Forget the Sci-Fi and Embrace

STRATEGIES the Engineering #connectedcar

AI in Automotive

Adaptivity, connectivity and artificial intelligence Adaptivity, connectivity and artificial intelligence Many auto-manufacturers have already made extensive use of AI and invested in machine learning for connected Many auto-manufacturers have already made extensive use of AI and invested in machine learning for connected cars and autonomous driving. cars and autonomous driving. However the race is on to make the smartest vehicle technology IN 30 SECONDS However the race is on to make the smartest vehicle technology

• Today we can design systems to Situation respond to unplanned yet known The automotive industry is transforming as data and events, but humans struggle to cope connectivity bring new with the challenges provided by opportunities for using AI in completely unexpected situations infotainment, self-driving cars and learning from vehicle data Solutions • Understand the kinds of AI • OEMs need to find the right prob- tools and techniques, know lems to solve and the right types of Complications what kinds of data and services and what specialist AI is in its early stages of AI to solve them expertise is required development and there are a plethora of solutions doing • Find the right problems to • Understanding how data and AI dierent things. AI also means solve and pair them with the can be used will help you to differ- dierent things to dierent right kinds of AI entiate your business people • Find opportunities to use AI internally and in customer- facing products and services • Set expectations: be prepared to Challenges to innovate with AI to fail - and to succeed - in unexpected Dierent kinds of AI with leapfrog the competition dierent applicabilities, ways idiosyncrasies, strengths and weaknesses, and when AI doesn’t deliver expectations and illusions are shattered

