to improve lives and transform businesses. our AI tools & technologies are deployed responsibly & earn people’s trust.

As shown in this report, Danish businesses are among the frontrunners in And yet, we realize that AI is one of the lesser understood modern tech- Europe when it comes to implementing and prioritizing AI. However, the nological break-throughs. Many questions remain. How are companies companies also have concerns before investing heavily in AI, pointing to applying this technology to empower employees, engage with customers, a lack of skills and a need for more clarity about data regulations. This re- transform their business and optimize their operations? Where are they flects a realistic understanding of the fact that AI is still a technology in its seeing benefits, and what are their blockers? early stages, undergoing rapid development and raising important legal and ethical questions. To provide answers, Microsoft commissioned this study to understand the AI strategy of major companies across 7 sectors & 15 countries in Europe. I believe Denmark has a solid foundation to be among the leading coun- It examines these companies’ readiness to adopt AI, how they rate the im- tries using AI, and I feel optimistic about our ability to continually benefit pact and benefits from AI implementations, and what they perceive as risks from the possibilities created by new technology. & keys to success.

We hope you find these insights inspirational for your own journey toward adopting AI & realizing its benefits for amplifying human ingenuity.

Vahé Torossian Marianne Dahl Steensen President, Microsoft Western Europe General Manager, Microsoft Denmark

6 7 Preface Preface

At a Glance

Noticeable potential for AI in many Danish companies slightly ahead compared to European peers corporate functions When looking across the 25 companies that have participated in the The most widely reported adoption of study in Denmark, it is clear that there are areas where they appear AI (47%) was in the IT/Technology func- slightly ahead on AI compared to the European aggregate. Across the tion, followed by R&D with 36%, and organizational hierarchical levels, more Danish companies report that AI customer service with 24%. Interestingly, is an important topic compared with the European aggregate. Further- several functions are hardly using AI at more, companies in Denmark self-report as being more mature – 96% all; most notably, the procurement func- of Danish companies are in the ‘piloting’ AI stage or beyond, compared While the hype of artificial intelligence Most impact expected from ‘opti- is almost as much as AI is expected to tion, where only 4% of the companies with 71% in the European aggregate. Expected impact is high as well: (AI) and its potential role as a driver of mizing operations’, with ‘engaging impact the core of these companies’ cur- currently use AI, followed by HR with 7% the vast majority of Danish companies report expecting AI to create transformational change to businesses customers’ as a close second rent business with 65% expecting AI to and product management with 9%. This some degree of impact across all business areas – core, adjacent and and industries is pervasive, there are have a high or a very high impact on the 89% of the respondents expect AI to is perhaps surprising, given the many new. limited insights into what companies core business. With AI presumably push- generate business benefits by optimiz- use cases and applicable solutions in are actually doing to reap its benefits. ing companies into totally new domains ing their companies’ operations in the these functional areas. This report aims at getting a deeper future. This is followed by 74% that ex- in the future, it is perhaps not surprising understanding of how companies cur- that AI is receiving attention as a key pect AI to be key to engaging custom- 8 key capabilities that are most im- rently manage their AI activities, and topic for executive management. ers by enhancing the user experience, portant ‘to get AI right’ What sets the most ‘AI mature’ how they address the current challeng- tailoring content, increasing response When asking the respondents to rank es and opportunities ahead. speed, adding sentiment, creating Very few of the 277 companies con- companies apart? the importance of 8 capabilities to ena- experiences, anticipating needs, etc. sider themselves “advanced” with AI ble AI in their businesses, ‘advanced ana- To get to the heart of this agenda, we Despite the apparent sizable impact that They expect AI will be beneficial in ‘empowering employees’ (76% lytics’ and ‘data management’ emerged received input from AI leaders in 277 C-suite respondents scored ‘engaging companies expect from AI, only a very of ‘more mature’ companies* vs. 42% of ‘less mature’ companies)*. as the most important. ‘AI leadership’ companies, across 7 sectors and 15 customers’ highest of the AI benefit small proportion of companies, consti- and having an ‘open culture’ followed. countries in Europe, via surveys and/or areas. Noticeably, 100% of the most ad- tuting 4% of the total sample, self-re- They report using a combination of structured and unstructured interviews. Below is the brief summary vanced* companies expect AI will help port that AI is actively contributing to data for AI (65% of ‘more mature’ companies vs. 15% of ‘less mature’ When self-assessing the capabilities of what they had to say. them engage customers, compared to ‘many processes in the company and companies), and data from both internal and external sources (68% where the companies are least com- only 63% of the less mature companies. enabling quite advanced tasks today’ of ‘more mature’ companies vs. 16% of ‘less mature’ companies). petent, they point to emotional intelli- AI is a “hot topic” - but more so on Using AI to ‘transform products and (referred to as ‘most advanced’ in this gence and AI leadership - defined as the C-level than in daily operations services’ comes out slightly lower with report). They expect AI will help them ‘engage customers’ (85% of ‘more (lack of) ability to lead an AI transfor- 71% of the companies respond that 65%, and ‘empowering employees’ mature’ companies vs. 59% of ‘less mature’ companies). mation by articulating a vision, setting AI is considered an important topic the lowest with 60% of the companies Another 28% are in the ‘released’ stage goals and securing broad buy-in across on the executive management level. expecting AI-generated benefits in where they have put AI selectively to ac- They see AI predominately being driven from a combination of the organization. This is significantly higher than on that area. tive use in one or a few processes in the technology push and business pull (61% of ‘more mature’ compa- the non-managerial / employee level company. The majority, 51% of companies vs. 32% of ‘less mature’ companies). To summarize, the challenge ahead where AI is only considered an impor- AI is expected to impact entirely nies, are still only planning for AI or are appears to be as much about culture and tant topic in 28% of the companies. new business areas in the future in early stage pilots. 7% of companies leadership as it is about data, analytics, * ‘More mature’ defined as companies that self-ranked as 4 or 5 on the maturity Interestingly, Board of Directors also are self-rated as least mature, indicating 57% of the companies expect AI to and technology. 5-scale, and ‘less mature’ defined as companies that self-ranked as 1 or 2. came out lower with ‘only’ 38% of have a high impact or a very high im- that they are not yet thinking about AI at respondees reporting that AI is impor- pact on business areas that are “entirely this stage. tant to their board. unknown to the company today”. This

Percentage of companies Share of companies that use that are still only in the Only 71% 57% acquisitions as a way to 80% 4% planning or piloting stages: of the companies of the companies obtain AI capabilities: of the most mature

of the companies are actively respond that AI is considered expect AI to have a high impact companies expect that AI using AI in ‘many processes ‘an important topic’ on the on ‘business areas that are will be beneficial by and to enable advanced tasks’ 61% executive management level entirely unknown today’ 10% only ‘empowering employees’

8 9 Setting the Scene Setting the Scene

About this Report What’s new?

Artificial Intelligence (AI) is not new. in AI, what they are investing in, and efit areas, how mature companies are Straight from the executives Contributions from open-minded and collaborative companies It has existed for decades: processing how they are managing the compli- in terms of adoption, and examining Where this report and extensive da- voice to text or language translation; cated process of adopting this new self-reported competence levels re- taset adds new insights is primarily We are extremely thankful for the time real-time traffic navigation; dynami- technology and deriving value across garding the capabilities required to into how leading companies are ap- and effort the many executives have cally serving targeted advertisements business opportunities. succeed when implementing AI. proaching AI on a very practical level. put into participating in interviews and based on personal data and browsing We hear straight from executives how providing data for this study. We’re par- history; predicting trends and guiding Perspectives, experiences, self- From the aggregate dataset we have their companies are addressing cur- ticularly appreciative of their willing- During the past few years, we investment decisions in financial in- assessment, and benchmarks been able to determine some bench- ness to openly share experiences and rent challenges, and how they apply AI have learnt what is easy, what stitutions. The current developments From new surveys, interviews and case marks across the covered markets, to unlock new value pockets. provide their perspectives on where is hard, what is realistic and have been fueled by an exponential studies gathered from approximately which we compare the individual the future is heading within AI. rise in computing power, increasing 277 companies, we provide a snapshot country with throughout the report. Based on the many interviews con- what is only hype. accessibility and sophistication of pow- of the current state of AI in 15 European The report also covers a full spectrum ducted, this report reveals some clear While this indicate a general interest in erful algorithms, and an explosion in of industry groups which tend to reveal the AI topic, it also speaks to the in- markets. This includes analyzing AI’s excitement and immense potential for — DNA the volume and detail of data available interesting insights. creasingly collaborative approach relative importance on the strategic using AI to bring new, improved prod- Telecommunications to feed AI’s capabilities. many leading companies are taking agenda, its expected impact and ben- ucts and services to market, create company exceptional experiences for customers when entering new technology do- Reality vs. hype and employees, and create ways to mains and embarking on journeys into unknown territories. Only recently started to see more operate that enhance performance widespread, scaled adoption of AI across the board. across sectors, value chains and eco- systems. Yet AI technology is quickly We learned that regardless of which approaching a point where it is be- use cases the companies pursue and coming a critical element in enabling the role that AI currently has, taking a companies across sectors to drive strategic outlook to assess the implica- revenue, increase profits and remain tions for the business and responding One of the key challenges is meeting the competitive. accordingly are increasingly seen as high expectations from the organization crucial for any executive agenda. We hear many people in numerous - AI is not magic, but takes considerable companies talk about AI. While the effort to successfully implement. hype is pervasive, not a lot of people fully understand its technological potential, where it can create value or — H. Lundbeck how to get started. This report aims at Pharmaceutical company providing a practical understanding of why European companies are investing

10 11 Artificial intelligence in Europe Setting the Scene

Rich Data Which sources of information is the study based on? When working with AI initiatives, it is important to focus on key business issues This report combines multiple sources We also present case studies of specific Recognizing and mitigating of data to answer the questions of why, companies, both local and internation- potential survey and interview bias where and how AI is currently being used al, to provide an understanding of what In terms of methodology, this report that benefit the whole and not just doing sub- in business. It provides an inside view they are doing with AI and why, draw- follows robust research design and across markets and sectors. It combines ing on lessons learned and obstacles to protocol. Doing so minimizes potential optimization at small scale. local and pan-European views, and adds overcome when putting AI to use for bias, but does not eliminate it, as it value through a quantitative perspective specific use cases and to derive value is inevitable in market research. One on how advanced companies are with on a strategic level. potential type is social desirability AI, and a qualitative perspective on how and conformity bias, as the topic of to develop the skills required to succeed Proprietary AI investment data AI receives lots of media and political — Novo Nordisk Pharmaceutical company with AI. We have received input from We have supplemented the prima- attention. Response bias, including over 300 people from 277 participating ry source input from the companies extreme responding, cultural bias, and companies. This has resulted in a range of with acquisition data from numerous acquiescence bias (“yea-saying”), are interviews and case studies as well as 269 sources, to take the pulse of the AI potential factors as we ask respondents company responses to our survey. investment market in Europe. These to self-report on their respective com- insights help provide a picture of the panies’ experience. Therefore, while Extensive online survey data from wider European AI ecosystem and its this report follows best practices, some business leaders in 269 companies development. bias is possible. We have surveyed people with a leading role in managing the AI agenda in all the AI expert perspectives Nonetheless, with the combination of companies that have contributed to the With this wider understanding of AI extensive survey data, interview data, study. This gives us an aggregate dataset start-up acquisitions, partnerships, and investment data, and expert perspec- that enables a perspective for each mar- investment funding, we outline how tives, we believe the report provides a ket and each sector, as well as compara- investments in AI are skyrocketing, solid foundation for an indispensable tive insights for the respective company where AI investment is taking place view of executive experience with – I don´t see why speaking openly about types, sectors, and countries in Europe. geographically, and which sectors are and future plans for – AI in business. making bets. As we are on the cusp of our ongoing AI intiatives should be a big Qualitative in-depth interviews with widespread change driven by AI, we senior business executives also reached out to AI experts from In addition, we conducted deep-dive academia for an outlook of AI technol- fuss. What really makes a difference from interviews to gain deeper, qualitative ogies going mainstream, and to gain insights into how AI is affecting the ex- an understanding of the macro scale a competitive perspective is a company´s ecutive agenda. Through conversations of business effects that they expect will with business leaders, we report on materialize when looking into a distant where they expect AI will have an impact, future. abilitity to execute. how important AI is to their current and future business strategies, what benefits they hope to realize from implementing AI, and which capabilities they believe are key to advance AI maturity in their — PFA Pensions and insurance company companies.

12 13 Setting the Scene Setting the Scene

Executive Perspective Large group of respondents Surveyed companies are well represented across Who are the respondents that have contributed to the study? with a specific AI/digital role each of the 15 European markets Organizational function of respondents Number of online surveyed companies per country in the online survey

The data approach used allows us to Functional diversity A combined annual revenue identify trends across industries and The respondents cover very different of $1.9 trillion countries based on input from various functions, of which the most common Participants come from both major Netherlands functional business areas. Consequent- are designated AI/digital department, listed companies, privately held com- 67 ly, we have captured a range of in- followed by IT, and strategy/general panies, and in some case relatively small sights, learnings, and perspectives from 60 Ireland management functions. This functional companies. In totality, they represent a Norway both strategic and technical points of diversity increases the breadth of the combined revenue of approximately view. report, with insights and perspectives $1.9 trillion. Despite covering a signifi- Sweden 22

covering widely different aspects of AI. cant part of total European business, our 20 Respondents predominantly in 45 21 selection criteria have also favored more Denmark 20 senior level positions Surveyed companies span niche oriented companies with extensive 39 Switzerland To ensure that these insights and per- multiple sectors AI experience and capabilities. 25 27 26 20 spectives are relevant at the executive The participating companies are spread 269 level, we surveyed and interviewed fairly evenly across seven sectors, with online survey Primarily listed companies 21 20 high-ranking officers with a responsi- the majority of companies belonging companies Italy represented in Danish data Austria in total bility for driving the AI agenda in their to Industrial Products & Manufactur- 5 The vast majority of respondents respective companies. With 60% of ing, followed by Financial Services, and 22 in Denmark are major listed com- 22 respondents being either part of top Transportation, Energy & Construction. panies or companies privately 15 Finland management or the executive man- 21 Services and Life Science are represented 20 held by foundations. All the par- agement team, their input is likely well to a lesser extent. Portugal

ticipating companies in Denmark Other attuned to the general perspective and United Kingdom, France & Germany overall strategic direction of the com- had a combined total annual reve- Digital/AI Luxenbourg & Belgium panies they represent. nue of over $138 billion in 2017. Spain Admin & Finance General IT/Technology General & Business Development R&D & Product Management Customer Service & Marketing General Management, Strategy More than 300 participants Majority hold a top management or executive position Number of participants interviewed Organisational level of person participating in the study and/or online surveyed in the study Seven major sectors covered in the study Representation of participating companies per sector category 30 of +300 are Danish companies C-suite/Executive 27% 24% + 9% 21% 17% 7%

Top Management Life Science Industrial Products Finance Services (non-executive) 33% 44% Pharmaceutical, Healthcare, Manufacturing, Banking, Insurance, Professional Services, Biotech Materials, Equipment Investments Hospitality, Public Services, Membership Organization

Management Level 37% 32% 13% 16% 17%

Employee (non-managerial level) 3% 0% CPR TMT Infrastructure Consumer Products Technology, Transportation, Energy, & Retail Media/Entertainment & Telecom Construction, Real Estate 15 European markets Denmark 15 European markets Denmark

