Artificial Intelligence In Supply Chains
A Work Project, presented as part of the requirements for the Award of a Master Degree inManagement from the NOVA – School of Business and EconomicsArtificial Intelligence in Supply ChainsMartin Zapke, 3806A Field Lab carried out on the Master in Management Program, under the supervision of:Professor José Crespo de Carvalho4th January 2019i
DisclaimerWith this disclaimer, Martin Zapke, ensures that the following work project to obtain the Masterof Science degree in Management is conducted by himself. The mentioned references have beenused solely. The copyright remains with the author and the contents must not be publishedwithout the approval of the author.AcknowledgementFirstly, I would like to express my gratitude to my advisor Professor José Crespo de Carvalhofor giving me the opportunity to contribute to this exciting research project. I am also thankfulfor his fantastic and continuous support throughout the field lab. Moreover, I would like tothank my interview partners, whose opinion and insights has been highly valuable for thecompletion of this work. Lastly, I would like to thank my fellow students and friends for thesharing of knowledge and mutual support throughout the master studies.Lisbon, 4th January 2019Martin Zapkeii
AbstractArtificial Intelligence (AI) is set to drive a new wave of digital disruption that redefinesindustries and propels unprecedented levels of innovation. As one of the most data-richenvironments within companies, supply chains create significant opportunities to harness thebenefits of AI. This study follows a qualitative research approach that aims to explore theimpacts and applications of AI within the supply chain. It was found that AI creates a broadspectrum of use cases that drive efficiency, automation and customer-centricity across allcomponents of the end-to-end supply chain.KEYWORDSArtificial Intelligence, Supply Chain, Supply Chain Managementiii
Table of ContentDisclaimer . iiAcknowledgement . iiAbstract . iiiList of Figures . viList of Tables. viList of Abbreviations & Glossary . vii1. Introduction . 12. Research Methodology . 23. Understanding Artificial Intelligence . 33.1. Defining Artificial Intelligence . 33.2. Branches of Artificial Intelligence . 43.3. Factors accelerating Artificial Intelligence. 63.4. Current Industry Applications of Artificial Intelligence . 73.5. Benefits and Risks of Artificial Intelligence . 103.6. Future of Artificial Intelligence . 114. Artificial Intelligence in the Supply Chain Context . 124.1. Defining Supply Chain Management . 124.2. Applicability of Artificial Intelligence within Supply Chains. 134.3. Current Applications of Artificial Intelligence within Supply Chains . 144.3.1. Back Office Automation . 154.3.2. Predictive AI . 154.3.3. Cognitive Robotics. 174.3.4. Virtual Assistants . 185. Assessing the Impact of Artificial Intelligence on Supply Chains . 195.1. Impact on Planning . 205.2. Impact on Sourcing . 205.3. Impact on Manufacturing . 205.4. Impact on Warehousing . 215.5. Impact on Distribution . 215.6. Impact on Customer Interface . 225.7. Benefits and Risks of Artificial Intelligence within Supply Chains . 225.8. Future Outlook of Artificial Intelligence within Supply Chains . 23iv
6. Conclusion . 24Bibliography . 26Appendices . 30v
List of FiguresFigure 1: Key Branches of Artificial Intelligence . 4Figure 2: Functionality of a deep Neural Network . 5Figure 3: Equity Funding of AI Startups worldwide, 2013-17 . 6Figure 4: Median Estimates of Artificial Intelligence exceeding Human Performance . 12Figure 5: Key Components of the end-to-end Supply Chain . 13Figure 6: AI potential Value Creation in the next 20 years . 13List of TablesTable 1: Artificial Intelligence Adoption and Use Cases by Sector . 7Table 2: Artificial Intelligence Use Cases and Impact by Company . 8Table 3: Artificial Intelligence Use Cases in the end-to-end Supply Chain . 14Table 4: Qualitative Study: Impact Assessment of AI on Supply Chains . 19Table 5: Benefits and Risks of implementing AI within the Supply Chain . 22vi
List of Abbreviations & GlossaryAIArtificial IntelligenceAMRAutonomous Mobile RobotsANNArtificial Neural NetworkASRAutomatic Speech RecognitionAVAutonomous VehicleBOABack Office AutomationDLDeep LearningMLMachine LearningNLPNatural Language ProcessingNLUNatural Language UnderstandingRPARobotic Process AutomationSCSupply ChainSCMSupply Chain Managementvii