Source: BearingPoint Institute Introduction Advances in computer power and the development of machine 2 umanity has been fascinated by learning offer huge potential for automation, prediction, and Hartificial intelligence since it has generation of insights from patterns in the data that humans been able to conceive of the idea. An fail to see. Ancient Greek myth imagined “Talos”, a STRATEGIES #connectedcar giant bronze automaton, built to protect the Cretan town of Europa from pirates. Wolfgang von Kempelen built “The Turk” Introduction in the 1700s, which he claimed was a But can we build machines that will deal Applications for the automotive sector mechanical man able to play chess. It with unexpected circumstances? You are exciting. Market research specialist Automotive AI in practice could indeed play chess – and win – but can design a system to respond to a Gartner now expects there to be 250 there was a real man hiding inside. number of unplanned yet known events million connected cars on the road by that might arise, but to cope with the 2020. Those connected cars will be Barriers to AI adoption Today we have chat bots we can talk completely unexpected is a challenge generating a high volume of data that to, robots that are taught by engineers many humans struggle with. Fully will be used to generate insights into Ambition with reality to build cars, and cars that can drive autonomous cars will come in time, but customer behavior; insights that can themselves (to some degree). But much work has to be done before they inform new product design, as well as while you can ask Alexa, Siri, Cortana, Conclusion can safely be put into situations that improvements and add-ons that can be Google Assistant, and other digital cannot be controlled. provided as part of Software Over The Academic perspective agents questions - or tell them to Air (SOTA) updates. Forrester Research control your lights or heating - try to We have come on in leaps and predicts “Artificial intelligence will drive Key takeaways hold a conversation with them and bounds, but we must be able to trust the insights revolution”2 and that “those most autonomous chat bots are lost for autonomous vehicles with our lives. So, that are truly insights-driven businesses words. About the author while to some it is magical and mystical will steal $1.2 trillion per annum from that machines can do such things, we Still, while we are a long way from their less-informed peers by 2020”. need to understand the technology we building anything that would be Notes and bibliography are dealing with: there is no room for As well as generating insights for what mistaken for human intelligence, we do admiring the magic – this is engineering, products to create, at least one attempt have versatile automated machines, pure and simple. has been made to “grow the actual not least in the automotive sector. product” by creating an autonomous car Navigation systems can learn your Advances range from a leap forward in based on a neural net that learned how preferred route. Cars can automatically computer power to the development of to drive by “observing” human driving respond to the environment through machine learning, in which big data can behavior3. mechanisms such as adaptive cruise be used to train neural nets and deep- control, where automobiles travel at learning, cognitive algorithms. These In fact, AI in automotive is just one the required speed but slow down if the offer huge potential for automation, part of a technology revolution with vehicle in front gets too close; some will prediction, and generation of insights the potential to transform the world in even overtake to maintain the desired from patterns in the data that humans which we live, changing the nature of speed. fail to see. our societies. As Gerlind Wisskirchen, vice-chair for multinationals at the As one might expect, there is fear relating to AI, from both 3 International Bar Association Global Employment Institute has explained, safety and work perspectives. “Jobs at all levels in society presently undertaken by humans are at risk of being reassigned to robots or AI, and STRATEGIES #connectedcar the legislation once in place to ‘protect the rights of human workers’ may be no longer fit for purpose”. Introduction It may even be that legislation is Automotive AI A second application of AI is the Automotive AI in practice required to ensure human beings get creation of new insights from big data: their “fair share of jobs”, alongside the in practice automotive manufactures source huge robots. Automation and AI can reduce The automotive sector is already making volumes of data from vehicles on the Barriers to AI adoption costs and improve productivity, boosting extensive use of AI technologies, with 7 road that it is difficult for humans the competitiveness of industries and million systems installed in vehicles by to comprehend. The connected car Ambition with reality economies, but there will be downsides 2015 (according to IHS Technology). As can provide a constant stream of to manage, including the prospect of communication networks improve, this information about how and where it is Conclusion widespread unemployment. number will increase rapidly, as extra being driven, and how it is performing. bandwidth becomes available from 4G Machine learning provides a powerful Academic perspective Moreover, amid the excitement, tool in the ongoing struggle to see the pragmatism is required. and 5G technologies as well as advances in Software Over The Air (SOTA). patterns within the data and to make Key takeaways As one might expect, there is fear sense of it. relating to AI, from both safety and work Many of these systems are dedicated to delivering infotainment and driver Manufacturers such as BMW have About the author perspectives. The reality of AI capability already begun using this data in new today, however, is nowhere near that interaction services, from smarter traffic and mapping services to voice and ways. One early application is predictive Notes and bibliography described in sci-fi novels. The potential gesture recognition, built on natural maintenance – the ability to anticipate for human-machine partnerships far faults at very early stages so they can outweighs the likelihood that people’s language processing capabilities and pre-trained neural-nets. GM, be corrected before breakdowns occur, jobs will be taken by machines. In the reducing cost and inconvenience. automotive sector, the challenge is to for example, is working with IBM to find AI applications that work, both incorporate its Watson AI technology, Increasingly, manufacturers are also technically and commercially. It’s easy made famous in 2011 by beating two incorporating data and analytics tools to get carried away by the potential former winners of the US TV show into their production technologies, for AI in connected cars, but imagining Jeopardy in a head-to-head version of using the insights gleaned from in-car the future is the stuff of science fiction the quiz (and winning $1m as a reward). data to power next-generation design – in realizing this vision, automotive IBM Watson will be built into GM’s in-car and engineering. For example, the Los manufacturers face engineering “cognitive mobility platforms”; these will Angeles-based start-up Hack Rod aims challenges. essentially be digital assistants capable to create the first ever car engineered of following drivers’ instructions and with AI that has been designed in a anticipating their needs. virtual environment. Then there is the essential role of Increasingly, manufacturers are also incorporating data and 4 AI in powering the advanced driver- assistance systems (ADAS) that will analytics tools into their production technologies, using the eventually turn fully autonomous insights gleaned from in-car data to power next-generation vehicles into a mainstream reality. ADAS design and engineering. incorporates tools such as camera-based STRATEGIES #connectedcar machine vision systems and radar- based detection units, but is ultimately powered by intelligent software capable Introduction of making instantaneous decisions on the basis of the complex information set Automotive AI in practice with which it is presented. Figure 1: Race for AI – major acquirers in artificial intelligence How M&A of AI start-ups has increased since January 2012 Opinions are split on when fully Barriers to AI adoption autonomous vehicles will be routinely Figure 1: Race for AI – major acquirers in artificial intelligence How M&A of AI start-ups has increased since January 2012 using public roads, but manufacturers 2016 Ambition with reality including Tesla, BMW, Mercedes, Nissan, Madbits TellApart Whetlab Magic Pony Ford and Volvo have all developed self- 2014 (Twitter) 2015 (Twitter) (Twitter) (Twitter)