14 15 Setting the Scene Setting the Scene

Indie Campers, Intesa Sanpaolo, ISDIN, ISS, Jansen AG, Julius Baer, 277 Companies Katoen Natie, KBC Group, Kemira, Kingspan Group, KLP Banken, Komplett, Kongsberg Gruppen, LafargeHolcim, LanguageWire, LEGO, LEO Pharma, Lerøy Seafood, Liga Portugal, L’Occitane, Lonza, A.P. Moller - Maersk, Acciona, Adamant-Namiki of Europe, Aegon, L’Oreal, Lusíadas Saúde, Luz Saúde, Länsförsäkringar, MAPFRE, Aena, Ageas, Agfa-Gevaert, Agrifirm Group, Ahlstrom-Munksjö, Merkur Versicherung, Metall Zug , Metro, Metso, M-Files, Millicom, AIB, AkzoNobel, Almirall, Alpro, ALSA, Amadeus, AMAG, Ambea, Mota-Engil, Mutua Madrileña Automovilista, Møller Mobility Group, APM Terminals, Aprila Bank, Arcelor Mittal, Ardagh Group, Neste, NH Hotel Group, Nilfisk, Nokia Corporation, NorgesGruppen, Arval BNP Paribas Group, Asiakastieto Group, Assa Abloy, Norstat, Novabase, Novartis, Novo Nordisk, Novozymes, Assicurazioni Generali, Atea, Audi, Austrian Airlines, Austrian Now TV, OBI, Oesterreichische Nationalbank, OP Financial Group, Federal Computing Centre, Autogrill, BAM Group, Barco, BASF, Opportunity Network, Orion, Paddy Power Betfair, Peltarion, BAWAG P.S.K, Baxter, BBVA, Besix, Bolloré, BTG, BUWOG, C&C Pernod Ricard, PFA, Philips, Planeta DeAgostini, Poste Italiane, Group, Campbells International, Capio, Carmeuse, Carnival Posti, PostNord, Proximus, Pöyry, Rabobank, Raiffeisen Software, UK, CEiiA, Cermaq, Chr. Hansen, Cirsa, City of Amsterdam, Raiffeisen Switzerland, Ramada Investimentos SA, Randstad, Rexel, Colruyt Group, Com Hem, Combient, Comifar Distribuzione, ROCKWOOL Group, Room Mate Hotels, Royal College of Surgeons in Constitutional Court of Austria, Coolblue, COOP Nederland, Ireland, S Group, Saipem, Saint Gobain, Sakthi Portugal, Salsa, Saxo Bank, Cosentino Group, Costa Crociere, Credit Suisse, Crédito Agrícola, Sbanken, SBB Swiss Federal Railways, Schindler, SEB, SGS, DAF Trucks, Danfoss, Danske Bank, Dawn Meats, DFDS, DNA, Siemens Mobility, SimCorp, Skandia, Solvay, Sonae, Sonae Arauco, DNB, DSM, DSV, Dümmen Orange, Dynamic ID, DAA, Edison, SpareBank 1 SMN, SpareBank 1 Østlandet, Sportmaster, Statkraft, EDP - Energias de Portugal, Egmont, EQT, Ericsson, Erste Group Stedin, Steyr Mannlicher, Stora Enso, Styria Marketing Services, Suomen Bank, ESB, ESIM Chemicals, Esprinet, Europac, Fazer, FDJ, Terveystalo, Swedbank, Swisscom, Taylor Wimpey, TDC, Teamwork, Federal Office of Meteorology and Climatology MeteoSwiss, Ferrovial, Telefónica, Telekom Austria, Telenor Global Shared Services, Telia, Fexco, Finnair, Fortum, Galp, Geberit, Genalice, Generali Versicherung, Tesco, Tetra Pak, The Navigator Company, TIM, Tine, Tokmanni, GetVisibility, Gjensidige Forsikring, Glen Dimplex Group, Globalia, TomTom, Tryg, TTS Group, TVH, Ubimet, UDG Healthcare, UniCredit, GN Store Nord, GrandVision, Grupo Antolin, Grupo Ascendum, Unilin, UPM, Vaisala, Valmet, Valora Group, Van Lanschot, Vattenfall, Grupo Codere Cablecom, Grupo Juliá, Grupo Nabeiro – Delta Cafés, Version 1, Visana, Vodafone Automotive, VodafoneZiggo, Voestalpine Grupo Pestana, Grupo Visabeira, GSK, GAA, H. Lundbeck, Hafslund, High Performance Metals, WABCO, WALTER GROUP, Western Bulk, Handelsbanken, Hera, Hostelworld, Husqvarna, IKEA Group, William Demant, Wind Tre, WIT Software, Wolters Kluwer, Zurich Ilmarinen Mutual Pension Insurance Company, Implenia, Impresa, Airport, Zurich Insurance, Öhman, Ørsted, Österreichische Post.

Danish companies All companies, excluding Danish companies Note: Of all contributing companies, 14 chose to be anonymous, 0 of them being from Denmark

16 17 Setting the Scene Setting the Scene

Companies are using a mix of Data Sources and Storage Companies are using a combination Solution: How are you primarily dealing with the computing demands Bits and Bytes of on-premise and cloud solutions needed for AI? Data Source: 1.Are you currently using unstructured or structured data Companies are increasingly using types in your AI process? 2.Are you currently using internal or external What technologies and data solutions are within the scope of the study? cloud-based AI solutions for both data sources in your AI process? storage and on-demand computing power - 83% of companies reporting using Cloud technology to some ex- AI can be defined as the ability of a not in common use by companies in While companies historically have tent to enable their AI capabilities. Key machine to perform cognitive func- Europe. Companies surveyed are cur- primarily have used internal data for benefits of cloud solutions mentioned tions which are normally associated rently focused on narrower and more supervised Machine Learning, many by many respondents are the flexibility with humans. This includes reasoning, specific use-cases that support existing have begun exploring the possibility of to swiftly scale systems up and down learning, problem solving, and in some business. These efforts will undoubt- combining internal and external data- to accommodate changing demand, a Solution cases even exercising human behavior edly help companies build capabilities sets in order to produce even deeper variable cost structure, and access to 27% 17% 56% such as creativity. that are necessary to deploy more insights. larger data sets. However, many com- In Cloud On premise Both advanced AI solutions in the future. panies are still relying on on-premise Advanced AI applications are not Machine Learning and Smart Robot- solutions, not least due to existing data yet widespread Machine Learning ics were found to be the most useful. infrastructure. AI holds the potential to transform The most commonly used AI technol- It is not clear from the study if this is because they are simply the most com- business in a radical way given its wide ogy among the surveyed companies Machine learning and smart mon starting points before deploying variety of use. Quite simply, business is Machine Learning. This is inarguably robotics most useful for Danish more advanced technologies, or if they leaders need to understand AI in order due to its wide-ranging applicabili- companies to grasp the opportunities and threats ty, making it relevant for a variety of also longer term hold the most wide On average, the underlying tech- the technologies pose. use-cases across the value chain. Of the and significant application potential. nologies that are most useful for different types of Machine Learning, Danish companies are concentrat- 32% 7% 43% While companies acknowledge the the most common is supervised Ma- ed in two areas: machine learning Structured Unstructured Both significant potential of broader, more chine Learning, where software is fed (84%) and smart robotics (52%). advanced AI technologies such as structured data and finds patterns that Additionally, 84% of Danish com- computer vision, speech recognition, can be used to understand and inter- Data Source panies selected more than one and virtual agents, they are currently pret new observations. type of AI technology.

A broad definition of technologies are included in this AI definition 38% 3% 44% Technologies included in the definition of AI used in this study Internal External Both

Text Analysis Computational analysis of texts, making it readable by other AI or Machine Learning and Smart Robotics found to be the most useful Natural Language Processing computer systems. Biometrics Computer interpretation, under- Analysis of human physical and Which of the following technologies have you found to be most useful in your company’s deployment of AI? standing, and generation of written emotional characteristics – used natural human language. also for identification and access control.

77% 44% 40% 39% 39% 26% 21% 20% 6% Virtual Agents Computer-generated virtual personas Machine Learning A computer’s ability to ‘learn’ that can be used to interact with people 84% 52% 40% 44% 44% 16% 0% 16% 4% in both B2C, C2B, and B2B contexts. from data, either supervised or non-supervised.

Speech Recognition Neural Networks and Deep Learning Enables computers to interpret spo- Machines emulating the human brain, ken language and to transform it into enabling AI models to learn like humans. written text or to treat it as commands Machine Smart robotics Natural Neural Text analysis Virtual agents Speech Computer Biometrics for a computer. learning language networks and recognition vision Computer Vision processing deep learning Gives computers the ability to Smart Robotics “see” images similar to how humans see. The combination of AI and robots to Affirmative responses, 15 European markets Affirmative responses,Denmark perform advanced tasks compared to traditional non-intelligent robots. Note: Remaining percent ‘Don’t know’ responses

18 19 Setting the Scene Setting the Scene

acquisitions, and is also much in line TMT most active, behind private Follow the Money with what we’re seeing when compar- Over $330 million invested in equity and venture capital ing with the US and Asia. AI companies in Denmark Investments into AI companies per sector, How much is invested in AI in Europe? Finland In Denmark, there were 21 trans- mUSD (accumulated 2008-2018)* $24M Investment activity concentrated in actions involving companies Norway 21 deals major European markets working with AI in the past dec- $30M It comes as no surprise that a lot of ade. Of these, 13 reported deal A few big AI transactions 5 deals influencing the overall picture investment activity is in the UK, France, value, which totaled $330 million, and Germany, having attracted 87% of implying the actual amount is Company AI investments in mUSD and $254M transaction volume per market 73 deals investment in AI companies over the even higher. A large portion of past decade. The UK leads significantly this amount was the purchase $7,453M (accumulated 2008-2018) Sweden 1,027 deals in this regard, with 533 of the total of Universal Robots in 2015 for Private Equity / 1,362 AI transactions in Europe. From $285 million. Of the AI compa- Venture Capital** an investment perspective, it is also nies in Denmark that received $330M* 21 deals worth noting that in April 2018, the EU investments or were acquired, committed to a 70% increase in invest- 52% focus primarily on machine Denmark $7262M ment in European AI by 2020, suggest- learning technology, likely due Ireland 533 deals ing further growth and potential in the to its wide applicability across a United Kingdom The Netherlands region. range of business problems and $39M $43M 37 deals 45 deals sectors.

$110M $520M 14 deals 140 deals Belgium Germany Steady increase in European AI investment AI companies invested into, transaction volume, Europe (from 2008-2018)** $1,843M 220 deals $107M TMT 31 deals $75M 17 deals $1357M Switzerland 165 deals Number of $494M Austria transactions 17 deals France Industrial Products

$3M 450 8 deals $368M 398 Portugal $47M 12 deals 29 deals 400 Infrastructure

European Denmark $131M Italy 350 327 markets 79 deals Total $254M UK bubble size not represenative 300 investment 21 deals $10.5bn Life Science *Universal Robots acquired for $285M Spain 250 228

The acquisition data from numerous alone. This trend is on track to con- tive investors and acquirers of AI than $70M corporates, accounting for 75% of deal 200 41 deals sources enabled us to explore the tinue, with an exponential increase Finance European AI ecosystem and gain in- in interest in AI driving more large volume in the last 10 years. This is an 148 150 sights into investment activity. companies to invest in AI or acquire AI indication that AI companies are in the early stages of high risk/high growth capabilities from innovative start-ups. 88 $38M dynamics. It also indicates that, for 100 An exponential increase in AI in- Of the 15 markets surveyed, some in- 64 10 deals vestment over the past decade clude one or two transactions that are large corporates, acquiring or invest- CPR ing in external AI businesses in order 50 29 27 Looking at AI transaction activity significantly large deals. 14 11 to obtain AI capabilities is relatively across Europe, there has been a steep $22M limited. This is confirmed by our survey 0 consistent growth trend over the past Majority of investments in AI from 14 deals results where only 10% of companies 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Services 10 years, totaling 1,334 transactions private equity and venture capital are seeking to obtain needed AI capa- involving AI by 2017 – with a six-fold Private equity (PE) and venture capital bilities through external investment or increase in activity in the last 5 years (VC) firms are significantly more -ac Europe **Including governmental investment

Note: Several transactions in the dataset did not have publically disclosed deal values, suggesting that actual total values are higher than what’s shown above *For all of Europe, 34 countries (not just the 15 markets focused on in this report)

20 21 Artificial intelligence in Europe ( Case Study ) Setting the Scene

Expert Perspective LEO Pharma What does the future look like according to AI analysts?

LEO Pharma has established a dedicat- Image recognition software allows LEO uses to identify underlying patterns. ed innovation lab that is tasked with Pharma to work towards being able These underlying patterns are then developing technological solutions to to conduct diagnosis of psoriasis via used to identify psoriasis on pictures improve the lives of psoriasis patients. applications, than can be fitted in ex- of unknown skin deceases. They are The solutions are not me- currently able to diagnose dicinal but rather focused with 91% accuracy, which We also spoke to a range of leading AI Agile culture enables AI task is to educate and improve un- on e-Health and add-on Image recognition software allows LEO they expect to improve experts from business and academia derstanding, from C-suite leadership devices. The Innovation Lab Pharma to work towards being able to further once more data is Culture was a recurring theme as well. to gain insights into the kind of change teams to employees at the coal face. has been established as an gathered. At this stage the It can either stifle forward momentum conduct diagnosis of psoriasis via applica- which we are on the cusp, and the This also ties in with the importance of independent unit acting as process takes some time, in organizations, or be the silver bullet tions that can be fitted in existing smart role AI is expected to play as part of a partnering to get started and access a smaller, more agile organ- since the data must be clini- that enables the potential of AI to be broader transformational wave. the expertise needed to use AI. While ization, where innovation devices with a camera. It is currently able cally validated by a team realized from top to bottom. partnering and collaborating solves can thrive and solutions be to diagnose with 91% accuracy of doctors who review and AI is entering the mainstream the perennial AI challenge concerning tested and brought to mar- classify the pictures - i.e. Some of the experts even argue that and here to stay the scarcity of talent, the significant ket faster. The lab has been label the data for machine it’s not only technical skills that hold cost and substantial benefit that can working with artificial intelligence for isting smart devices with a camera. The learning purposes. Furthermore, they One thing was clear from the experts up AI projects, it’s also the need for a be gained from AI means that organ- two years, and it has grown to become image recognition software uses su- have entered into a data partnership we spoke to: as far as the peaks and culture of experimentation. izations also need to be cognizant of the Lab’s top priority, focusing primari- pervised learning in which the software to speed up the process of gathering troughs of hype and technological building capabilities in-house for the ly on image recognition. is given pictures of psoriasis, which it data. leaps surrounding AI go, there is no Companies that are more natively doubt that we are living through a digital or have gone down that road long-term. particularly prominent peak, with no understand the value of experimenting indication that the buzz nor the po- and iterating. They don’t think in tra- Finally, as AI develops, we are also tential will fade away any time soon. In ditional terms of committing to year- going to see innovation and expertise a world increasingly dominated, dis- long projects that need to produce spreading outside of the dominant What next? rupted and driven by innovative tech specific outputs, but rather to explore clusters of the likes of Silicon Valley, powerhouses, large and small, it is no and test ideas before scaling. as governments, businesses and uni- In the future, LEO Pharma will leverage AI to continue to develop understatement to suggest that AI will versities increasingly invest in building virtual treatment solutions and services that complement its be a chief protagonist in the change When it comes to AI, knowledge, resources and capabilities. current product portfolio. Image recognition capabilities will LEO Pharma is a Danish innovation driven pharmaceutical transcending all elements of business knowledge is power allow patients to be diagnosed via smart-connected products company privately held by the LEO Foundation. TThe company in what has been labelled the Fourth Expert opinion also seemed unani- that can prescribe treatments and monitor progress. This will develops, produces and markets products within treatment of Industrial Revolution. mous in that most people not directly create a continuous feedback loop of outcome data, which is psoriasis and other skin diseases. The company sells its products involved with AI must still have quite a the first step towards personalized treatment that is expected to in more than 100 markets. LEO Pharma has 5,000 employees Business-minded people will basic understanding of what AI is and dramatically improve the quality of life for chronically ill patients. across 61 countries and had a 2017 revenue of €1.4 billion. drive the transformation what it can actually do. Therefore, the The AI experts confirmed some of the key ingredients necessary for AI in organizations: a combination of domain and technical expertise, the ap- propriate technology, the right talent, and lots and lots of data. While letting tech-savvy individuals drive innovation Farmers and growers are still reasonably is great for building understanding, conventional, with an average age of 55 years. The AI will radically change the industry as the The biggest challenge with AI is data quantity true transformation will not come until chances are that this will change significantly in the R&D function will fundamentally change. A and quality. Pharma companies have plenty of business people start suggesting prob- future. It could just be that technology companies lot of research will be model-based, which data, but it is seldom structured and labeled in lems for AI to solve - not the other way round. will become the disruptors of our market. will significantly increase the speed of drug an adequate manner. discovery. — Royal Agrifirm Group Agricultural cooperative

22 23 Setting the Scene Setting the Scene

We believe that every organization is going to have to write their own AI manifesto: what they believe about AI, how they’re going From the Horse’s Mouth* to use or not use data, how they’re going to publish data, and *From the highest authority make the consumers of their products and services aware of that. The creation of those manifestos is going to become a gateway to the success of AI.