1. Introduction“Artificial Intelligence is probably the most important thing humanity has ever worked on. It ismore profound than electricity or fire.” – Sundar Pichai (CEO of Google).This quote emphasizes that we are facing another transformational time period. Analogous tothe agricultural and industrial revolution, the digital revolution is having a profound impact onmany facets of our society (Gesing et al. 2018). At the center of this revolution is ArtificialIntelligence (AI), which has expanded beyond research labs to become omnipresent in oureveryday lives. Already today, AI-driven applications such as speaking and perceiving devices,smart robots or self-driving cars are starting to deliver real-life business and consumer benefits.In the context of the data-driven economy and the technology’s disruptive impact, companiesneed to reevaluate all aspect of their organization. This includes what many consider as thebackbone of every company: the supply chain (SC). According to McKinsey (2018a), it is oneof the business functions, in which AI can create the most value. To harness this enormouspotential, SC managers need to understand AI’s possible use cases, including the benefits andrisks that come along with it. The objective of this work is to facilitate this understanding bytaking a practical approach, in which potential applications of AI within the SC context arepresented and assessed.The first part of this work aims to establish a fundamental understanding of AI by analyzing itsgeneral context. This includes the examination of AI’s definition, key branches and acceleratingfactors. Subsequently, the analysis of use cases across various sectors and companies will helpto illustrate the technology and its potentials. A comparison of AI’s benefits and risks, as wellas a future outlook, will conclude the first part. The second part will build upon these insightsand examine the technology in the context of SCs. The definition of generic SC componentswill serve as the basis for the subsequent analysis of AI use cases within the field. Finally, the1
last part leverages qualitative expert interviews to assess the overall impact, benefits, risks andfuture implications of AI on the respective building blocks of the SC.2. Research MethodologyThis paper contributes to a joint research project of exploring how major digital technologiesare affecting SCs by focusing on the impact that AI has on the field. Essentially, this study isguided by the subsequent research questions:(1) What lies behind the term Artificial Intelligence and where does it apply?(2) What are the impacts and applications of Artificial Intelligence within the supply chain?(3) What are the major benefits and risks of Artificial Intelligence within the supply chain?To develop a comprehensive answer to these questions a qualitative research methodology waschosen for this study. This is because the research questions are exploratory by nature andcannot be answered through quantitative methods (Stebbins 2001). Moreover, a qualitativeapproach should be preferred when examining a subject that does not aim to find one single“truth” but instead seeks to investigate potentials of an uncertain future (Sargeant and Sullivan2011).At first, secondary research is conducted to develop a fundamental understanding of thetechnology’s general and SC context. In the next step, the findings will be complemented byprimary research in the form of five expert interviews. This will help to gain various insiderperspectives on the subject from experts that work at the intersection of AI and SCs (seeappendix A). However, even though this approach is appropriate to gain an in-depthunderstanding of the subject, it is important to mention that the answers are based on a smallscale sample that is not statistically representative. Moreover, as the qualitative data relies onindividual perspectives, it cannot be entirely objective. The expert interviews will follow asemi-structured approach, that uses a predefined set of open-ended questions as a reference butleaves room to extend the conversation beyond those if necessary (Patton 2002). The following2
topics will serve as guidance for the expert interview questions: a) how AI is changing SCsfrom an organizational perspective and the major trends at present; b) the current and futureapplications of AI in SCs; c) areas of SCs where AI will have the biggest and least impact onand where it will be more and less useful; d) the benefits and risks of applying AI in SCs; e) thehurdles of adopting AI in SCs and how to overcome them; f) the impact of AI on the workforce.To conduct the proposed methodology, the study relies on primary and secondary data sources.While the primary data is collected through the expert interviews, the secondary data is basedon published scientific papers such as journals or books as well as reports from majorconsultancy companies.3. Understanding Artificial Intelligence3.1. Defining Artificial IntelligenceAlthough the term AI is deeply embedded in today’s academic and corporate environment, itstill lacks a distinct and generally accepted definition. This is mainly due to the interdisciplinaryand complex nature of the field. The term “Artificial Intelligence” was initially coined by thecomputer scientist John McCarthy who organized the first