Conclusion driving cars, while entrants from other Causata Novauris Nervana Nexidia Vocal IQ Perceptio Turi Emotient Tuplejump (NICE Technolgies Systems (NICE 2013 (Apple) (Apple) (Apple) (Apple) (Apple) sectors - including Google and Uber - (Systems) (Apple) (Intel) Systems)

Medio have proved that the technology works IQ Engines SkyPhrase LookFlow Desti Orbeus Angel.ai Expertmaker SalesPredict Academic perspective Systems (Yahoo) (Yahoo) (Yahoo) (Nokia) (Amazon) (Amazon) (eBay) (eBay) in a series of public trials. Mobileye, (Nokia) DeepMind Granata DNNresearch Emu Jetpac Timeful Moodstocks api.ai Technologies Decision acquired by Intel for $15bn in March (Google) (Google) (Google) Systems (Google) (Google) (Google) Key takeaways (Google) 2017, claims its partnership with BMW (Google) Mobile Dark Blue Vision Face.com Wit.ai Masquerade Zurich Eye Bit Stew Wise.io 2012 Technologies Labs Factory will begin mass production of fully (Facebook) (Facebook) Tech. (Facebook) Systems (GE) About the author (Facebook) (DeepMind) (DeepMind) (Facebook) (GE) autonomous vehicles in 2021. OCULUSai Sociocas Saffron Indisys Convertro Gravity Encore Alert Barricade.io SAIPS Movidius Itseez (Meltwater Networks Technology (Intel) (AOL) (AOL) (Meltwater (Sophos) (Ford) (Intel) (Intel) Group) (AOL) (Intel) Group) Notes and bibliography The race is on, in other words. The Netbreeze Cognea AlchemyAPI Explorys Equivio SwiftKey Genee Crosswise Palerra engineering challenges now are to make (Microsoft) (IBM) (IBM) (IBM) (Microsoft) (Microsoft) (Microsoft) (Oracle) (Oracle)

the smartest, most convenient, adaptive, Tempo AI PredictionIO MetaMind Geometric Otto (Salesforce) (Salesforce) (Salesforce) Intelligence (Uber) intelligent self-driving electric vehicle (Uber) with fully personalized infotainment, Source: BearingPoint Institute, CBInsights and to get as much meaningful data from such vehicles to inform a cycle of continuous improvement. Barriers to AI adoption There are many types of AI and each is suited to different kinds 5 For all the potential of AI in automotive, of problems with different forms of data and decision-making substantial challenges remain and requirements. disappointments are inevitable. Indeed, the latest Gartner Hype STRATEGIES #connectedcar Cycle for Emerging Technologies puts autonomous vehicles right at the top of the curve, squarely in the “Peak of Introduction Inflated Expectations” zone where early success stores are accompanied by Fully autonomous driving requires a Then there is the question of safety. many failures. complex skillset. The system must have Autonomous vehicle manufactures will Automotive AI in practice powerful perception, to understand be tested on every journey as to their The reality is that automotive autonomy exactly what is happening in the real- claims to be producing safe vehicles, Barriers to AI adoption is at a relatively early stage, with human time environment, but also the ability to with painful reputational and adoption drivers still required to monitor journeys model intention – to anticipate how that consequences following setbacks. Ambition with reality and take control of vehicles in certain environment is likely to change when, for Delivering perfect safety records will circumstances. Uber, for example, example, other vehicles change position. be impossible, particularly given the Conclusion admits its cars struggle to navigate presence of non-autonomous vehicles bridges, where an absence of buildings Moreover, the judgements made on the roads. Agreement is also by humans are often very difficult Academic perspective makes it hard for the vehicles to find necessary on difficult ethical questions: reference points. Also, Ford and Tesla to explain; building algorithms to where some form of crash is inevitable, replicate the processes by which certain Key takeaways point to extreme weather problems such which moral choice should be made – as snow that settles on lane markings, judgements are arrived at is therefore for example, should the car swerve to covering them from view. very difficult. It may be that neural nets, avoid another vehicle if doing so will About the author say, will also deliver answers that do not mean hitting a pedestrian? This creates further problems. The appear logical. That will require a high Notes and bibliography Toyota Research Institute has pointed to degree of trust among users. There are many types of AI and each the challenge of giving drivers enough is suited to different kinds of problems warning to take control of their vehicles A related problem is that connected cars with different forms of data and when they need to intervene. generate increasingly vast data volumes, decision-making requirements. We for which automotive manufacturers will are still learning how to apply these One issue is that the business of need new capabilities to collect, store, technologies as each type of AI has its replicating human judgement is a organize and analyze. Many may find own limitations. Pattern recognition is developing science. AI developers have their existing data management abilities only as good as the example patterns made huge progress with ideas (such as are not fit for purpose. In these early with which computers have been neural nets) and they are experimenting days of autonomous driving, collecting trained. Inference engines work only as increasingly with deep-learning and analyzing these large volumes of well as the rules and the variables set techniques and concepts such as natural data will be essential for improving for them. Genetically evolved designs selection, where the right solution is AI training, as well as improving the depend on the chosen criteria for arrived at through complex trial-and- competency of AI engineers. natural selection and the granularity of error processes. the learning algorithm. There are many ways to combine Building an intelligent machine that It’s imperative for 6 different kinds of AI to compensate can transport people along the obstacle manufacturers to think in for their weaknesses and to make courses of our highways requires best use of their strengths. The use of engineers who understand the problems terms of the problems to be different layers of processing is also involved and how the different forms solved– and which tools are important. More manually crafted layers of AI can be combined effectively, to