— Norm Judah, Chief Technology Officer of Worldwide Services at Microsoft

The full extent of the AI story remains in its early stages. What we do know is that big data, computing power and connectivity If you have a ton of data, and your problem is one of classifying pat- are changing the industrial landscape. The opportunity rests in terns (like speech recognition or object identification), AI may well accelerating the digitization of businesses, making them more be able to help. But let’s be realistic, too: AI is still nowhere near as data driven by building applications that deliver machine-assist- flexible and versatile as human beings; if you need a machine to ed insights. read, or react dynamically, on the fly, to some kind of ever changing problem, the technology you seek may not yet exist. Intelligence is a — Mona Vernon, CTO, Thomson Reuters Labs really hard problem.

— Gary Marcus, Founder & CEO, Geometric Intelligence [acquired by Uber] professor, NYU, contributor to The New Yorker and The New York Times In some cases, there is too much hype, but paradoxically, the potential opportunities and benefits of AI are still, if anything, under-hyped. Often, the impact of new technologies is overes- AI is a general purpose technology, so will eventually affect all in- timated in the short term and underestimated in the long term, dustries. However, this impact can be slowed by the lack of data and while there is a lot of noise regarding AI, there’s been a lack in particular industries. There’s also more innovative cultures of in-depth discussion and analysis of how it’s actually going to inside different organizations, that can either drive adoption or transform businesses. prevent it.

— Nigel Duffy, Global AI Innovation Leader, EY — Marc Warner, CEO, ASI Data Science

24 25 Artificial intelligence in Europe Role of AI in European Business

A Strategic Agenda Where is the AI conversation currently taking place?

A good starting point to understand Active C-suite and Board of Direc- both pertain to job insecurity and to how large European companies are tors involvement the fact that AI is still a highly abstract handling AI is to look at who in the In 71% of the companies surveyed, AI topic for many when it comes to prov- organization is driving the AI agenda, is already an important topic on the ing day-to-day business value. whether it be the Board, the C-suite, C-suite agenda and across various Role of AI managers, or employees. roles - from cost-focused CFOs looking AI is an important topic among for efficiency through automation, to executives in Denmark AI is particularly relevant at CDOs with customer-oriented ambi- In Denmark, AI is an important higher organizational levels tions as part of wider digitalization topic across most levels of the or- From driving strategic considerations efforts, to the CTOs who is often still ganization. This is particularly the at the Board level to being a topic of in- responsible for a type of AI Center of case at the C-suite level, where terest or concern at the employee level, Excellence. 76% of Danish companies sur- in European the results are clear: AI is important veyed report AI is an important and resides across all levels at many of Companies more advanced in AI tend item on their agenda. Similarly, the organizations we interviewed. to have stronger involvement of the 68% of companies in Denmark C-suite and the Boards of Directors consider AI to be an important Only a few companies stated that AI is than the rest. They focus less on the topic at the managerial level. not currently an important topic at any technology itself and more on the busi- level of the organization - while the ness problems that AI can addresses. Business vast majority of companies view AI as generally important regardless of how Relatively speaking, the AI topic seems advanced they are, or how much AI is to not yet having reached the same being considered for deployment in level of importance at the non-mana- the near future. gerial level (employees) than at the top. Speculating about the reason, it could There is a lot of hype surrounding AI at the moment, and few doubt its potential. We examine how important is AI compared to AI is an important topic on the C-suite level in particular other digital priorities and where AI fits on the strategic agenda. On what hierarchical levels in your company is AI an important topic? AI is in particular an important topic at the Execu- tive Management level We look at the impact of AI on the company’s core business, as STRATEGIC LEVEL Board 44% well as adjacent and new areas of business. of Directors 38% level

Executive 76% Management 71% We also examine the current AI maturity levels across sectors and level markets, the potential drivers for deploying AI, and where AI is 68% Managerial 56% applied within organizations, across customer-facing functions, level

Employee 32% operations, product development, and internal business support. (non managerial 28% level)

OPERATIONAL LEVEL

Affirmative responses, 15 European markets Affirmative responses, Denmark

26 27 Role of AI in European Business Role of AI in European Business

Among Friends Push or Pull What is the importance of AI against other digital priorities? How is AI predominately deployed into the organizations?

To understand the drivers behind Business and IT drive AI advancement in Denmark In a business era driven by innovation The participating companies are gen- AI seen as slightly more impor- the adoption and deployment of AI Among Danish companies surveyed, 60% drive AI deployment by and tech-led disruption, AI is obviously erally in the process of understanding tant vs. other digital priorities in the companies, we took a closer both pull from the business needs as well as push from IT’s ca- not the sole priority. the potential of existing data, includ- in Denmark look at how AI is approached in a top pabilities and innovations. In addition, 36% of Danish companies ing to what extent it can be used, Many companies surveyed in down-bottom up management con- manage AI via a combination of both top-down and bottom-up AI as a digital priority what it can be used for, and how to Denmark are engaging in suc- text, and from a functional tech- vs. approaches, above the European share (28%). This outcome re- capture and leverage it. When asked on a scale of 1 to 5 how cessful pilot projects and Proofs business driven dynamic. flects the results from the survey, showing that AI is an important important AI is to the business relative of Concept or have AI initiatives topic across different organizational levels for Danish companies. Furthermore, many of the companies to other digital priorities, the majority that are released into production. AI driven from a combination of are focused on building the appropri- of respondents told us that it is about When it comes to their prioriti- technology push and business pull ate data infrastructures or modern- equal. Very few organizations said it zation, respondents in Denmark The contributing companies are quite izing legacy systems as a top digital was their most important digital priori- on average consider AI slightly evenly split across deploying AI as a priority, both being prerequisites ty, or not formalized as a digital priority more important than other digital top down process, as a bottom up, or for introducing AI into the company. AI deployed and managed in a balanced way at all, with the spread of responses priorities, a ranking marginally as a combination of the two. However, Considering that AI is heavily reliant How would you characterize the way AI is being managed in your com- leaning slightly towards the upper end above the European aggregate. when looking at the self-reported most on data as its fuel, this development pany? How would you characterize the way AI is being deployed in your of the importance spectrum. Additionally, among the countries advanced companies, they are more suggests that the foundations are company? surveyed, Denmark is the only top down than bottom up in their ap- being laid for further AI integration in This slant is likely to increase as many country where no company re- proach. It was clear from speaking with the years to come. companies expect AI to become more ported that AI is not an important them, that this is partly a result of AI important, as the technology develops digital priority. Respondents are being increasingly important enabler and use-cases become more clear to in the company, and playing an in- also focusing on fixing legacy Top Down Bottom up Both companies. systems, collecting and storing creasingly significant role in the overall the right data, and building their strategy. general digital strategy. AI driven from a combination of technology push and business pull According to a large part of the companies. and despite still being a techni- cally complex thing that requires many 34% 28% 29% 32% 28% 36% specially skilled employees, AI is most AI is seen as one of many digital priorities - but rarely the most important The majority consider AI to be important often deployed as a combination of How important is AI relative to your company’s other digital priorities? business pull and technology push.

This resonates well with one of the Business Pull IT Push Both Avg. Score most consistent inputs from the execu- 52% 44% tives on the most sought after AI pro- Deployment Approach 28% 24% files which centered in on the hybrid 16% 12% 3.2 3.1 9% 8% 7% profile that understand the business 0% needs and the ability to match them to the technological possibilities. 1 2 3 4 5 24% 32% 23% 8% 45% 60%

Not important Important Most important AI is not formalised AI is one of many AI is the most important as a digital priority digital priorities digital priority

15 European markets Denmark

15 European markets Denmark Note: Remaining percent ‘Don’t know’ responses Note: Remaining percent ‘Don’t know’ responses

28 29 Role of AI in European Business Role of AI in European Business

TMT sector with largest percentage of companies that are either released or advanced Ready, Set... How would you describe your company’s general AI maturity? Sectors arranged by maturity based on Advanced and Released What is the maturity of AI in different sectors?

TMT 2% 10% 40% 45% 2%

While working with AI should be con- the structure of existing data, collec- Services 5%6% 22% 27% 27% 18% sidered a continuous journey, the AI tion of new data, and data access in maturity of surveyed companies pro- general. However, the trend is clear: vides a tangible indication of the level AI maturity is on the rise as adoption of Finance 4% 22% 34% 36% 4% of advancement of current initiatives. key technologies accelerates and internal capabilities grow. Multiple use cases, limited scalabil- Infrastructure 5%9% 21%17% 32% 46% 28% ity and advanced use The vast majority of European busi- The majority of companies have begun nesses are currently either conducting exploring use-cases, while some com- pilot projects to test selected use- Industrial Products 4% 21%25% 53%44% 21% panies have made early investments cases, or have commenced implement- with the intention of taking a leading ing AI in the business. When talking position in AI. The levels of advance- with executives, it is evident that many Life Science 4%7% 25% 45% 49% 34% 17% 4% ment also vary in that some companies companies are struggling with how to are focusing on narrow use-cases to integrate pilot projects into daily op- support their existing business, while erations. CPR 25% 29% 29% 9% 9% others are taking an explorative approach. Among the small group of Clear sector patterns, with TMT, Services, and Finance on top companies with no or only little AI 0% 20% 40% 60% 80% 100% activity to date, several respond that Companies currently leading the way in they are planning to drastically ramp terms of AI maturity are in TMT, Servic- None Planned Piloting Released Advanced up efforts soon. es & Hospitality, and Financial Services. Companies in those sectors gravitate Technology immaturity and internal towards grading their AI maturity as data quality are key obstacles ‘Released’ (AI in active use, though Many companies that have already selectively or not with very advanced This indicates slower technology ‘Advanced’ stage of AI maturity. As open source implemented AI initiatives in their tasks), or ‘Advanced’ (AI actively con- adoption lead times in these slight- Several companies in both Consumer technologies have matured businesses are seeing tangible benefits. tributing to many processes and en- ly more conservative sectors. Yet, Products & Retail and Services & Hos- abling advanced tasks). A logical ex- we are working on Consequently, many of them are ex- with 74% of companies being in the pitality cite the challenges of knowing ploring more use-cases and structuring planation for the maturity in TMT and ‘Piloting’ or ‘Released’ phases, the what relevant AI technologies are avail- establishing synergies with their learnings from previous AI pro- Finance is their tendency to be digitally Infrastructure sector also seems to able, utilizing unstructured data, as well our enterprise software jects into a modus operandi that can advanced and more savvy with analyt- be evolving onto more advanced AI as affording the payback period where – something which we speed up new initiatives. ics, favoring these companies to pro- maturity. there may be large upfront costs and gress beyond piloting by having data believe is a prerequisite undetermined returns on investment. Meanwhile, a substantial number of science capabilities in place to evolve Life science and CPR have fewest for developing the best AI- companies have intentionally chosen to towards more advanced AI stages. released projects driven digital services as take a ‘follower’ position, reporting the Consumer Products & Retail compa- well as attracting the best perceived immaturity of AI technolo- Infrastructure and IP with relatively nies have a broad spread in terms of AI many projects in ‘piloting’ phase people gies as a key reason. Another reported maturity, where 25% state they have obstacle to rolling out broader AI The Infrastructure and Industrial Prod- no plans at present for how and when initiatives are rooted in data and data ucts sectors both stand out as having to use AI – much higher than other — TDC infrastructure, where companies have no companies responding that they are sectors – while others in the same Telecommunications separate projects aimed at improving ‘Advanced’ in AI at this stage. sector are already at the ‘Released’ or company

30 31 Role of Ai in European Business Role of Ai in European Business

AI Maturity Curve Companies in Denmark are among the most AI mature In terms of AI maturity, Denmark has a strong position compared to the European aggregate. Many Majority of companies are in the ‘Piloting’ or ‘Released’ stage Danish companies are using AI to some degree – among them, 8% report to be in an advanced stage, where AI is actively contributing to several processes across the organization. However, the majority We asked companies to self-report their current AI maturity level, grading themselves at None, are still in the piloting phase (52%). Some of their use-cases include the employment of machine Planned, Piloting, Released, or Advanced - as defined below. learning for product performance optimization, tailored marketing initiatives, and sales support to identify cross-sales opportunities or predict churn.

2 / 25 LEVEL OF MATURITY (8%)

Advanced AI is actively contributing to many processes in the company and is enabling quite advanced tasks 9 / 25 (36%)

Released AI is put to active use in one or a few processes in the company, but still quite selectively, and/or 13 / 25 not enabling very advanced tasks (52%)

Piloting AI is put to active use, but still only in early stage pilots

1 / 25 (4%) Planned AI is being planned, but not yet put to active use, not even in early stage pilots

0 / 25 None (0%) Not yet thinking about AI

20 / 269 59 / 269 106 / 269 74/ 269 10 / 269 (7%) (22%) (39%) (28%) (4%)

15 European markets Denmark

32 33 Role of AI in European Business Role of AI in European Business

State your Business Where is AI currently deployed across the companies’ value chains?

Looking at the business functions that towards taking an experimental, agile AI technologies in the future. which in the case of HR include talent most commonly use AI provides a approach which is key to AI; and the Operations and back-end functions acquisition (avoiding human bias), good indication of where companies R&D function often sits on significant use AI to increase efficiency by au- onboarding (Q&A), performance eval- We are applying AI in some are placing their bets. These functions amounts of useful data leading to high tomating processes and informing uation (analyzing data), etc. but rather selected areas related to are driving the company AI agenda, potential use-cases. decision-making. The key enabler is seems to be a result of prioritizing other influencing the future direction of the data infrastructure, and many com- functions and priorities first. consumer experience, but company’s AI efforts. Online customer interactions panies – currently limited by legacy wide-ranging ambitions generating front-end data systems and processes that impede are emerging across our Many AI in R&D and IT/Digital capture and retrieval of data – need to AI mostly applied in R&D & Product Development, IT, Tech, & Digital and Customer-facing, commercial functions value-chain. Currently, we functions such as Marketing, Sales and Customer upgrade their infrastructure. Marketing in Denmark are defining the future data On top of an expected high prevalence Service are also heavier users of AI, Among companies surveyed in Denmark, usage spans 11 out of the 13 business of AI within IT departments, AI is also partly driven by their digitization levels. Limited use in HR and Procurement functions presented. The distribution of AI usage is primarily concentrated in architecture as a foundation commonly used within R&D functions. Although AI is generally adopted more There are several functions where AI is R&D & Product Development (56%) and in IT, Technology & Digital (52%). In for extensive use of AI. This primarily comes down to three slowly in customer facing interactions hardly in use among the participating addition, Denmark reports a higher usage of AI in Marketing (40%), above the factors: employees in R&D are often than in back-end functions, the abun- companies. This includes people- European aggregate (22%). These results partly reflect the large Danish phar- — LEGO engineers who tend to have a good dance of data from increased use of ‘intensive’ functions such as HR and maceutical industry using AI to transform R&D, as well as the general interest of Toy company understanding and appreciation of AI; online channels is expected to make Procurement. This is not due to lack companies to understand and approach their customer base with the assistance the R&D function is often already wired these functions obvious candidates for of potentially valuable AI use-cases, of AI.

AI most commonly applied in IT & R&D functions Which of your company’s business functions currently use AI?