academic conference on the subjectat Dartmouth College in 1956. The conference is recognized as the birthplace of AI as anacademic discipline, in which McCarthy described it as the study of “machines simulatinghuman intelligence” (McCarthy et al. 1955). For the purpose of this research, we will buildupon this idea and define AI as an area of Computer Science that deals with the developmentof systems, able to carry out cognitive functions, which we typically identify with human minds.This involves fundamental abilities such as learning, understanding natural language,perception or reasoning (McKinsey 2018). The extent to which AI systems perform theseabilities distinguishes Narrow (or “Weak”) from General (or “Strong”) AI. Narrow AI systemsonly use certain aspects of human cognition and focus on a particular problem they have beentrained to solve, as opposed to General AI systems, which are capable of applying the full3
spectrum of cognitive functions (like humans) to solve any task they are confronted with. Sinceall current AI applications are designed around specific problems, and general AI has yet to beaccomplished, the term “AI” will subsequently always refer to the narrow version of thetechnology. Moreover, AI should not be regarded as a single technology but as an umbrellaterm for a variety of technological branches, that are often interrelated and build on top of eachother.3.2. Branches of Artificial IntelligenceIn this chapter, we will present some of the key technology branches of AI, that each applyparticular cognitive abilities such as learning, understanding natural language or perception. Itis important to mention that these branches are not exhaustive but instead have been the focusof AI research and business applications in recent years (Stanford 2016). These includeMachine Learning (ML), Natural Language Processing (NLP), Computer Vision and Robotics(see Figure 1 for the respective definitions).Figure 1: Key Branches of Artificial IntelligenceFigures 1:Figure1: KeyKey Branchesbranches of Artificial IntelligenceFigures 2: Functionality of a deep Neural NetworkFigures 3: Key branches of Artificial IntelligenceSource: own illustration based on Norvig and Russel 2009Some branches focus on processing external information, such as NLP and Computer Vision;some use information to act upon it, such as Robotics; and others use information to learn fromSource: own illustration based on Norvig and Russel 20094
it, such as ML (McKinsey 2017). However, for many AI applications these branches oftenmutually reinforce and complement each other. For instance, ML models are the coretechnology for many advanced NLP, Computer Vision, and Robotics applications. EspeciallyDeep Learning (DL), which is a subfield of ML, has significantly contributed to the advent ofmany AI applications in recent years. DL uses deep Artificial Neural Networks (ANN) toresemble the operation principle of the human brain (Deng and Dong 2014). Similar to howrelations between neurons in the brain adjust and improve through experience, connectionswithin the ANN are strengthened or weakened as new data inputs are received by the network(Gesing et al. 2018). By reinforcing connections that achieve good results and weakening theones that lead to inferior results, the output quality gradually improves with every learningcycle. Figure 2 simplifies the structure and functionality of a deep ANN for two differentproblems types. ANNs consist of connected neurons, arranged in a series of layers. They areFigure 2: Functionality of a deep Neural NetworkFigure 2: Functionality of a deep Neural Networkcapable of processing all forms of inputdata such as pixel, audio or textualdata (Hecker et al. 2017). In a popularcase for computer vision, namely therecognition of faces, the input layersintroduce the pixel data into thenetwork. Hidden layers then breakdown the visual components toidentify the distinctive features of aSource: own illustration based on Gesing 2018certain face. At the output layer, theANN predicts whether the face belongs to Person X, Y or Z. As mentioned above, the predictionaccuracy increases with every learning iteration. A deep ANN with numerous hidden layers canSource: own illustration based CB Insights 2018solve more complex problems as it can identify increasingly subtle features of the input data.Source: own illustration based CB Insights 20185