STRATEGIES best placed to help – rather #connectedcar of “intelligence” have been combined: interpret the wealth of problem-data than to start with the for example, image-recognition being spewed out every millisecond algorithms often demonstrate improved by the myriad of sensors on the car to technology. Introduction performance when images are make driving decisions. prefiltered to accentuate key features, Automotive AI in practice reduce noise, or remove irrelevant parts of the data (sometimes a black-and- Barriers to AI adoption white image is sufficient, as color merely distracts). Ambition with reality

Conclusion Figure 2: From infotainment to traffic and mapping services, the automotive sector is already making extensive use of AI technologies Figure 2: From infotainment to traffic and mapping services, the automotive sector is already making extensive use of AI technologies Academic perspective

Streaming Software Over Key takeaways media The Air (SOTA) • Determining mood from driving style • Updates to AI algorithms • Selecting music depending on mood • Retrained neural nets About the author • Selecting music depending on who is driving

Notes and bibliography • Driver intent prediction Other cars Your car Roads, • Optimizing route based on • Crowdsourcing traffic data from traffic lights, and where traffic lights are, how cars, phones in cars, third-party other congestion often they are green, whether traffic services they are green now • Route-planning based on • Predict road black spots from • On-phone media driver braking behaviors congestion data, map layers, • In-car media learning from repeat journeys, • Select optimal speed for corner • Navigation crowdsourcing learning gradient and road camber Manufacture Car parking

• Genetic algorithm vehicle design • Likelihood of finding a parking space • Remote data collection, usage data, • Distance to end point diagnostics and prognostics inform • How long people tend to stay here design choices for future vehicles