6% 4% 7% 14% 36% 9% 4% 12% 23% 47% 22% 19% 24%

56% 52% 40% 36% 28% 24% 16% 12% 8% 8% 0% 0% 4% General Management General HR Admin / Finance Management Product Procurement Manufacturing Operations / Logistics IT / Technology / Digital Marketing Sales Service Customer Strategy R&D & ProductR&D Development

Group Product Operations Commercial

Affirmative responses, 15 European markets Affirmative responses, Denmark

34 35 Artificial intelligence in Europe ( Case Study ) Artificial Intelligence in Europe

A.P. Moller – Maersk The European

There is no doubt that AI has the po- and services and improve existing Treating AI as a distinct part of wider tential to transform Transportation & products and services); enhanced digital initiatives, Maersk established Logistics, giving rise to a new class of customer experience (service delivery, an in-house software development and intelligent logistics assets and opera- issue resolution, empowering custom- innovation unit, consisting of 100 em- tional models. Data science ployees and growing. The aim Business is not new to A.P. Moller – is to deliver AI products and AI Landscape Maersk, yet only recently has solutions rooted in the group AI become a part of Maersk’s As a designated new discipline posi- business strategy, building on core strategy as a functional well-defined use-cases with technology with tangible tioned close to the core of group strat- deep sponsorship from the applications. As a designated egy, Maersk is developing AI capabili- business, thereby avoiding the new discipline positioned ties as part of a broader transformation trap of living separately from Benefits and close to the core of group of the business. the business and not adding strategy, Maersk is develop- value. ing AI capabilities as part of a Maersk’s early investment in broader transformation of the agile transformation and peo- business. ple capabilities has resulted in Maersk takes a broad view of er-facing employees); and operational the organizational structure AI, applying intelligent technology to efficiencies (for example via network and concentration of talent necessary Risks three main areas: product offerings, optimization). to drive AI forwards in a large global (using AI to develop new products organization. As a number of industries are beginning to reap the benefits of AI, we investigate what AI is actually doing for businesses today What next? and what is expected in the future.

A.P. Moller – Maersk is a Danish conglomerate with activities Maersk is developing a platform to leverage company data to in two sectors: Transport & Logistics and Energy. Maersk is develop products, partly by optimizing the company’s data the largest company in Denmark, and the world’s largest architecture to ensure faster development. To meet these We takeWe looka look at at how how big important an impact the executives digital transformation expect AI will have in operator of container ships and supply vessels. The company demands, Maersk is changing its approach to attracting talent. terms of driving growth or causing disruption in their industry, has approximately 88,000 employees, a fleet of more than 1,100 AI also requires an entire new skillset among company leaders, agenda is on the highest executive level vis-à-vis other vessels, and subsidiaries and offices in 130 countries. Its 2017 transforming them into AI leaders who are deeply engaged in and examine AI’s basic and more advanced uses - highlighting revenue was $31 billion. implementing AI in the business. strategic priorities. examples of these functionalities in operational mode. We dig deeper to understand whether digitalization is primarily a keyWe lever also to present improving a strategic and sustaining approach the to current understanding core busi AI’s- four ness,benefit or rather domains a lever forfrom building a business tomorrow’s perspective, business summarizing focusing the on adjacentvalue executives or even entirely expect new to generate business by areas. using AI, and touching on At Maersk, we build things that have deep There’s good awareness about what AI can what business leaders see as the most prevalent business risks. business sponsorship and add value. bring to the shipping industry. In the future, Maersk won’t just be a shipping company, but And we summarize how progressed the companies are on an integrated forwarder of logistics. their overall digital transformation maturity journey.

36 37 Business Benefits and Risks Business Benefits and Risks

Another World What is the expected impact from AI within the next 5 years? We are not too worried about FinTechs. They are developing Of the surveyed companies, 81% believe tors, more than 30% expect the indus- Countries expect different impact Companies in Denmark are in the middle of the pack nice products but are serving that AI will have a high or significant try to be disrupted. from AI At 40%, companies in Denmark are in the middle section along with a niche market. By partnering impact on their industry within the next When approaching impact from a companies in Sweden, Spain, and Ireland in terms of expecting AI to with FinTechs we can both five years. Digging deeper into the data, Limited sync of maturity and country perspective, the tendency have a significant impact in the future. However, when including com- many of these companies expect AI to expected impact make better propositions. We remains; very high expectations across panies that reported a 4 on a scale of 1 to 5, 88% of Danish companies fundamentally change their competitive are more concerned with the The biggest disparity is within Finance, the board. Portugal stands out with report that AI will have a high or significant impact on their industry. landscape, driven by increasing risk of specifically Pension and Insurance, most ‘high’ impact responses. This result can be partly explained by the sector configuration in tech giants. They have a lot competition, including from new types where ambitious companies are mak- Denmark, which is concentrated in Life Sciences, Finance and TMT. All of data and a lot of money, of start-ups and companies from adja- ing significant investments in building In the opposite end of the expected these sectors score on average relatively high across Europe. Accord- cent industries. The majority of compa- so if they decide to enter the data infrastructure and AI capabili- impact scale, Ireland, Austria, and ing to the executives, some of the ways in which AI will disrupt indus- nies also believe that AI will play a key insurance market that can be ties, while others are taking a waiting Spain, in that order, are the countries tries relate to automation of core business processes, new platform role in their efforts to continuously cut stance, and will jump on the AI train where most companies expect only and service business models, or new digital offerings. a problem. costs to stay competitive. when the technology is more mature. ‘some’ impact from AI or less.

Strongholds and premiums to — Aegon change as AI gains ground Financial services group Many companies expect competition Services the sector with the highest expected impact from AI High expected impact from AI consistently across countries to intensify due to the ‘winner takes all’ How much impact do you expect AI will have on your industry How much impact do you expect AI will have on your industry dynamic often associated with the mas- within the next 5 years? within the next 5 years? Portugal has the highest share of companies sive scale that AI and digital can create. expecting ‘significant They also expect significant impact on impact’ from AI their products, increasingly in the form of new services, and they believe the UK, France & Germany Belgium & Luxemburg Portugal Sweden Austria Italy Norway Netherlands Denmark Spain Ireland Switzerland Finland Infrastructure Life Science Services Finance Industrial Industrial Products TMT speed of developing new products and CPR taking them to market will drastically 100% 5 100% 5 decrease - making current competitive Significant impact 9% Significant impact strongholds less viable in the long-term. AI will disrupt the industry, re- AI will disrupt the industry, resulting in entirely new products, sulting in entirely new products, 25% 29% 33% services, and business models 33% services, and business models This is particularly clear in R&D intensive 33% 35% sectors such as Pharma, where big data- 80% 42% 40% 40% 4 80% 40%43% 40% 40% 40% 4 43% 45% sets and intelligent algorithms to speed High impact 48% High impact 55% up the drug discovery process (10x 50% 64% mentioned as realistic) can impact the dynamics towards existing peers, while new AI based entrants (e.g., intelligent 60% 3 60% 3 devices) can influence how premiums Some impact Some impact are distributed in future value chains. AI will create significant industry AI will create significant change to the industry, but key 33% 77% 25% change to the industry, but key 44% structures will remain as is 42% 57% 65% structures will remain as is Across sectors, executives expects 46% 30% 35% significant impact 40% 2 40% 2 45% 33% Limited impact 48% Limited impact Services comes out on top in the ‘High 47% 41% 50% Impact’ category, but all sectors expect 39% 54% a significant degree of impact from AI. 18% 35% An overwhelming share also anticipate that AI will result in entirely new prod- 20% 1 20% 29% 1 15% 35% ucts, services, and business models. No impact No impact 24% AI will not recognizably change 14% 22% AI will not recognizably change 21% 10% 17% products, services and business products, services and business Companies from Industrial Products 9% 18% models in the industry 5% 14% 5% 5% 14% 5% 14% models in the industry 11% 12% and CPR expect relatively least ‘high’ 3% 4% 5% 5% 5% 5% 5% 5% 5% 4% 3% impact from AI, but even in these sec- 0% 3% Don’t know 0% Don’t know

15 European markets Denmark

38 39 Business Benefits and Risks ( Case Study ) Artificial Intelligence in Europe

AI Here, There, Everywhere What is the proximity of AI’s future impact to core business? PFA Pension

Companies expect impact across all horizons PFA Pension has been working with AI bot that matches the inquiry to one are sent to the client. Due to the high To what degree do you expect AI will create impact for your company within each of the following areas? since 2014, focusing on a wide range of 300+ response templates. This in- complexity of pension products, PFA’s of applications spanning process au- creases consistency while dramatically vision is that AI will eliminate routine tomation, intelligent advisor tasks and empower its em- platforms and prediction for ployees with insights that free Avg. Score meaningful client engage- up time for interaction with Core Business ment. Its models predict im- PFA’s models predict important life clients. 36% 40% 29% 28% 37% Primary areas of 24% portant life events and score events and scores customer loyalty, Like many other businesses, 4.2 4.0 the company’s customer loyalty, enabling enabling pension advisers to engage the main challenge for PFA is current business 1% 4% pension advisers to engage how to integrate successful 0% 0% in informed conversations in informed conversations with clients pilot projects into daily oper- with clients based on fast and based on fast and effective lead prior- ations. To make this process Adjacent Business effective lead prioritization. easier and to create the basis 52% itization 32% 31% 33% This relevant dialogue with for realizing the cross-organ- Business areas on 22% clients has already shown izational potential of AI, PFA the edge of the 12% 3.7 3.7 7% company’s core 4% results, with lower churn and is building the proper data business 1% increased salesforce effectiveness. reducing time spent per inquiry, since infrastructure. To deliver on this, a third 0% For in-bound inquiries, PFA has cre- PFA employees only need to confirm of the IT organization is working on a ated a natural language processing and tweak the templates before they range of agile projects. New Business 40% Business areas 24% 32% 20% 22% 24% 3.7 3.8 entirely new to 10% 8% the company 3%

0% 1 2 3 4 5 What next?

Not at all To some degree To a very high degree PFA Pension is the largest provider of pension and insurance PFA Pension will use AI to further customize client interaction and services in Denmark, serving more than 1 million customers offer new services catering to evolving client needs. A “Next Best 15 European markets Denmark Note: Remaining percent ‘Don’t know’ responses and managing €80 billion in assets. In addition, the company Action” algorithm will make it possible to deliver personalized has begun offering asset management, investment funds, recommendations to clients based on their and their peers’ and banking services. The company was founded in 1917 as an behavior, scenario probabilities and data patterns. Pension Many of the participating companies Europe are more split and contain more Danish executives split on new independent pension company, which remains customer-owned advisers will execute on this by leveraging insights from AI tools. are expansive, with diversified business “Don’t Know” responses than for “Core” business impact today. It employs 1,300 people and has one of the lowest cost- PFA also plans to leverage AI technology to digitally reach clients units offering a range of products and – perhaps because there is an inherent Around 50% of companies expect levels per customer in the industry. 24/7 with intelligent self-service solutions. services. We questioned where they challenge in making predictions about AI to have a high or very high expect AI to have an impact - in their AI’s impact on new business areas impact across core, adjacent and core, adjacent and/or new business. where business results are not yet real- new business areas. Some com- ized, and where the role of current and panies have a designated unit to AI will impact across the board, but upcoming AI technology is not clear. identify new business areas; oth- less consensus on timelines ers are skeptical about whether Yet, interestingly 32% feel confident AI Companies expect AI to have a relative- their organization will create new AI will fundamentally change our industry. Increasingly, we find it necessary to articulate ly equal impact on core, adjacent and will impact areas that are “entirely new business streams with the help Historically, pension has low customer our vision of AI to ensure organizational new areas of their business. In inter- to the company.” This is not far behind of AI. views, they say impact depends on the the 37% of respondents who expect a involvement, but I believe intelligent digitally buy-in. We do not plan to cut FTEs – we plan timeline, for instance AI impacting the very high degree of impact on the core enabled self-service solutions can change this. to leverage this technology to empower our areas of the current business. core business now, but adjacent and This could ultimately lead to better coverage employees. new business later on. The range of an- for our customers. swers for “Adjacent” and “New” across

40 41 Business Benefits and Risks Business Benefits and Risks

Use It or Lose It Predict We use image recognition to predict how enzyme batches will develop at very early stages in the process. This makes us more Anticipate events How is AI put to use in companies today? effective, as we can take preventative measures. and outcomes ― Chr. Hansen Bioscience company

AI enables a wide range of uses, broadly tasks can be done without human inter- Prescriptions’ potential is big split into personalizing, automating, vention, a substantial number of com- Prescription is the laggard among the predicting, prescribing and generating panies are currently in the process of five AI uses, with current use-cases insights. We asked companies how training chatbots to transform the way typically being early stage, such as relevant each was to their business and information is acquired. suggestion engines and decision rec- found a significant degree of variance in Automate ommendations for salespeople and Any manual process should be looked into to see if it is really terms of what executives expect to use Generating insights to make in- advisors. AI for advanced prescription neccesary Everything that is a repetitive, slightly boring task AI technologies for. formed decisions Handle tasks without such as complex decision making lies in human intervention should be taken away if possible. This frees time that can be Focusing on generating insights based the future, as it requires collecting large Prediction is the top use spent engaging with customers. on internal and external data, 58% of amounts of data and understanding With 74% of companies seeing predic- companies view AI as a way to make which variables are significant, including tion as a relevant use of AI, this func- better decisions. This requires a sophis- some that are difficult to digitize. ― SEB Financial services group tionality, which includes churn analysis, ticated data infrastructure. Companies predictive analysis, and predictive reliant on R&D are using AI to speed up maintenance, comes out as the top the process of analyzing data for new Prediction and automation use. Companies with a large customer product development and to inform most relevant in Denmark base use churn analysis to identify and future research. At least 50% of respondents in proactively engage customers with exit Denmark consider four of the five Insights We use AI to find trends across R&D datasets that would potential. Sales teams use predictive Personalization is becoming a com- main uses of AI relevant for their normally be very time consuming or even impossible to find. analysis to identify leads with the high- mon feature Identify and understand company. The most common The potential for AI within R&D is huge, as it will speed up the est likelihood of conversion. Companies Among the surveyed companies, 44% are to predict and to automate, patterns and trends pace of drug discovery. that sell or use advanced costly machin- are using AI to personalize the user followed closely by generating ery use predictive maintenance to save experience, for instance by tailoring insights. Current use-cases high- money through decreased downtime. content to individual interactions as an lighted by executives include ― H. Lundbeck Pharmaceutical company effective way of driving mass-person- creating prediction models that Intelligent automation for effec- alization. Next steps in personalization allow companies to estimate de- tively dealing with routine tasks include chatbots and virtual assistants, mand, and automation of routine Smart automation is seen as widely ap- where some companies already have back-office tasks and operations. plicable by 74% of companies surveyed. fully automated customer front-end With estimates that 20-30% of current solutions in place. Personalize We use AI to personalize the client experience on our trading platform based on the individual user’s behaviour and Tailor content and preferences. This allows us to offer a more tailored trading user-experience Prediction and automation relevant to most companies experience for all of our customers regardless of size or What are the relevant uses of AI in your company? activity level. In summary, it is all about democratization of services that often is only available to the top end clients.

74% 72% 58% 44% 24% ― Saxo Bank Investment Bank 92% 88% 80% 64% 32%

Prescribe We use Natural Language Processing to group customer inquiries and suggest which of our 300+ templates we should Suggest solutions to use in response. Our employees only need to confirm the defined problems choice or tweak it slightly. This dramatically lowers the time it To predict To automate To generate insights To personalize To prescribe takes to respond.