3.3. Factors accelerating Artificial IntelligenceSince the “birth” of AI in 1956, the field has gone through various cycles of excitement andhype, followed by so-called “AI winters” – periods characterized by declining interest, research,and funding (Norvig and Russel 2009). The two major AI winters in the 1970s and 1990s canbe ascribed to AI not being able to live up to its hype, mainly due to the technological limitationsduring these periods (see appendix B.1.). Current AI activities suggest that we are again in an“AI spring”, with the technology being embedded in our everyday lives. This becomes obviouswhen looking at the increase in equity funding of AI startups in recent years (see Figure 3). Theannual funding volume in 2017 ( 15,25B) was nine times higher than in 2013 ( 1,73B).Figure 3: Equity Funding of AI Startups worldwide, 2013-17Figures 3:Figure4: EquityEquity Fundingfunding ofof AIAI StartupsStartups worldwide,worldwide, 2013-172013-17Simultaneously, there hasbeen a ninefold increase inthe number of annuallyFigures 5: Median Estimates of Artificial Intelligence exceeding HumanPerformanceFigures 6: Equity funding of AI Startups worldwide, 2013-17publishedAIresearchpapers since 1996 (Scopus2017).AccordingtoGesing et al. (2018) andSource: own illustration based on CB Insights 2018McKinsey (2018a),thecurrent acceleration of AI can be ascribed to the convergence of three technologicaldevelopments: increased computational power, access to big data and algorithmicSource: own illustration based CB Insights 2018advancements. As computer processing-intense technology, AI has benefited from theexponential growth in chip efficiency suggested by Moore’s law as well as the use of graphicalSource: own illustration based CB Insights 2018(GPUs) instead of computer processing units (CPUs). GPUs allow for large parallel workloads,significantly reducing the time required to train AI algorithms (Gesing et al. 2018). Secondly,Source:own illustrationbasedCB Insights2018theproliferationof “bigdata”regardingvolume, velocity, and variety is a crucial part of AI’ssuccess. Traditional computing techniques were not able to process such large and oftenSource: own illustration based CB Insights 20186Source: own illustration based CB Insight
Artificial Intelligence in Supply Chains Martin Zapke, 3806 A Field Lab carried out on the Master in Management Program, under the supervision of: Professor José Crespo de Carvalho 4th January 2019 . ii Disclaimer With this disclaimer, Martin Zapke, ensures that the following work project to obtain the Master of Science degree in Management is conducted by himself. The mentioned references .
Artificial Intelligence -a brief introduction Project Management and Artificial Intelligence -Beyond human imagination! November 2018 7 Artificial Intelligence Applications Artificial Intelligence is the ability of a system to perform tasks through intelligent deduction, when provided with an abstract set of information.
SERIES Integrating the supply chains of vaccines and other health commodities This document examines the benefits, challenges, and rationale for integrating vaccine supply chains with the supply chains of other health commodities. It provides agencies, donors, decision-makers, and partners with a brief overview of supply chain integration
and artificial intelligence expert, joined Ernst & Young as the person in charge of its global innovative artificial intelligence team. In recent years, many countries have been competing to carry out research and application of artificial intelli-gence, and the call for he use of artificial
BCS Foundation Certificate in Artificial Intelligence V1.1 Oct 2020 Syllabus Learning Objectives 1. Ethical and Sustainable Human and Artificial Intelligence (20%) Candidates will be able to: 1.1. Recall the general definition of Human and Artificial Intelligence (AI). 1.1.1. Describe the concept of intelligent agents. 1.1.2. Describe a modern .
IN ARTIFICIAL INTELLIGENCE Stuart Russell and Peter Norvig, Editors FORSYTH & PONCE Computer Vision: A Modern Approach GRAHAM ANSI Common Lisp JURAFSKY & MARTIN Speech and Language Processing, 2nd ed. NEAPOLITAN Learning Bayesian Networks RUSSELL & NORVIG Artificial Intelligence: A Modern Approach, 3rd ed. Artificial Intelligence A Modern Approach Third Edition Stuart J. Russell and Peter .
Peter Norvig Prentice Hall, 2003 This is the book that ties in most closely with the module Artificial Intelligence (2nd ed.) Elaine Rich & Kevin Knight McGraw Hill, 1991 Quite old now, but still a good second book Artificial Intelligence: A New Synthesis Nils Nilsson Morgan Kaufmann, 1998 A good modern book Artificial Intelligence (3rd ed.) Patrick Winston Addison Wesley, 1992 A classic, but .
BCS Essentials Certificate in Artificial Intelligence Syllabus V1.0 BCS 2018 Page 10 of 16 Recommended Reading List Artificial Intelligence and Consciousness Title Artificial Intelligence, A Modern Approach, 3rd Edition Author Stuart Russell and Peter Norvig, Publication Date 2016, ISBN 10 1292153962
PA R T 1 Introduction to Artificial Intelligence 1 Chapter 1 A Brief History of Artificial Intelligence 3 1.1 Introduction 3 1.2 What Is Artificial Intelligence? 4 1.3 Strong Methods and Weak Methods 5 1.4 From Aristotle to Babbage 6 1.5 Alan Turing and the 1950s 7 1.6 The 1960s to the 1990s 9 1.7 Philosophy 10 1.8 Linguistics 11