Source: BearingPoint Institute Ambition with reality Successfully exploiting AI in automotive will depend on 7 How will manufacturers move forward manufacturers’ ability to incorporate key concepts from with AI? It’s imperative to think in multiple disciplines. terms of the problems to be solved – and which tools are best placed to STRATEGIES #connectedcar help – rather than to start with the technology. In other words, think of AI and the connected car as just another Introduction engineering challenge. In practice, that means beginning with automotive sector, with manufacturers It may also be necessary to develop Automotive AI in practice an AI opportunity assessment – an buying niche providers to acquire new ways of working and even new audit of the business’s most pressing specialist knowledge. Ford’s recent $1bn business models. Concepts such as “fail Barriers to AI adoption challenges and opportunities, combined investment in the AI business Argo is fast” and “agile” can help manufacturers with a scoping exercise to identify one example. deploy new technologies rapidly, testing Ambition with reality where AI technologies will deliver the and iterating them in the marketplace, greatest leap forward in understanding Successfully exploiting AI in automotive and then discarding the less successful Conclusion and value. In some cases, the answer will depend on manufacturers’ ability ideas. The automotive sector is used to may derive from customer-facing use to incorporate key concepts from working on extended production and multiple disciplines – demand pressure Academic perspective cases, but it will also be important not development cycles, but for AI and to neglect the potential for AI to deliver in automotive, but also emerging trends related technologies, these will need to in the application of AI, including Key takeaways more internal benefits, such as process accelerate. efficiency, product design optimization, chat bots, intelligent agents, voice and cost control. recognition and adaptive behavior, Nor should automotive companies About the author as well as big data and analytics ignore the potential for new Automotive manufacturers also need capabilities. It’s important to learn collaborations, including with Notes and bibliography to consider whether they possess the from industry lessons. For example, academic partners. Such links are now skills and resources to exploit these Microsoft’s Tay Twitter chatbot aimed strengthening, with new ideas from opportunities. In some areas, services to showcase advances in AI and natural research scientists finding commercial have become commoditized – voice language processing capabilities, but it applications. Toyota, for example, recognition is a good example – and very quickly learned from other Twitter already has a research partnership with the simplest solution may be to buy users to disparage women and ethnic Stanford University and Massachusetts them in from third-party suppliers minorities. Microsoft was forced to Institute of Technology (MIT) that is (or buy the OEMs outright). In other decommission Tay, making it one of the working on systems to assist vehicles cases, closing the expertise gap may year’s biggest machine learning and AI in interpreting real-time challenges in be more difficult; this has been one busts.4 urban environments. factor influencing M&A activity in the 8 Figure 3: Successful exploitation of AI in Automotive depends on key concepts from multiple disciplines

Figure 3: Successful exploitation of AI in Automotive depends on key concepts from multiple disciplines

Embedded Autonomous Sensor technology vehicles Internet of STRATEGIES #connectedcar calibration Things (IoT)

Cognitive computing Intent Voice Chatbot prediction recognitionAutomotive Introduction Automation Engineering Safety discipline case Deep learning Artificial Automotive AI in practice Automotive Intelligence (AI) Infotainment Evolutionary Barriers to AI adoption algorithms Systems thinking Collective intelligence Ensemble Affective Ambition with reality learning computing Hacking Conclusion the future Fail fast Pattern Machine Innovation Data Science recognition learning Academic perspective Big data Formal Key takeaways methods

About the author Disruption: Tech & Disruptive Actionable business model design insights

Notes and bibliography Analytics and Anomaly visualization detection Source: BearingPoint Institute Conclusion For the automotive sector, the challenge is to understand what 9 While it’s easy to slip into the realms of the connected car could achieve for the customer base – and science fiction, AI actually represents how to secure those benefits. just another business challenge and opportunity. Other industries recognize STRATEGIES #connectedcar exactly this point. For example, in financial services, the hedge fund Bridgewater Associates is working on an Introduction AI project that will automate decision- Figure 4: Key AI terminology making and strip out human emotion Automotive AI in practice from its management processes. Figure 4: Key AI terminology Reinforcement learning The automotive supply chain is already Barriers to AI adoption being disrupted, as technology firms get Intent

ever larger slices of the automotive pie. Learning preferences Supervised learning prediction

Gradient boost machine Heuristic evaluation Ambition with reality Back propagation Voice recognition

Frost & Sullivan points out: “Technology Genetic algorithms Machine learning set Training Neural net Neural GPU companies are expecting to be a new tier tuning Hyper-parameter Forward chaining Game theory Knowledge engineering Conclusion 1 for OEMs… 13 OEMs will be investing Morphology Intelligent Kernel trick analysis Convolutional neural networks networks neural Convolutional Sentiment over $7.0 billion in the development of Turing test Support vector machine agents Object identification Backward chaining various AI use cases. Hyundai, Toyota, Digital signal AI Academic perspective processing and GM will account for 53.4% of the Ensemble Deep learning Adaptive behaviour

5 Statistical learning theory Feedback Image Reward Tensor Tensor total investment share.” Feature extraction Neural coding Neural Key takeaways Chat bots recognition

Big data Natural language processing For the automotive sector, the challenge Finite state machine Distance Siamese network Logistic regression About the author is to understand what the connected car