Affirmative responses, 15 European markets Affirmative responses, Denmark ― PFA Pensions and insurance company

42 43 Business Benefits and Risks Business Benefits and Risks

Making AI Simple Improved production and efficiency ented sectors, AI enables provision of through optimized operations new services via multilingual cognitive What is a good framework to map the potential benefits from AI? While digital transformation in general tools, geo-location suites, sentiment is based on customer engagement, op- analysis, cognitive robotic advisory ca- timizing operations is what companies pabilities, personalized service agents first look to when putting AI to use. It and more to transcend the sectors to draws on multiple levers such as: intelli- a new level of value-add -with signifi- The contributing companies generally text, voice, and images; and ‘interact- gent prediction, e.g., identifying chron- cantly increased scale and reach in real expect to benefit in all four key doing’ with employees, customers and ic diseases, anticipating non-perform- time. mains as outlined in Microsoft’s Digi- other stakeholders in natural ways. ing products, or adaptive modelling tal Transformation framework: opti- to flag corrective actions; operational Enabling employees to be more mizing operations; engaging customers; Applying AI to these domains can be efficiency, e.g., optimizing forecasting efficient and capable transforming products and services; transformational to a business, ulti- and order-to-fulfilment flows across Across sectors, numerous AI use-cases and enabling employees. Each domain mately changing the landscape of the the value chain, or processing huge focus on increasing employee produc- draws on underlying AI functionali- business itself and the industries and sets of documents in a fraction of the tivity or serve to enhance the human ties – ‘reasoning’ through learning and eco-systems in which it operates. time; and deep insights, e.g., detecting ingenuity and the ability to fulfil a forming conclusions with imperfect anomalies to surface irregularities such given function. AI helps employees in data; ‘understanding’ through inter- Let’s look in more detail at what that as fraud, or identifying new pockets of B2C companies expand organizational preting the meaning of data including entails. opportunity before competitors do. knowledge by analyzing vast customer behavior datasets in order to adapt Engaging customers more effec- online and offline store layouts, driving tively through AI conversion and sales. Customer per- After optimized operations, companies sonalization is used at scale, powered look to customer engagement as the by AI solutions that reveal real-time Artificial Intelligence impacts business in four benefit domains domain in which to seek most business customer insights, identifying the best Companies must consider how they approach the benefit domains in their AI strategy formulation benefits. Early examples of AI appli- next actions for up-sell and cross-sell cations in the customer engagement opportunities, as well as predictive space involve levers such as conversa- models that obtain a 360-degree view tional agents, e.g., bots providing per- of the customer by integrating cus- sonal recommendations and transac- tomer data and sentiment to generate tional advice; personal assistants, e.g., targeted offers. guiding decision-making, shortening conversion cycles; and self-service, e.g., options to help customers reduce time Engage your customers Transform your to resolution. E.g., provide customers products & services advice, shorten conver- E.g., speed up prod- Staying ahead of the competition by sion cycles, and reduce uct innovation cycles, transforming products and services time to resolution Artificial enable new value add services, and provide Transforming products and services, Intelligence real time support and enabling employees, came out on the same level, slightly below the two benefit other domains when it comes to where domains companies expect to generate future business benefits. It is important to ensure proof of concept and a clear indication of value creation Transforming products and services, before moving forward. AI has to drive ultimately giving rise to entirely new business models, is mostly favored in measurable impact and not just hype. Enable employees Optimize your R&D-heavy sectors where companies E.g., increase employee operations consider AI and advanced analytics as efficiency through pre- E.g., improve plan- — Ørsted levers to speed up the product innova- dictions, enabled sup- ning and reduce costs Energy company tion and discovery process. In B2C-ori- port, and automation of through intelligent repetitive tasks prediction, operational efficiency, and deep insights, predictive maintenance

44 45 Business Benefits and Risks ( Case Study ) Artificial Intelligence in Europe

Where Value Hides What benefits do business leaders particularly expect from AI? Carlsberg

Carlsberg has a strong history of re- what they’re calling a ‘flavor finger- products at a much higher speed and search and innovation at its Research print’ for each sample. Insights from quality. Respondents were asked to assess the Fewer expect products and services Higher expectations to engage Laboratory established in 1876 – from this dataset can then inform the devel- potential of AI within each of the four and employee engagement customers than other countries isolating yeast cultures for beer fer- opment of new brewing organisms and One of the collaborating partners, benefit domains. Although executives speak of the po- Among Danish companies sur- mentation and inventing the pH scale, ultimately entirely new beers. Working iNano at Aarhus University, has already tential in making sense of existing and veyed, 96% expect AI to optimize to cracking the genome sequence in partnership with universities and conducted a Proof of Concept that can Optimizing operations and engag- new sources of data to introduce high- their operations through the of barley and developing differentiate between four ing customers to deliver most value er margin services to product portfo- automation of processes, or by sustainable packaging Carlsberg beers: Carlsberg Among all companies surveyed, 89% lios, expedite new product develop- forecasting capacity and 96% solutions. The Laboratory Pilsner, Tuborg Pilsner, expect AI to prove beneficial in opti- ment, and introduce innovative new believe AI will benefit customer creates more than 1,000 Wiibroe and Nordic. De- mizing operations, with use-cases most offerings, only 65% expect AI to help engagement – across all countries beer samples daily and with The study ‘The Beer Fingerprinting Pro- veloping methods for fast highlighted by executives being mon- transform products and services. surveyed, Denmark has the high- its new research study ‘The ject’, is harnessing AI to develop how new and reliable assessment of itoring results, predicting trends, and est share of companies expecting Beer Fingerprinting Pro- flavors in complex mixtures prescribing future solutions. A lot of Even fewer (60%) expect AI to provide beers are created and enjoyed. benefits in these two domains. At ject’, it is harnessing AI to are hugely valuable from a focus is given to intelligent automation, benefit from empowering employees 84%, Danish companies expect develop how new beers are product development per- such as making compliance cheaper to improve productivity, enable inno- AI to boost their transformation created and enjoyed. spective as well as for qual- and more robust, improving risk analy- vation, support problem solving, etc. of products and services, for ity control and assessment sis, optimizing supply chains, providing instance by making offerings The project uses a series of high-tech technology companies, AI solutions will purposes. Structuring and speeding up predictive maintenance capabilities, What we did hear overwhelmingly, more data-driven and interactive. sensors which can accurately gauge help the research teams to select and this process using AI can significantly and more. however, was the importance of bring- Lastly, 72% of Danish respondents the delicate nuances and aromas in the develop brewer’s yeast for application enhance Carlsberg’s product offerings ing all employees along on the com- expect AI to enable employees beer, mapping out data constituting in both alcoholic and non-alcoholic and the time it takes to go to market. Not surprisingly, the ability to struc- pany’s AI journey. This involves getting by freeing up their time and thus ture repeatable processes and reduce internal buy-in that AI will be a force allowing them to focus on more human error and bottlenecks is some- for good, generating excitement about interesting and value adding thing most executives can get behind working with intelligent technologies, tasks, or by providing personal- from a cost-saving perspective. and making existing jobs easier and ized and tailored training. 74% of companies surveyed expect AI more engaging. What next? to help them engage customers and With plans to use the project to strengthen its position in the enhance the user experience, including Established in 1847, the Carlsberg Group is one of the leading global beverage market, Carlsberg is also expecting the three-year tailoring content, increasing response brewery companies in the world today, with a large portfolio of project to spark innovations and lead to new start-ups beyond speed, adding sentiment, creating beer and other beverage brands. Their flagship brand, Carlsberg, the brewing world that can use technology that supports the experiences, and anticipating needs. is one of the best-known beer brands in the world and, reflecting evolving consumer desires for various tastes and styles, the mapping of complex mixtures. Carlsberg, having once developed Group is also growing their focus on specialty and craft brands. one of the original methods for brewing beers that is still in The Carlsberg Group employs over 41,000 people globally and use by most companies today, is looking to build on this latest technology-driven innovation for use in other industries, such as Most companies expect to generate benefit from optimizing operations operates in more than 150 markets. Headquartered in Denmark, its the environmental, pharmaceutical and food industries. What business benefit do you expect AI to generate? annual revenue in 2017 was 62 billion DKK.

89% 74% 65% 60%

96% 96% 84% 72% This research study puts advanced analytics No rapid assays exist today for the and intelligent cloud technology as a determination of flavor compounds in cornerstone of the project and combines beverages but it is crucial that we can do this expertise within several fields of research. to ensure that the Laboratory continues to Optimizing operations Engaging customers Transforming products & services Empowering employees We are excited to see the project unfold and develop beer of the highest possible quality determine how it will impact faster go to and provide a model for brewing in Denmark Affirmative responses, 15 European markets Affirmative responses, Denmark market processes for Carlsberg. and the rest of the world.

46 47 Business Benefits and Risks Business Benefits and Risks

Sector Benefits Landscape Front to Back We asked companies across sectors what business benefit they expect AI to generate What are the expected benefits by sector? across Engaging customers, optimising operations, Empowering employees, and Transforming products & services Executives surveyed and interviewed wearables, paving the way to valuable TMT expects AI to increase engage- in the various sectors recognize the data collection and even entirely new ment, insights, and connectivity distinct benefits of AI, speaking about business models. The focus in many Telecom, Media the myriad of ways they see AI trans- and Technology companies seem to forming their businesses and indus- Engaging customers in new ways in be on using AI to reduce costs of re- Engaging customers Optimising Transforming Empowering tries. Although there are clear patterns Consumer Products and Retail taining and growing customer bases. operations products & services employees to discern, executives from different The Consumer Products and Retail AI is projected to help build seamless sectors often speak to different ben- companies we spoke to rank lowest in experiences across devices, predicting Tailoring content, increasing Automating processes, Adding data services, gener- Improving productivity, en- response speed, adding sen- monitoring results, pre- ating new business models, abling innovation, exploring efit areas from which they particularly terms of expecting benefits from AI, churn, and automating customer ser- timent, creating experiences, dicting trends, prescribing extending reach, etc. new capabilities, supporting hope to capitalize from. pulled down by only 44% expecting vice capabilities to solve some of the anticipating needs, etc. solutions, etc. problem solving, etc. benefits from AI to empower employ- sector’s longstanding challenges while Services companies expect the most ees. However, with multilingual cogni- bringing down costs. Life Science benefits from AI tive tools and being able to bring tar- Pharmaceutical, Services companies reported the high- geted, tailored offerings to customers, AI to revolutionize Financial Servic- 71% 71% 88% Healthcare, 54% Biotech est expected benefits across all four many spoke of the potential to engage es firms domains, expecting significant value customers, and of using AI for crucial Finance companies reported some of from AI through engaging customers activities such as understanding brand the highest expectations for AI benefits CPR and empowering employees, for ex- performance and sentiment analysis. across the four domains, which can Consumer ample via improving resource and skills explain the sector’s current frontrunner Products 78% 78% 58% 44% allocation across their large human Virtually all Industrial Products and & Retail when it comes to current AI maturity. capital pools. (Note: the Services sam- Infrastructure companies look to From using machine learning to detect ple is the smallest of all sectors.) optimize operations fraud and automation to streamlining Companies from the Infrastructure and KYC efforts in the back office, and to Industrial Products Expedited drug discovery and dis- Industrial Products & Manufacturing reducing compliance and regulatory Manufacturing, ease prediction in Life Science Materials, sectors top the list at 96% respectively costs via technologies that digest vast 61% 70% 56% Equipment 96% Executives in Life Science are among in terms of expecting efficiency gains quantities of legal documents, banks those most excited about benefits through AI optimized operations. The and other financial institutions are pertaining to transforming products heavy focus on equipment, complex looking to provide higher quality ser-

TMT and services. Many see AI leveraging supply chains and materials means vice at faster speeds and lower costs. Technology, existing internal and external datasets there is ample scope for intelligent Similarly, mortgage applications can be

Media/Entertainment 81% 52% 88% 64% to speed up the drug discovery process optimization. Yet, there is a relatively approved in a matter of minutes, and & Telecom and enable the transition towards pre- small focus on engaging customers investment decisions can be guided cision medicine. and empowering employees. This is by robo-traders to transform products likely due to the frequent B2B nature of and engage customers in the front Finance Banking, Deep learning with huge datasets is these businesses, and the potential for office. Insurance, also expected to assist with disease automated machinery to play an ev- 67% 73% 78% 84% Investments prediction. Customers can be engaged er-growing role in the industrial sector. using new health-oriented IoT-related

Infrastructure Transportation, Energy, Construction, 70% 74% 96% 54% Real Estate As a railway company, we have significant physical assets that need to be maintained. With AI we see significant opportunities, like automatically detecting faults in railway tracks and Services Professional Services, predicting maintenance needs. This improves not only efficiency but also security. Hospitality, Public 78% 78% Services, Membership 83% 83% Organization — SBB Swiss Federal Railways Railway company Affirmative responses by sector

48 49 Artificial Intelligence in Europe Business Benefits and Risks

Risky Business? What do business leaders need to pay attention to when implementing AI?

There are inevitable concerns about need to take advantage of solutions employees, rather than replace them AI is at the top of our agenda. Currently, we the business risks associated with AI, in accordance with everything from altogether, allowing for more peo- as many of the applications of the rel- GDPR to cybersecurity concerns. For ple-oriented or creative work. There is primarily use it to enhance and support R&D, atively new technology are still in their the latter, the lack of clarity around AI also a larger task in training employees early development while receiving sig- regulation can slow down scaled im- to work together with AI, usually a nificant media and political attention. plementation as business leaders worry challenge and risk in itself. but it also has significant potential for other However, from what business leaders about investing in solutions when the tell us, they are balancing their excite- rulebook is still being written. Many Seeing the wood for trees business areas, such as enhancing market and ment about AI’s potential with some first movers within our AI report feel A further dominant risk articulated by healthy reflections on key business they need to write the rules themselves several surveyed business leaders is risks, not at least the risk of investing and hope for the best. about feeling information overload. AI patient insights. in a technology that may not prove its can help make sense of huge quantities commercial value if not done correctly. Concern with the human in the new of data, but setting up AI and learning machine age to use it effectively requires feeding Broad concern with regulatory A prevailing risk many companies were the technology the right data and landscape also concerned with was impact on working out what is useful versus what — H. Lundbeck Pharmaceutical company Over half of all companies surveyed ex- personnel. The need for employees is noise. A further element in the risk of pressed concern regarding regulatory across the organization to buy in and overload is understanding the different requirements. This concern can broadly adapt to working with AI touches on all AI technologies and solutions available be split into compliance with existing industries and markets. The instinctual and making sense of technological as requirements and navigating the nas- fear of job losses among personnel is well as market developments to know cent, often ill-defined regulatory land- one that needs to be managed as AI where to make strategic use of AI. scape for AI. For the former, companies will often transform the daily tasks of

Top 3 business risks in Denmark We are working closely with senior sales Regulatory Impact on Upkeep of the and relationship staff to better understand 1 Requirements 2 Personnel 3 system what they are looking at when they screen 64% 36% 36%

leads and clients – this allows us to emulate, Danish companies expressed concern AI is not just about technology. The The pace of change in AI technology about regulatory requirements, and in adoption of AI is equally about change is very rapid. Companies fear that a automate, and often also optimize the human particular, the need for clear guidelines management, including culture and technology they invest in now will be and regulations regarding AI. With- mindset shifts. It requires balancing outdated before it has achieved its out such clarity, investment in AI can employees’ fear of losing their jobs expected ROI. Additionally, as with decision making with AI. be perceived as risky for companies with the awareness that once AI frees any system, it requires planned and because they may invest in something them from menial tasks, they will gain unplanned maintenance, including allowed at the time that may not be more time to help customers and de- updating its servers and the data it later on. Danish companies shares this velop products and services. It is also uses. Companies tell us it is difficult to Investment Bank concern with other European countries about recruiting new employees and demonstrate business cases with small — Saxo Bank surveyed: for almost all of them, regu- supporting current employees as they pilot tests, and therefore, hard to know latory requirements was one of the top adopt new kinds of technology. where to invest initially and over time. three risks.

Note: Affirmative responses, Denmark. The respondents were asked to select all that applied of the following response options included: Diffusion of resources, 50 51 Loss of control, Upkeep of the system, Information overload, Regulatory requirements, Impact on personnel. Artificial Intelligence in Europe Learn fom the Leaders

Capabilities. How? What competencies are required to get AI right?