Automation measure could achieve for the customer base – Gesture Robotic process automation Semantic hashing Semantic network Memory

Notes and bibliography and how to secure those benefits. The recognition Source: BearingPoint Institute, CBInsights key to meeting that challenge will lie in engineering, innovation, and product and service development. There is urgency if OEMs don’t want to be left behind, but there is also a place for traditional engineering discipline, and investment portfolio and risk management. 10 ACADEMIC PERSPECTIVE

Professor Mark Skilton, Professor of Practice, Information Systems Management & Innovation, Warwick University STRATEGIES #connectedcar Mark Burnett: Mark, can you give us your perspective on the impact AI is having and will be having in the future? Mark states: “We are in a race towards a new artificial intelligent automotive ecosystem Introduction that will impact how transport will work across all industries and countries. AI is the new competitive advantage that is seeing existing brands and new ones rapidly redefine their business model and Automotive AI in practice vision from a product to a service enabler as the mobile platform of the future. The automobile will be the proxy for everything we do on the move, defining how vehicles are designed and assembled in intelligence factories and Barriers to AI adoption connected supply chains, to redefining the role of travel in the smart city, connected home, automated retail and personal lifestyles. Vehicles will change human employment through self-driving capabilities within the next decade. Ambition with reality It will change car ownership combined with a massive change to clean energy and new possibilities of onboard connected living and work experience. All these automotive innovations will reinvent the car and automotive companies place in society as we move into the fourth industrial revolution.” Conclusion Mark Skilton has 30+ years’ experience as a professional consultant with a track record in top 1000 Companies Academic perspective in over 20 countries and across multiple public, private and start-up sectors. He has direct Industrial experience of commercial practice leadership, boardroom and investor strategy to program team and transformation Key takeaways management at scale. Mark is a recognized International thought leader in digital, company strategy, telecoms, digital markets and M&A strategies, CxO practices and is the author of many books and international papers, About the author including Building the Digital Enterprise Building the Digital Enterprise and Building Digital Ecosystem Architectures. His work and views have been published in the Financial Times, New York Times, Wall Street Journal, Washington Post, Bloomberg, Associated Press, Mail, New Scientist, Nature, Scientific American and broadcast via many television Notes and bibliography and radio channels around the world, including BBC, Sky, ITV, Al Jazeera. 11 KEY TAKEAWAYS • Get to grips with the different kinds of AI to discover which problem domains are best suited to each type • Understand your data and what you want to get out of it: how AI can help, and the additional infrastructure, aggregation, data-cleansing or pre-processing you require to make it usable by AI STRATEGIES #connectedcar • Identify specialist expertise you will need and decide whether you are going to build your own AI competency, outsource this to a specialist firm, or buy/partner with a third party • Look at how your competitors are using AI and explore what it could do to differentiate your business Introduction • Find the right problems to solve and the right types of AI to solve them • Identify existing solutions or roles that could be replaced or augmented by AI Automotive AI in practice • Find opportunities to use AI internally and in customer-facing products and services Barriers to AI adoption • Ensure someone in your organization has responsibility for exploiting AI in your business • Understand how AI fits into the jigsaw puzzle of the broader connected-car digital ecosystem1 Ambition with reality • Set expectations: be prepared to fail - and to succeed - in unexpected ways

Conclusion

Academic perspective How BearingPoint can help you Key takeaways BearingPoint can help in a variety of different ways: we are working with various kinds of AI in different business areas and as part of Digital Ecosystem Management, we are generating business insight from big data, and we are applying AI and Robotic Process About the author Automation (RPA) to various real-time system, process and decision-support problems. • Artificial Intelligence or Robotic Process Automation Notes and bibliography - AI and RPA Opportunity Assessment - AI and RPA tooling selection • Due diligence on AI investments and acquisitions • Decision support, including use of IBM Watson • Generating insight from data with our HyperCube solution, which helps you use artificial intelligence to navigate an ocean of complex, dynamic data and gain deeper understanding of your key risks and opportunities. • Digital Ecosystem Management (DEM) with BearingPoint’s Infonova R6 • Ninja IT to quickly deploy new classes of software like BearingPoint’s R6, HyperCube and IBM’s Watson About the author 12