This section explores the necessary As the majority of companies we spoke eight capabilities to develop AI matu- to are looking to supplement their 8 capabilities rity, realize tangible business benefits, in-house skills with external partners 1. Advanced Analytics and minimize risk. As exhibited in the when building their AI solutions, par- Obtaining and deploying specialized chart on the following page, we asked ticularly for pilot projects, it is not due data science skills to work with AI by the companies to rank the importance to a general lack of relevance. attracting talent and working with external parties of these capabilities in terms of incor- Learn from porating AI into their business, as well Bringing behavioral science into play 2. Data Management as to self-assess how competent their via Emotional Intelligence to build Capturing, storing, structuring, companies are with regards to each AI solutions that understand and mimic labeling, accessing and enabling capability. human behavior, and make it easier for understanding data to build the humans to interact with the technolo- foundation and infrastructure to work with AI technologies The human element and technology gy, is seen as the relatively least important AI enabling capability. An explana- the Leaders Some of the eight capabilities center 3. AI Leadership around human elements: AI Leader- tion for this could be that the technical The ability to lead a transformation ship; Open Culture; Agile Development; skills are still so relatively complex for that leverages AI technology to set companies to grasp and establish, that defined goals, capture business Emotional Intelligence. Others are value and achieve broadly based more technology oriented: Advanced more advanced human cognitive skills internal and external buy-in by the Analytics; Data Management; Emerg- become less of a priority at this stage. organization ing Tech; External Alliances. The promise of AI lies in creating business value. Noticeable sector deviation 4. Open Culture Ranking of key capabilities for real- As exhibited in the following chart Creating an open culture in which where business leaders are asked how people embrace change, work to izing AI potential break down silos, and collaborate competent their company is in relation We have identified the eight most recognized capabilities needed Advanced Analytics comes out on across the organization and with to the most important AI enabling external parties top as the most important AI enabling to successfully create value from AI, and assessed how compe- capabilities, the sector aggregate capability among the companies sur- scores land at or just above the medi- 5. Emerging Tech veyed. Data Management is second. tent companies are within each. an, with a fairly close spread. Sectors The organizational-wide capability AI Leadership is perceived as the third to continuously discover, explore that are more mature in using AI are most important capability. Open Cul- and materialize value from new those that report higher competency solutions, applications, and data ture refers to collaboration and the in Advanced Analytics - particularly platforms Perhaps more importantly, the executives we spoke with high- ability to embrace change and uncer- TMT (Telecom, Media/Entertainment & tainty. lighted the importance of these 8 competencies as those needed Technology), as well as Finance (includ- 6. Agile Development An experimental approach in which ing Banking, Investment & Insurance), Understanding how to deploy the collaborative, cross-functional teams to successfully create value from AI. and Life Sciences (including Healthcare work in short project cycles and right Emerging Technologies in a future & Pharma) all report lower competency iterative processes to effectively proven way is ranked fifth, followed by advance AI solutions in AI Leadership. A possibility is that in Agile Development, where self-organ- the pharmaceutical industry, AI chiefly ized teams are characterized by shorter 7. External Alliances resides in R&D, and has yet to affect project cycles, the ability to work with Entering into partnerships and the broader organization on the wider alliances with third party solution constantly evolving technology, and strategic level. providers, technical specialists, and transparency regarding success and business advisors to access technical failure that leads to wider buy-in and capabilities, best practices - and Companies intend to use various levers talent scaling. to obtain these AI capabilities. Compa- nies are relatively evenly split between 8. Emotional Intelligence Entering into External Partnerships using recruitment (60%), training Applying behavioral science ranks second to last in terms of im- capabilities to understand and mimic (56%), partnering (57%). Only 10% of portance, perhaps because it’s the human behavior, address human the companies use acquisition of teams needs, and enable ways to interact area that resonates most with existing or businesses as a way to fast track with technology and develop more capabilities and where business leaders human-like applications building much needed AI capabilities. perceive themselves most in control.

52 53 Learn fom the Leaders Learn fom the Leaders

AI Competency Model

Advanced Analytics and Data management considered most important AI capability TMT leads the other sectors in AI competency How competent is your company within these organizational capabilities? How competent is your company within these organizational capabilities? How important is each of the organizational capabilities for your success with AI?

Competency Importance

IndustrialLife ScienceProductsFinanceCPR Infrastructure Services TMT

15 European 15 European Denmark markets markets Advanced Denmark Advanced Analytics Analytics

Data Data Management Management

AI Leadership AI Leadership

Open Culture Open Culture

Emerging Tech Emerging Tech

Agile Agile Development Development

External External Alliances Alliances

Emotional Emotional Intelligence Intelligence

2,0 2,5 3,0 3,5 4,0 4,5 5,0 2,0 2,5 3,0 3,5 4,0 4,5

Note: ‘Don’t know’ answers not included in average score. Note: ‘Don’t know’ answers not included in average score. Average competency and importance for Denmark and 15 European markets (1: lowest – 5: highest). Average competency by sector (1: lowest – 5: highest). Capabilities ranked according to highest importance in 15 European markets. 54 55 Learn from the Leaders Learn from the Leaders

Companies consider themselves moderately competent within Advanced Analytics How competent is your company within Advanced Analytics? 1. Advanced Analytics Avg. Score

4.5 4.4

36% 36% Importance 28% 27% 20% 18% Obtaining and deploying specialized data science, data engineer- 12% 14% 4% 3.2 3.3 ing, data architecture and data visualization skills by training em- 0% Competency ployees, attracting talent and co-creating with external partners 1 2 3 4 5

Not competent Moderately competent Highly competent Limited analytics skills beyond tra- Semi-autonomous examination of Significant data science activity using ditional business intelligence based data using sophisticated techniques techniques such as complex event primarily on historical data and tools to predict future events processing, neural networks, etc. The backbone of AI is made up of In other words, the longer you wait, the skilled, intelligent minds who are ca- harder it can be to get the right people. 15 European markets Denmark Note: Remaining percent are ‘Don’t know’ responses pable of understanding business prob- Consequently, a ‘wait-and-see’ strategy lems at the granular level, and deploy- can be risky for companies that are AI ing AI to effectively solve or support followers due to the scarcity of talent, Co-creating to compensate for blind others in solving these problems. This which may prove impossible to attract spots - while avoiding the black box Advanced Analytics is a key requires technical data science and once the company is ready to make a The scarcity of available talent has led priority for companies What to learn mathematical engineering skills, to more ambitious move into AI. companies to increasingly co-create in Denmark hybrid profiles with sufficient business from AI leaders: solutions with external partners who Across all markets surveyed, Ad- acumen to decode problems and abil- While many companies struggle with bring with them specialized know-how. vanced Analytics is considered ity to tackle them using quantitative acquiring AI talent, we also experi- 1. Providing interesting However, executives very clearly point one of the most important of methods. enced companies - even in traditional problems, good data, and a to the need for internal AI capabilities the eight capabilities necessary industries such as Transportation and freedom to thrive in a non- in the receiving end to understand the for success with AI – on a scale A self-fulfilling talent prophecy Industrial Products - with AI teams of corporate environment is real problems and evaluate the perfor- of 1 to 5, the average score in +25 experienced data scientists hold- key to attracting talent. It is evident from the study that there mance of external partners. Denmark is slightly above the ing Ph.D’s in mathematics, astrophysics, is a major lack of technical data skills to European average (4.5 vs. 4.4). 2. A wait-and-see follower etc., from high profile universities. Most meet the drastically rising demand for Companies find that AI solutions im- In terms of their competency, strategy can prove risky and often, these companies have been AI. As a result, the hunt for AI experts plemented by external parties become 76% of Danish companies report put companies in a talent first movers on AI and attracted senior has become extremely competitive, black boxes unless the organization to be moderately competent scarcity trap. practitioners tasked with building out and it is far from uncommon that func- is capable of contributing and taking or above (3.2 average). This sizeable AI communities to work on the tional AI experts are paid higher sala- over the solutions after delivery. Avoid- demonstrates the room for 3. Training existing staff with most strategic business agendas. ries than their superiors are - in some ing black boxes is a general concern growth in getting companies deep business intrinsics is cases leading to new HR policies to among executives. Consequently, in- ready for something they con- key to make AI work - and Hybrid profiles becoming the reflect evolving requirements. ternal data scientists must be able to sider to be important. The com- effective when access to The talent pool is getting hardest currency decode and dissect AI applications to panies interviewed talk about talent is challenged. heavier on statistics Several business leaders state that the One of the most consistent inputs from explain of the underlying rationales. their efforts to increase their capabilities - this is getting lack of AI talent is the greatest barrier the executives was the need for people competency in this area, and in to implementation within business op- with deep domain knowledge com- mainstream, it is no longer Such rationales are important in mak- particular, mention challenges erations. Interestingly, companies that bined with strong technology profi- ing AI driven solutions creditable, and around finding the needed skills just for nerds. What really have chosen an early adopter strategy ciency. This hybrid profile is essential to greatly reduce the risk that an AI appli- and personnel, as well as appro- makes a difference is for AI have been successful in attract- identify relevant use-cases in the busi- cation draws wrong conclusions based priate applications of advanced The real challenge of today hybrids that understand ing senior professionals who again ness with possible AI solutions. on false assumptions. analytics. is to recruit people with the have been able to build out sizeable AI both the technical and right skillsets. We either teams in their companies – based on Contrary to data scientists, software business aspects. the premise that talents seek talent – engineers, and even data architects have to train more people making AI recruitment a self-fulfilling that can be recruited externally, the ourselves or recruit people — Chr. Hansen prophecy for these pioneering com- hybrid profile is often nurtured by from abroad. Bioscience company panies. training existing employees from the line of business and adding AI skills. To succeed however, a fundamental ap- — VodafoneZiggo preciation for technology is required. Telecommunications company 56 57 Learn from the Leaders Learn from the Leaders

A significant share of companies consider themselves moderately to highly competent within Data Management 2. Data Management How competent is your company within Data Management? Avg. Score

4.7 4.4

Capturing, storing, structuring, labeling, accessing and govern- 36% 36% 38% 31% Importance 19% 16% 12% ing data to build the foundation and infrastructure to work with 8% 3% 3.0 3.2

AI technologies 0% Competency 1 2 3 4 5

Not competent Moderately competent Highly competent Limited discipline and method to en- Traditional DM with embedded busi- Agile governance, processing of data sure currency and quality of reference ness rules to validate, reconcile, and from disparate sources incl. external data across subjects and systems align data to corporate policy networks, fast query access, etc.

Companies tend to focus their AI Companies reported that they typically 15 European markets Denmark Note: Remaining percent are ‘Don’t know’ responses efforts in areas where they already spend 2-3 years building the appropri- have relevant data. We found that the ate data infrastructure for AI, and many amount of available data varies sig- respondents with the most ambitious AI nificantly by sector but regardlessly, a visions are still spending the majority of significant proportion of the time com- their time fine-tuning their infrastructure. use mostly structured data from internal Similarly, 60% of these self-rated most panies dedicate to AI is spent on data data sources, a significant 80% of the advanced companies report use of hy- management related tasks. Data privacy regulations most advanced companies also use both brid architectures of on-premise and What to learn Data infrastructure is not only a prereq- structured and unstructured data, and cloud based storage, while the less ad- from AI leaders: Data governance is no trivial task uisite for effectively working with AI, but an equivalant 80% use data from both vanced predominatly rely on internal and external sources. on-premise platforms. One of the major hurdles companies is increasingly needed to comply with 1. Make sure that the value face regarding data is governance, data privacy regulations, which respond- of data is understood and particularly who ‘owns’ it, how data is ents see as a key risk. The recent imple- prioritized throughout Data Management is the most important capability in Denmark stored, how to access it, and who may mentation of GDPR in the EU has high- the organization. access it are all essential questions lighted the need to govern what data is Danish companies rate Data Management as the most important of the when working with AI. Questions that being used for. AI-specific regulation is in eight capabilities necessary to succeed with AI (4.7 average on a scale of 1 to 2. Engage the C-suite in used to be about efficiency suddenly many ways still immature, and AI leaders 5) – above the European average (4.4). Despite the high level of importance, defining data governance become highly strategical and com- find that a lack of clear guidelines can Danish respondents report on average to be only moderately competent and strategy - it is key to plex to respond to without rethinking limit their progress. in this capability (3.0). This suggests that some companies have developed getting AI right. governance structure and policy. Gov- a Data Management foundation but are still midway before achieving the ernance aside, the most common ob- Advanced companies (also) appreci- capability level that will fully back their AI systems. Introducing an adequate 3. Build your data structure stacles to using data are organizational ate external and unstructered data data governance structure and finding the right quantity and quality of data to embrace unstructured data, also from external silos or legacy systems built for specific To build precise and useful AI solutions, is essential according to many of the companies interviewed. sources - advanced purposes, resulting in decentralized companies not only need a lot of data, companies indicate that storage that limits access. but also accurate data that is appro- you may soon need it. priately structured and labeled. Data is often reported to be in a state that it is We can provide a more personalized service to simply unusable, as it could lead to unde- our guests, both before check-in, during the stay sirable or unreliable outcomes. and after check-out. Content personalization and While most companies are preoccupied recommendations will further improve customer with cleaning, structuring and migrating The ethical use of data is a challenge or risk. Data must be stored properly. The person who engagement. historical data, some have chosen to build new data structures from scratch to generates the data is also the owner of the data, and that person has to decide what to do with it. collect the correct data going forward. — Grupo Pestana Interestingly, we found that while com- Hotel chain — Royal Philips panies that are less mature in AI tend to Health technology company

58 59 Learn from the Leaders Learn from the Leaders

A large proportion of companies consider themselves to have limited or no AI Leadership competency 3. AI Leadership How competent is your company within AI Leadership? Avg. Score

4.1 4.2

32% 32% Importance The ability to lead an AI transformation from top to bottom - by 24% 20% 24% 23% 12% 10% 8% 9% articulating a vision, setting goals and securing broad buy-in 2.9 2.9 across the organization Competency 1 2 3 4 5

Not competent Moderately competent Highly competent Limited strategic priority assigned Substantial resources deployed to AI, AI recognized as a key strategic pri- to AI activities and only vague fo- mandates assigned, and leadership ority, with strong C-suite sponsorship cus on AI in the management team articulation of an AI vision and high tolerance of uncertainty As with any corporate transformation, enough in itself for understanding how the foundation for successful deploy- AI is impacting the business. As AI tech- ment of AI is executive leadership nologies become increasingly complex, 15 European markets Denmark Note: Remaining percent are ‘Don’t know’ responses buy-in and sponsorship. The C-suite leaders must be able to launch, sup- must be aligned in what they want port and, where necessary, challenge to achieve, and AI must be placed on relevant AI initiatives against strategic Accepting loss of control AI Leadership is a high priori- the strategic agenda to ensure that AI business imperatives. The disruptive As new technological opportunities ty for companies in Denmark efforts are an integrated part of the potential that companies believe AI will foster innovative, dynamic business What to learn Danish companies consider AI company’s overall strategic goals, that have also means that leaders should models, organizations will need to tear Leadership to be the third most from AI leaders: capital is allocated, and employee time anticipate and prepare for a broader down silos to become more agile and important capability to succeed is dedicated. change management exercise aimed collaborative. To achieve this change, 1. The organizational with AI, (4.1 average on a scale of at embracing the change from AI on it is paramount for leaders to create transformation driven by 1 to 5). However, Danish compa- AI Leadership among the lowest multiple levels. and convincingly articulate a vision so AI will be continuous - nies rate their competence with competency of all capabilities stakeholders understand the bigger this requires seeing AI as a AI Leadership among the lowest Significant variation in AI conversa- picture. process, not a project. Given the relative importance of AI of the eight capabilities (2.9 tions from top to bottom Leadership (avg. 4.2 across all sectors), average). The European average A general characteristic of this chal- 2. Leadership must it is interesting to see that business Interestingly, data revealed that AI is (2.9) is also similarly well below lenge is that leadership needs to accept be accustomed to leaders self-assess their level of com- considered an “important topic” on the the importance average rating. that it will lose some control. Projects AI technologies to petency as among the lowest of all C-suite level among 73% of the compa- This likely reflects that many of will increasingly be explorative, bot- understand how it will eight AI enabling capabilities, with an nies surveyed. However, less so on the the companies surveyed have tom-up and have less certain out- affect the company. avg. competency of only 2.9; 66% of Board of Director level where it is only gone through an initial digital comes, requiring leaders to be ready to respondents state that their companies considered an important topic in 38% transformation and are now 3. Articulating a clear AI adjust the overall direction of the com- have moderate, little or no AI Leader- of companies, and even less so on the starting to develop their AI lead- vision is key to achieving pany more frequently. Increasingly, AI ship competency. Many executives are operational employee level with 28%. ership competencies. Further- buy-in and motivating projects will rely on open source code realizing that business acumen is not more, many of the companies in- exploration of use-cases and off-site cloud solutions, building We observed in the interviews that terviewed referred to major tech with uncertain outcomes. on collaborative capabilities outside companies very rarely have AI capable companies when assessing their the company. leaders across the Board of Directors, AI position; with those reference Executive Management, and Functional points, many companies consid- Management layers. Senior AI leaders ere themselves not to be highly can sometimes be found on one of the competent in AI Leadership. We have already collected our data in a data lake for levels, but rarely with any speaking almost 3 years. We have worked with data collection and leadership colleagues to challenge ide- adding analytics into our processes for some time and as. This leadership vacuum was often pointed to as an issue from lower level that is why we are now ready for further developments. AI experts.