Mark Burnett, Head of Innovation and R&D, BearingPoint, London STRATEGIES #connectedcar Mark Burnett has over 20 years’ experience in technology and innovation. He has published articles on technology trends, innovation and problem-solving Introduction techniques. His background is in solution design, enterprise architecture, strategy and innovation. Mark Automotive AI in practice champions the use of artificial intelligence and machine learning to augment our understanding of the world Barriers to AI adoption from big data and drive the next revolution towards more integrated human-machine partnerships. Ambition with reality [email protected] Conclusion

Academic perspective Acknowledgements The author would like to thank Professor Mark Skilton for Key takeaways his valuable input, Tanja Schwarz and Sharon Springell from the BearingPoint Institute; Michael Agar from Agar Design; Christopher Norris from CopyGhosting; and About the author AngéliqueTourneux from BearingPoint. Notes and bibliography Notes and Bibliography 13 1. Mark Burnett, ‘How can vehicle manufacturers fit into the new connected car ecosystem’, BearingPoint, London, PDF, 2017 http://bit.ly/ CCecosys 2. James McCormick, ‘Predictions 2017: Artificial Intelligence will Drive the Insights Revolution’, Forrester Research, Cambridge, MA, PDF, November 2, 2016 http://bit.ly/P17AIdrive

STRATEGIES 3. Will Knight, ‘The Dark Secret at the Heart of AI’, MIT Technology Review, Cambridge, MA, April 11, 2017 http://bit.ly/TR_Nvidia - NVidia’s #connectedcar experimental autonomous vehicle learnt to drive by observing human driving behavior 4. Ed Burns, ‘Not All Machine Learning and AI Projects Panned out in 2016’, SearchBusinessAnalytics, December 23, 2016 http://bit.ly/SBA_ AIpan16 Introduction 5. Frost & Sullivan press release, Self-Learning AI Poised to Disrupt Automotive Industry Giving Rise to New Business Opportunities for OEMs, PR Newswire, London, December 15, 2016, http://bit.ly/PRN_AIauto - Technology firms expecting to be tier 1 for OEMs, AI high on the list Automotive AI in practice

Barriers to AI adoption

Ambition with reality

Conclusion

Academic perspective

Key takeaways

About the author

Notes and bibliography About the BearingPoint Institute 14 At the BearingPoint Institute, our ambition goes beyond traditional CONNECT ‘thought leadership’. We aim to contribute original ideas to the science of business management whilst equipping decision makers with practical advice gained in the field and through our research projects. Follow us on Twitter at @institute_be STRATEGIES #connectedcar#Digital www.bearingpointinstitute.com Join us on LinkedIn at www.inst.be/linkedin www.bearingpointinstitute.com

Introduction Send us your comments, thoughts and feedback: About BearingPoint [email protected] Automotive AI in practice BearingPoint is an independent management and technology consultancy with European roots and a global reach. The company operates in three www.inst.be/feedback Barriers to AI adoption units: Consulting, Solutions and Ventures. Consulting covers the advisory business; Solutions provides the tools for successful digital transformation, regulatory technology and advanced analytics; Ventures drives the Ambition with reality financing and development of start-ups. BearingPoint’s clients include DOWNLOAD many of the world’s leading companies and organizations. The firm has Conclusion a global consulting network with more than 10,000 people and supports Read the BearingPoint Institute on your tablet clients in over 75 countries, engaging with them to achieve measurable and with the mobile app: Professor Mark Skilton sustainable success. iOS version from the iTunes AppStore Key takeaways www.bearingpoint.com Android app on Google Play

About the author Amazon Kindle (Fire only) store

(to guarantee a good reading experience, these apps are for Notes and bibliography tablets only)

© 2017 BearingPoint Holding BV. All rights reserved. Printed in the EU. The content of this document is subject to copyright. Changes, cuts, enlargements and amendments, any publication, translation or commercial use for the purpose of trainings by third parties requires the prior written consent of BearingPoint Holding BV. Any copying for personal use is allowed and only under the condition that this copy right annotation will be mentioned on the copied document as well. All information supplied in this document is correct, to the best of our knowledge, at the time of publication.