— Telia Telecommunications company

60 61 Learn from the Leaders Learn from the Leaders

Most companies rate themselves moderately competent in Open Culture How competent is your company within creating an Open Culture? 4. Open Culture Avg. Score

3.8 3.9

32% 36% 41% Importance 24% 23% 18% 13% Creating an open culture in which people embrace change from AI, 4% 4% 2.9 3.2 navigate confidently in uncertainty and ambiguity, work to break 0% Competency down silos, and collaborate seamlessly across the organization 1 2 3 4 5

Not competent Moderately competent Highly competent Gravitation towards linear processes, Emerging transparency, participation, Truly collaborative and open approach, and organization of AI in separate empowerment, community building to with clear accountability across both silos within existing structures increase speed and agility internal and external domains New technologies have often disrupted drive a fundamental transformation

how work is conducted. AI is no differ- and increasingly assist in tasks previ- 15 European markets Denmark Note: Remaining percent are ‘Don’t know’ responses ent. Establishing an open, collaborative ously performed by humans. culture to minimize resistance and enable human performance can prove Interestingly, the companies that efficient to prepare the organization self-rated as most advanced see a ken down in order to promote a culture for transition. However, this may be lower risk to personnel than the less where AI-teams work in conjunction Denmark rates Open Culture as difficult, as the magnitude of impact advanced (only 20% of advanced re- with the rest of the company to create an important capability What to learn driven by AI can imply a fear of uncer- ported this risk as a concern vs. 43% value, circumventing needless complex- Even though Danish companies from AI leaders: tainty, ambiguity, and a general resist- for the companies still in the “planning” ity and time-consuming processes. on average consider Open Culture ance to change. phase). to be important (3.8), it has one 1. Establish cross- Another issue relates to the concept of of the lowest importance levels organizational projects to Risk to employees less of a concern Relatively small competency gap sharing data openly, when the value of compared to the other capabili- foster collaboration and among most advanced companies the data largely remains unknown until With a relatively small gap between ties. In terms of competency, only learning across functions. Companies reported that employees importance (avg. 3.9) and competency it has been treated, processed or com- 4% of Danish companies report to generally have a positive attitude (avg. 3.2), creating an Open Culture is bined with other datasets. be highly competent, and overall 2. Ensure employee buy-in towards AI. Yet, one thing is having a one of the capabilities where business they report to be below moder- by being open and clear positive attitude in general, another is leaders feel most comfortable. Cooperation across the organization ately competent in Open Culture about on-going projects to retain an open attitude once new Many of the most advanced companies (2.9 average). The results are and desired outcomes. technologies start impacting the way An obstacle mentioned by many re- that have been able to produce several quite surprising given the general work is done. spondents is the ability to work col- AI projects have also managed to es- Nordic approach to cross-func- 3. Ensure that governance laboratively across the organization tablish links and cooperation across the tional, cross-organizational ways structures support To achieve buy-in, business leaders despite AI most often being put to use organization. These cases indicate that of working, but perhaps suggest collaboration through must make the changes due to AI towards quite narrow use-cases. With the benefits of an open work culture far awareness of further room for projects co-owned by tangible to reduce organizational un- benefit areas being limited to specific exceed the difficulties and associated improvement. AI experts and business certainty. However, companies expect domains or functions, it is often not risks. leaders. a significant impact from AI which will seen as relevant to involve the organization in a broad and collaborative An obvious obstacle to an open culture approach on AI. is the fear of job losses with the intro- You cannot only have data scientists do it. They have a super duction of AI. According to respond- My philosophy is that experiencing and experimenting is important role but you also have to complement them with Furthermore, many companies have ents, the fear of workforce redundancy the only way forward. We can spend an infinite amount designers. Because you need to find the use cases where had difficulties in carrying out effective has some merit, but the concern should of time and resources on thought leadership but when we AI programs, which are closely mod- not overshadow the significant benefit you apply those types of technologies. Even though it is the first get stuff done, when we start managing the craft and elled on the lean processes of startups. potential of AI. A pivotal task for com- fantastic technology that can bring fantastic results, it has The primary purpose of such programs pany leaders is to proactively articulate how to pivot, it will not be about the technology but about to be embedded in a design approach to meet the customer is to enable brief, agile projects to a tangible vision for AI initiatives. This what we are asking and looking for. needs and solve real problems. gauge the applicability of AI use-cases, will make it easier for employees to requiring a substantial change to com- understand the AI opportunities on a — EQT pany culture. Silos between depart- personal level, and thereby embrace — IKEA Group Private equity group ments in the company have to be bro- the change ahead. Furniture retail company

62 63 Learn from the Leaders Learn from the Leaders

Emerging Technology is the AI-enabling capability with most ‘Moderately Competent’ replies 5. Emerging Technology How competent is your company within adopting Emerging Technology? Avg. Score

3.9 3.9

48% 42% Importance 28% 29% 13% 12% 12% 8% The organization-wide ability to continuously discover, de- 3% 2.9 3.3 ploy, and create value from intelligent solutions, applications, 0% Competency and data platforms 1 2 3 4 5

Not competent Moderately competent Highly competent Legacy oriented tech stack, typically Increasingly AI enabled tech stack with Tech stack optimized for AI across on-premise based and with limited on-demand cloud computing, agile hardware, interface, algorithms, future proofing of AI technologies software, scalable architecture, etc. architecture, people, services, etc.

15 European markets Denmark Note: Remaining percent are ‘Don’t know’ responses Evidence of the rapid pace of techno- that being unable to quickly integrate wider digital puzzle, where dots need to logical change are plentiful in today’s innovative trends and cutting edge be connected across technologies. This digital world. What we have seen is that technology due to the burden of means success with established technol- there is a definite correlation between legacy systems, siloed business units, ogies, from cloud and SaaS platforms to The importance of execution Emerging Technology seen as companies that are ahead of the pack and complex governance processes getting the basics right with analytics, is Finally, this capability is also effective an important capability for with AI and with the wider technolog- is proving a real challenge for their AI key to building on what is already there. execution. Many companies we sur- What to learn Danish companies ical adoption. That AI benefits from adoption. veyed across Europe had developed from AI leaders: being able to identify and implement Working with emerging technology also prosperous use cases supported by Danish companies consider emerging technology may seem intu- While there is some truth behind such relates to agile development and the abil- Emerging Technology an im- robust concepts and AI applications - 1. Build a radar to pick up on itive and obvious, yet finding the right stereotypes, we also heard from several ity to trial, test and experiment in iterative, portant capability (3.9 average), on paper. But technical limitations tend merging tech trends and formula is no trivial exercise. executives who are able to build radars short cycles. This kind of working culture same as the European average to get in the way of implementation. connect them to market that pick up what’s happening in tech- allows companies to work with less stable, (3.9). In terms of competency, opportunities. How strong is your tech radar? nology domains and applications that untested technology. Enabling innovation Employees with limited technical ability same as with Open Culture, With an average score of 3.3, the ability this continuous explorative process is requires an outlook from the very top often need upskilling to work with new Danish companies report to be 2. Look past the technology to explore and implement emerging serving them well to get an overview of of the organization that accommodates technology. IT and business may need on average below moderately hype and remember the technology is an area where business workable AI solutions that could prove longer investment horizons and at times to work closely together and speak competent (2.9). This result business model - it may leaders perceive their companies to successful in production. uncertain financial returns. This is particu- each other’s languages to reach com- suggests companies are very likely need to change in be most competent across the eight AI larly key when working with AI technology mon goals. In addition, organizations much aware of the importance the not so distant future. enabling capability areas. Do you enable or hinder innovation? that, according to the executives, is often need to learn to move more quickly of many of these capabilities; Once companies are able to selectively not as mature as the digital solutions and nimbly in this space - whether to however still consider there to 3. Cloud solutions can be One factor in working with emerging source new solutions from the outside deployed for other purposes. complete an acquisition of new tech, to be a gap until they reach the helpful to engage with and rapidly developing technology to world, the challenge is then how to ensure compliance with IT standards, desired level of competency to multiple datasets across build a stack fit for AI is a well-calibrat- enable it. This can be a case of actively Not all that glitters is gold or simply to pair new tech with legacy leverage them. sources - increasingly a ed ‘radar’ by which large companies encouraging enablement, or at the Despite the need to explore and navigate systems. This ability is often also about priority to capture value pick up on the trends outside of their very least not hindering it. Many com- a tech sea characterized by uncertainty, a speed, not far from the development from new pockets. own walls. Many companies mention panies treat AI as a crucial piece of a recurring theme when interviewing exec- pace that characterizes the emerging utives is the importance of balancing ex- tech itself. citement with new technology and com- mitment to an innovative mindset, with one foot planted firmly on the ground.

A big challenge is to follow all the rapid Seeing past the hype, remembering the Treating AI as a distinct part of wider digital initiatives, Maersk established business model, and not wasting finite evolutions in the market and match that to the an in-house software development and innovation unit, consisting of 100 resources on every shiny object is also im- right business initiative. portant. In other words, remembering as a employees and growing. leader that while experimenting is crucial, — DAF Trucks not all that glitters is gold. — A.P. Moller - Maersk Manufacturing company Shipping company

64 65 Learn from the Leaders Learn from the Leaders

Companies seem relatively competent within Agile Development How competent is your company within Agile Development? 6. Agile Development Avg. Score

3.9 3.8

44% Importance 37% 28% 30% 16% 12% 10% 6% 8% An experimental approach in which collaborative, 4% 3.2 3.2 cross-functional teams work in short, iterative project cycles to Competency effectively progress AI solutions 1 2 3 4 5

Not competent Moderately competent Highly competent Linear approach without iteration Applied agile principles and method- Agile development fully deployed, with loops, and limited continuous plan- ologies, but only limited use outside people empowered to make quick deci- ning, testing, integration and feedback software development functions sions together across functions Considering that many AI technologies use-cases. Of the most advanced com- are still in their infancies, working with panies, 80% deploy AI into the organi- 15 European markets Denmark Note: Remaining percent are ‘Don’t know’ responses them is far from plug and play. To over- zation via top down only or a via hybrid come this, many of the companies that of top down and bottom up. are successfully working with AI tend to take an agile, iterative approach to It varies whether these central units projects. Using this approach, these take a leading role in pushing the Agile development new to many Agile Development among high- companies greatly increase their ability agenda, or instead focus on gather- business departments est competencies in Denmark What to learn to explore AI potential due to a dras- ing knowledge and experience from Most companies fully understand the Out of the eight capabilities pre- tically reduced project cycle time and already existing efforts that are decen- need for agile development, but less sented, companies in Denmark from AI leaders: dynamic risk reduction. Short project tralized in the organization. reckon that they have the necessary feel most competent in Agile 1. Agile development is cycles result in project teams receiv- capabilities for it. Working in an agile Development and Advanced effective in engaging ing constant feedback on what works Agility provides the opportunity for manner is very different from what Analytics (3.2 average for both). people across functions, and what does not, to continuously informed changes of direction most organizations are used to. While Among Danish respondents, fostering collaboration, steer the direction of the project. This the department running an AI project 84% report to be moderately Taking an iterative approach can also and bridging tech and creates a process centered on learning might be accustomed to following an competent or above, including help mitigate risks. Frequent feedback business. and experimentation, helping to build loops allow the project team to better agile approach, the vast majority of 12% who consider themselves internal knowledge and capabilities. identify, understand, and correct unde- project teams consist of people from highly competent. Agile De- 2. Iterative processes sired outcomes before the AI applica- other parts of the business. velopment is considered an promotes quick internal Most advanced companies deploy tion is put into production, potentially important capability to succeed learning due to their top down or via a hybrid model doing harm. This flexibility does not Several IT and AI departments indi- with AI (3.9 average) Many of the frequent feedback loops. With an average competence level only apply to risks, as agile projects can cate that this collaboration can be companies in Denmark talked of 3.2, Agile Development is an area generally use continuing knowledge difficult, but still see it as pivotal to about AI pilot projects being 3. Fast experimentation with where companies are self-reported and experience to make informed drive value from the projects. Getting introduced in at least some ar- pilot projects and use- to be reasonably skilled. Quickly es- changes of direction and avoid the the business accustomed to working eas of their organization, which case testing can quickly in an agile manner is not easy, as it can explain the higher reported show how to create value We have learned a lot from tablishing proof of concept is key to “black box” syndrome. organizational buy-in, and many com- requires acceptance of new ways of level of competency in Denmark. through AI. the pilot projects we have panies report that an agile, iterative ap- Contrary to agile projects, ‘big bang’ governing and evaluating projects. According to the respondents, had. You need to take proach helps them build evidence and projects are more destined to fail as in most cases agile AI projects an explorative approach proof in a fraction of the time it takes they skip the learning process, and lack The outcome of agile projects is typ- originate in IT, Data Science or ically less predictable than for tradi- R&D departments. to this and accept that for a more traditional project, the important feedback loop pivotal to developing good AI solutions. The tional projects, and for stakeholders governance and project This has great significance, as they world of AI is simply too complex for to fully embrace an agile approach, management is very find that tangible proof of concept humans to foresee potential issues, and they have to accept this randomness different for these types of instrumental in achieving buy-in and therefore an agile approach is better. and recognize the value of learning. projects. understanding in the wider organization. Efforts to develop proof via agile development processes are often or- — William Demant chestrated by a central unit that collab- Healthcare company orates with business units to identify

66 67 Learn from the Leaders Learn from the Leaders

Companies generally consider themselves moderately to highly competent forging External Alliances How competent is your company within building External Alliances? 7. External Alliances Avg. Score

3.6 3.7

36% 37% Importance Entering into partnerships and alliances with academia, solution 20% 20% 28% 18% 12% 12% providers, and AI specialists to access technical capabilities, best 4% 3% 2.9 3.2 practices and talent Competency 1 2 3 4 5

Not competent Moderately competent Highly competent External relations mostly based on Multiple strategic alliances, often be- Significant alliances with AI partners traditional sourcing of external ven- tween equal partners working collab- in open eco-systems enabling access dors providing specific AI services oratively to mutually benefit from AI to external assets and (big) data AI leaders are increasingly opening up To address one of the biggest prerequi- to create collaborative alliances with sites of working with AI, access to large external partners, enabling them to amounts of data, companies state that 15 European markets Denmark Note: Remaining percent are ‘Don’t know’ responses tap into a significantly larger pool of they are increasingly looking to entering capabilities and talent, and to reduce into data partnerships where they either the time it takes to develop or deploy buy or exchange data with other parties. tioned as very practical issue with AI in working solutions. This is a way for companies to get hold general. This led some companies to of data that they are unable to capture prefer internal teams and individuals in What to learn from This trend seems to be the new modus themselves, or simply a way of quickly order to ensure that despite poor doc- operandi, unfolding across markets and increasing the size of their datasets. umentation, the knowledge about the AI leaders: sectors. It is also the capability with the code at least stays inhouse. 1. Make sure to have internal people in the receiving end before smallest gap between perceived impor- Others report that they look to pre-de- widely engaging with external partners. tance and competence level among the veloped, out-the-box algorithms, in participating companies. order to increase the speed of bringing Denmark below moderately 2. Academic partnerships are an increasingly sought after way quality solutions in to product. competent in External Alliances to access innovative eco-systems, gain new insights, and Technology, data, and service Companies in Denmark consid- explore emerging AI opportunities. delivery partnerships Academia playing a more noticeable er themselves to be on average Development of AI and delivery of re- role in collaborating with companies slightly below moderately com- 3. Partnerships can pose a challenge to many business lated projects are most often done with It is becoming increasingly common for petent in Agile Development (2.9), processes; consider involving key functions like legal early, a mix of internal and external stake- companies to enter into partnerships below the European average (3.2). to ensure a productive partnership structure and effective holders. The rationale is multifaceted – with universities in order to position As with the other capabilities, collaboration model. some companies are simply struggling themselves within AI and get access to Danish companies rate their com- to obtain the needed talent, whereas crucial knowledge. petence with External Alliances others see a partnership approach to be below their rating of its impor- a faster, more flexible solution. These Companies also see this as a way of es- tance (3.6 average). The results external alliances typically come in two tablishing a pipeline of AI talent already suggest that Danish companies forms: being focused on technology familiar with their business and the prob- have engaged in some partner- and technical AI know-how, or focused lems they face. Some of the more ambi- ships and gained some experience on strategy and business development. tious companies have a strategy of posi- from it, yet are in the early phases tioning themselves within AI, comprised of deploying AI and are still try- of active conference participation and ing to figure out what to develop The end-user will expect similar user-experience multiple university partnerships in which internally and when to collaborate and connectivity from hearing aids that they get they actively participate in developing with external parties. We will definitely pursue a partnership strategy. Instead of from smartphones, etc. To go beyond what we courses and programs. trying to build everything in-house we will join forces with offer on today’s connectivity, we need to look for others to build a strong ecosystem. Documentation of code is proving a new partnerships. challenge - also to externals — Nilfisk The lack of code documentation for — William Demant Manufacturing company self-learning algorithms was often men- Healthcare company

68 69 Learn from the Leaders Learn from the Leaders

Companies consider themselves least capable within Emotional Intelligence How competent is your company within applying Emotional Intelligence? 8. Emotional Intelligence Avg. Score

3.1 3.3

32% 34% Importance 24% 29% 28% 14% 15% Applying behavioral science to understand and mimic human be- 8% havior, address needs, improve human-machine interactions, and 2% 2.0 2.4 Competency ultimately create more human near applications 1 2 3 4 5

Not competent Moderately competent Highly competent Limited experience with adopting Appreciation of behavioral science Human-centered approach, ethical mind- behavioral science into the AI devel- and cultivation of uniquely human set and values that engender trust and opment process, at best outsourced skills to increase value from AI human ingenuity beyond business effects

AI has for long focused on cognitive they are still at a relatively low maturity capabilities and skills within math- stage where more immediate require- 15 European markets Denmark Note: Remaining percent are ‘Don’t know’ responses ematics, statistics and logical rea- ments such as Advanced Analytics, soning. Adding human emotion and Data Management and AI Leadership intelligence, these capabilities move are more relevant and prevalent. to a new, more complex level: the un- irony, anger, and frustration. This will Emotional Intelligence: even derstanding of human behavior, and However, when taking a deeper look obviously become more valuable as it the lowest importance rating is What to learn the ability to interact accordingly with at the companies that have assessed is increasingly applied in customer- above moderately important facing solutions with the ability to from AI leaders: technology. themselves to be ‘Advanced’ in terms Danish companies rate Emotional learn and improve. of general AI maturity - meaning that Intelligence moderately impor- 1. The most advanced Changing the way people interact AI is actively contributing to many tant for their success with AI (3.1 companies are putting Human centrism requires strong with technology processes and enabling quite advanced average.), slightly below the Euro- emotional intelligence leadership One of the limits of traditional AI has tasks in the company - it is interesting pean average (3.3). Even so, in the to use within their AI been the inability to understand hu- to see that they perceive the Emotional While emotional intelligence holds respective Danish and European applications, despite its man traits such as emotional state, for Intelligence capability as more im- great potential that could lead to early samples, these ratings are lowest relatively infant stage. instance exhibited in writing, physical portant with a score that is noticeable adopters gaining a competitive advan- of the eight capabilities. Similarly, condition, or tone of voice. With AI’s higher than the average score for all tage, long-term success is dependent the competency with Emotional 2. Companies must develop cognitive intelligence capacities within companies. on not only technological develop- Intelligence is also the lowest of their behavioral science ment, but also leadership. capabilities to mimick It is given that we will see reach, machines are increasingly able Danish and European averages to sense, recognize, and decode hu- Many advanced companies perceive (2.0 vs. 2.4.), which for both is be- human behavior and the rise of AI. All the big tech man traits. this to be either ‘very’ or ‘highly’ Leaders must drive the transformation low moderately competent. The translate it to technology. companies are spending important. Notably, these companies that will make humans comfortable ability to adopt behavioral science 3. Many have virtual money on AI capabilities. This holds the potential to fundamen- come from five different markets and a with intelligent technology, as a pre- in the tech development process assistants, chat bots, and tally change the way people interact wide variety of industries, including Life requisite for harvesting its potential is in its infant stages for most local They have explicit visions NLP a powerful way to with machines, making technology ca- Sciences, Financial Services, TMT, CPR, benefits. What the most advanced companies, which for the most to master human thinking get started with building pable of handling more complex tasks and Services & Hospitality. companies have shown is that this part are still developing their AI emotional intelligence and behavior. That may and ultimately augmenting humans to transformation must augment human strategies. Notably, Emotional In- into AI solutions. or may not happen in the an extent previously unachievable. Value in customer-facing ingenuity to become truly effective. telligence is the only capability in applications next five years but certainly which no Danish company rated Emotional Intelligence in its infancy The need for behavioral science to itself as highly competent. within a certain time frame. Except for advanced companies, survey understand human needs is expected When you combine it with results indicate that companies view to increase with the integration of AI in computing power, it will be the adoption of emotional intelligence smart devices, and in customer facing inevitable. in AI processes as the least important applications such as chat bots, roboad- capability, and the one where they visories, customer inquiries processing, have the lowest competency. When etc. The most advanced companies’ AI — Skandia asked to address why this is, companies technologies are beginning to decode Pensions and insurance across sectors and markets note that human emotions from text, such as company

70 71 Artificial Intelligence in Europe ( Case Study ) Successful Value Creation

EQT

EQT’s AI journey started in 2015, with with addressing commercial challenges looking at a small subset of companies Working with AI requires a completely new fundamental questions about how to for a few companies, with others natu- that are introduced to EQT or found via drive long-term value from digital for rally following. research, to screening a global startup its investors. To address this, manage- ecosystem. It tracks multiple struc- way of working, new governance, project ment brought in talent from leading EQT believes that AI is a disruptive tured and unstructured data points, tech companies to drive change in- technology, and it wanted to leverage public and proprietary, for millions of evaluation, accepting more variation in ternally and across portfolio companies, including web companies. traffic, media mentions, and Internal change was push- AI algorithms are deployed to establish financials. outcomes, and requires longer time horizons. based and transformative. Focus was put on the Cloud, whether successful companies have a AI algorithms are deployed to new tools, and decommis- unique ‘digital signature’ that can be establish whether successful sioning old ones. From a spotted early on, so that EQT can ap- companies have a unique fully outsourced model, EQT proach them proactively. ‘digital signature’ that can — Ørsted Energy company insourced key functions and be spotted early on, so that now owns the source code the fund can approach them itself. Experience, results, and proactively. The platform is overcoming bottlenecks helped build it. It started building a tool to rede- being used in venture capital and pri- its position as a trusted advisor. This fine the core of it business: investing. vate equity research, with predictive was used to drive organic pull-based Motherbrain, a platform that has been analytics under development. change in portfolio companies, starting live since 2016, allows switching from

What next?

EQT is a Swedish private equity group of 27 funds, with over €50 EQT is addressing data-driven challenges in investment decision It’s about having the right mindset. It’s not billion in capital raised. Since its establishment in 1994, EQT has flow. Currently, data gathering and prediction are the most time invested in 210 companies, and exited over 100 investments. The intensive, however, there is a shift in commoditizing these. group has several investment strategies within private capital, Motherbrain’s AI algorithms are used to build predictive analytics that tomorrow everything will be different. real assets and credit. EQT has 540 employees in 14 countries with continual learning loops to enable dynamic, better quality across Europe, Asia and North America. Portfolio companies have predictions, allowing resources to refocus on judgement. It It’s all about building up capabilities and combined total sales of over €19 billion and a combined total of is a lengthy process, as the entire investment flow has to be over 110,000 employees. considered, including identifying companies and comparing them speeding up constantly. The power of to successful companies that EQT looked at or invested in. technology in general is overestimated in the short term and underestimated for the long- There are so many things you have to fix before To be successful with AI, you need data and term and I think that’s the case with AI too. even starting to think about AI. The way we talent. The missing component is the mental see digital and companies that are successful state when you are prepared to try it out, in this space is that they understand their when the moment comes where you choose — VodafoneZiggo Telecommunications company customers and their market in a very granular between recommendations based on old way. methods and AI. If you choose AI, that’s when change truly happens.

72 73 What’s Next for You? What’s Next for You?

4. Build a data strategy and technology stack purposefully fit-for-AI AI Fast Forward Training your AI products essentially requires significant data. Useful data. Valid data. Establishing a solid data strategy and practice in your organization to proficiently acquire data, identify data, clean data, measure data, and manage data will ultimately make your organization flourish with AI. Build How to get started and take AI to the next level? your AI resources around data engineers who organize the data, data scientists that investigates the data, software engineers who develop algorithms and implement applications. Make sure that your structure and governance harness the power of data, and that your technology stack across products, solutions, and applications nimbly enables your AI priorities. When doing so, remember that your business model is likely to change.

Learn more about how to build a flexible platform and portfolio of AI tools and next generation smart applications where your data lives - whether in the intelligent cloud or on-premise

1. Choose a step-by-step approach in getting familiar with AI Given the wide scope of AI and variations in use cases, it is key to start out by identifying what prob- 5. Beyond all, engender trust and enable human ingenuity lems to solve and what opportunities to pursue. High level prioritizing between engaging customers, When designed with people at the center, AI can extend companies’ capabilities, free up creative and optimizing operations, empowering employees and/or transforming products and services adds clar- strategic endeavors, and help achieve more. Humans are the real heroes of AI – design experiences ity, is helpful to structure the discussion on a strategic level, and ensures a step-change approach to that augment and unlock human potential. Opt for a “people first, technology second” approach. This taking the company to the next AI level. Identify the problems you aim for AI to solve, prioritize the entails designing AI for where and how people work, play and live, bridging emotional and cognitive value with business owners, and acknowledge the capability gaps to get there. You need to get on the intelligence, tailoring experiences to how people use technology, respecting differences, and cele- AI train, but do not jump on the AI wagon blindly. AI should serve your business plan, not vice versa. brating the diversity of how people engage, Thereby putting people first, reflects human values and promotes trust in AI solutions. Read more in the blog on Linkedin about “AI for businesses: Not if, but when and how” by Michel van der Bel, Microsoft President, EMEA Learn more in the Microsoft Trust Center and the book ‘The Future Computed’ by Brad Smith and Harry Shum from Microsoft on artificial intelligence and its role in society

2. Display executive leadership and approach AI from a position of strength Leadership comes from the top, also in the case of AI. For this to happen, executives must understand AI essentials and strategic perspectives, and they must communicate a clear AI ambition to the organiza- Designing for people tion. AI leaders must actively sponsor and mobilize AI adoption on all levels, from the Board and Exec- utive levels, through Management and the operational employees. Staying ahead in the accelerating AI At Microsoft we believe that, when designed with people at the center, AI can extend your capabilities, free you up race requires executives to make nimble, informed decisions about where and how to employ AI in their for more creative and strategic endeavors, and help you or your organization achieve more. business. When doing so, look to strongholds before bringing in the AI ‘twist’. Amplifying existing company strengths is an excellent way to catalyze motivation and internal support. The following principles guide the way we design and develop our products: Read more customer stories to see how others are using AI to transform their business, and learn from Microsoft Research on how AI is solving the most pressing challenges • Humans are the heroes. People first, technology second. Design experiences that augment and unlock human potential. • Know the context. Context defines meaning. Design for where and how people work, play, and live. • Balance EQ and IQ. Design experiences that bridge emotional and cognitive intelligence. • Evolve over time. Design for adaptation. Tailor experiences for how people use technology. • Honor societal values. Design to respect differences and celebrate a diversity of experiences. 3. Hire new skills ahead of the curve – or focus relentlessly on training existing talent Innovation is what creates tomorrow. A key challenge for putting AI to productive use and accelerate intended outcomes is the war for skills and talent. This not only relates to data scientists and software engineers, but also to skill sets and expe- Learn about our AI platform to innovate and accelerate with powerful tools and services that bring AI to every rience within human and behavioral science. Opting for a follower strategy and being late to the game developer. can prove risky, as talent seeks to go where talent is already. If aggressive poaching for insourcing talent Explore Intelligent applications where you can experience the intelligence built into Microsoft products and is difficult to embrace, then work bottom-up by training the engineers you already have on the new AI services you use every day. paradigm and collaboratively ride on the backs of the others. Regardless of strategy, focusing relentlessly on building required skills and talent is key to staying ahead and progressing along the learning curve. Learn about AI for business. Use AI to drive digital transformation with accelerators, solutions, and practices that empower your organization. Learn more in the Microsoft AI School about the open-source cognitive toolkit (previously known as CNTK) and how to help train deep learning algorithms

74 75 What’s Next for You?

Who to Contact from Microsoft

The team in Denmark that can empower your company to achieve more with AI

Frederik Braun Peter Kyvsgaard Nana Bule Director Enterprise Commercial, Solutions Sales Manager, Data & AI, Director for Marketing & Operations Microsoft Denmark Microsoft Denmark Microsoft Denmark

Frederik Braun is Director for En- Peter Kyvsgaard is Sales Manager Nana Bule is COO/CMO in Microsoft terprise Commercial in Microsoft for the Data & AI Solution Specialist Denmark, and has lead the unfolding Denmark. He has a passion for team in Microsoft Denmark, espe- of Microsoft’s digitization strategy driving digital transformation with cially focusing on AI and data driven and approach in the Danish market. Microsoft clients while having years business development. He has been She has worked with some of the of experience in closing complex a part of the company since 2015 largest Microsoft clients on their solution sales for the benefit of the and is currently leading our AI Am- digital journey. clients’ business development. bassadors Program. Nana has been in the tech industry Frederik has been a part of Micro- Peter has more than 20 years of ex- for almost two decades, starting her soft for 10 years, latest as Director perience within data and advanced career in eCommerce and leading for the Technical Solution Sales. His data analysis. different sales and marketing teams background is in the financial sec- at Microsoft across the consumer and tor leading business development commercial business. within Corporate & Institutional Banking.

76 77 Contributors from EY

Team responsible for the Denmark edition of the study ‘Artificial Intelligence Report: Outlook for 2019 and Beyond’

Thomas Holm Møller Dr. Ellen Czaika Martin Vester-Christensen Henrik Axelsen

Partner EY | Innovation, Analytics, Digital Senior Manager, EY Partner, EY Advisory Co-founder EY-Box Deputy Leader, EMEIA Denmark

[email protected] [email protected] [email protected] [email protected]

EY-Box is focused on digital Ellen holds a PhD in technol- Martin is a Senior Manager in Henrik rejoined EY’s Advisory strategy, growth ventures, ogy, policy and management EY’s Analytics and Technology practice in Denmark in 2018 innovation architecture from MIT. She has masters practice where he is responsi- to focus on advanced analyt- and tech-led transactions. degrees in engineering man- ble for strategy consulting in ics. He most recently led EY’s Thomas works with leading agement and system design EY’s machine learning team. European prudential practice companies to uncover plau- from MIT and in applied He has hands-on experience in out of London, focusing on sible futures, launch new statistics from the University implementing machine learn- future of risk for European businesses, and rewire their of Oxford. Ellen advised this ing and AI in various industries, banks, more specifically the core through data and digital study on research design, particularly retail, consumer impact of AI on risk, capi- in the search for new profit methodology, and analysis. goods, pharma and financial tal, liquidity, processes and pools and business mod- service organizations. Martin’s governance. Henrik holds a els. He serves on the board Ellen is engaged in the EY contribution to EY’s AI teams master in law from University for several entrepreneurial EMEIA Center of Excellence and practices builds on his sol- of Copenhagen and an MBA growth-stage businesses. on innovation, analytics, and id theoretical foundation from from Copenhagen Business digital. She has worked with a PhD in Applied Mathematics School. He has had a long Thomas is responsible for the global organizations and and an MBA in Management consulting career across sec- AI study across 15 markets start-ups, having recently of Technology from the Tech- tors working with analytics in collaboration with central served as the head of R&D for nical University of Denmark. prior to focusing on AI, main- and local EY strategy teams a precision Ag startup that ly in the finance, industrial and AI specialists. uses AI to assist farmers. products and energy sectors.

Based in Copenhagen. Based in Zürich. Based in Copenhagen. Based in Copenhagen.