Endorsements “Much has been written about the mechanics of modern AI – what it is, what it isn’t, and how to build and train one – but this fascinating book is among the few that explain what it means for the enterprise – where it fits in to business needs, how it can be integrated into existing systems, and how it can create opportunities for new value.” —Grady Booch, IBM Fellow, Chief Scientist for Software Engineering “I recommend Beyond Algorithms: Delivering AI for Business to anyone that is responsible for leading and transitioning AI to achieve and maintain a competitive advantage. The book addresses many of the important issues faced in the application of AI to businesses. It is written in an easy-to-read style, while providing enough technical content that many people working in the field can capture the main messages.” —David R. Martinez, MIT Instructor and Laboratory Fellow “Books about AI too often tend towards the extremes – either praising or condemning unconditionally. Beyond Algorithms offers a more nuanced, nontechnical introduction from experienced practitioners. Filled with guiding principles and real-world examples, this book offers a balanced view of AI’s strengths and limitations and provides practical advice for problem selection, project management, and expectation setting for AI initiatives.” —Emily Riederer, Senior Analytics Manager at Capital One “Beyond Algorithms is one of those wonderful story-driven IT books that distils decades worth of real-world experience into intoxicating wisdom that is refreshingly easy to consume. The book plows through and around the hype of the ‘all powerful’ deep neural nets to provide an engineering approach for the field. It won’t tell you how to establish the Singularity, but its ‘A to I checklist’, accuracy and monitoring advice, and ‘doability method’ could make you an AI delivery hero.” —Richard Hopkins, FREng FIET CEng, Former President of IBM Academy of Technology “Important and timely. A practical guide for realizing business value through AI capabilities! Rather than focusing on hard to access details of the mathematics and algorithms behind modern AI, this book provides a roadmap for a diversity of stakeholders to realize AI business solutions that are reliable, responsible, and sustainable.” —Matthew Gaston, AI Engineering Evangelist Beyond Algorithms Beyond Algorithms Delivering AI for Business James Luke David Porter Padmanabhan Santhanam First Edition published 2022 by CRC Press 6000 Broken Sound Parkway NW, Suite 300, Boca Raton, FL 33487-2742 and by CRC Press 4 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN CRC Press is an imprint of Taylor & Francis Group, LLC © 2022 James Steven Luke, David Porter, Padmanabhan Santhanam. Reasonable efforts have been made to publish reliable data and information, but the author and publisher cannot assume responsibility for the validity of all materials or the consequences of their use. The authors and publishers have attempted to trace the copyright holders of all material reproduced in this publication and apologize to copyright holders if permission to publish in this form has not been obtained. If any copyright material has not been acknowledged please write and let us know so we may rectify in any future reprint. Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying, microfilming, and recording, or in any information storage or retrieval system, without written permission from the publishers. For permission to photocopy or use material electronically from this work, access www.copyright.com or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. For works that are not available on CCC please contact [email protected] Trademark notice: Product or corporate names may be trademarks or registered trademarks and are used only for identification and explanation without intent to infringe. Library of Congress Cataloging‑in‑Publication Data Names: Luke, James, author. | Porter, David (Data architect), author. | Santhanam, Padmanabhan, author. Title: Beyond algorithms : delivering AI for business / James Luke, David Porter, Padmanabhan Santhanam. Description: First edition. | Boca Raton : CRC Press, 2022. | Includes bibliographical references and index. Identifiers: LCCN 2021056146 | ISBN 9780367622411 (hardback) | ISBN 9780367613266 (paperback) | ISBN 9781003108498 (ebook) Subjects: LCSH: Business—Data processing. | Artificial intelligence—Industrial applications. | Electronic data processing—Management. Classification: LCC HF5548.2 .L843 2022 | DDC 658/.05—dc23/eng/20220110 LC record available at https://lccn.loc.gov/2021056146 ISBN: 978-0-367-62241-1 (hbk) ISBN: 978-0-367-61326-6 (pbk) ISBN: 978-1-003-10849-8 (ebk) DOI: 10.1201/9781003108498 Typeset in Minion Pro by codeMantra Contents Authors xi Acknowledgements xiii PROLOGUE 1 Chapter 1 Why This Book? 3 ◾ AI IS EVERYWHERE 3 ENTERPRISE APPLICATIONS 4 AI WINTERS 5 WHAT IS DIFFERENT NOW? 7 PROCEED WITH CAUTION! 8 DELIVERING AI SOLUTIONS 8 BETTER UNDERSTANDING OF AI IS CRITICAL FOR SOCIETY 8 TARGET AUDIENCE FOR THE BOOK 9 AN OUTLINE OF THE BOOK 9 REFERENCES 10 Chapter 2 Building Applications 11 ◾ WHAT’S DIFFERENT ABOUT AI WHEN BUILDING AN APPLICATION? 11 PROMINENT AI APPLICATIONS OF THE LAST SEVEN DECADES 15 AI OR NO AI? 23 THE PRESENT – THE DOMINANCE OF THE WEB 24 THE FUTURE – THE ENTERPRISE STRIKES BACK 27 EXAMPLES OF REAL ENTERPRISE APPLICATIONS 31 WHERE DO YOU INTRODUCE AI? 38 ACTIVITIES IN CREATING AN AI APPLICATION 39 COMPLEXITY OF AI APPLICATIONS 42 ARCHITECTURAL AND ENGINEERING CONSIDERATIONS 44 v vi ◾ Contents THREE STAGES OF AN ENTERPRISE AI APPLICATION 46 ENABLING ENTERPRISE SOLUTIONS AT SCALE 48 IN SUMMARY – ARE YOU READY TO START BUILDING APPLICATIONS? 51 REFERENCES 52 Chapter 3 It’s Not Just the Algorithms, Really! 55 ◾ INTRODUCING ALGORITHMS 56 ALGORITHMS IN AI 57 ALGORITHM ADDICTION 59 APPLICATIONS VERSUS THE UNDERLYING TECHNOLOGY 60 ALGORITHMS AND MODELS 60 OBJECT DROPPING PROBLEM 63 UNDERSTANDING THE OBJECT DROPPING DATA 65 FOUR MODELS TO PREDICT OBJECT BREAKAGE 67 COMPARING THE TWO ML APPROACHES 75 COMPARING PHYSICS MODEL WITH ML 78 WHAT ARE THE ML ALGORITHMS ACTUALLY LEARNING? 79 FEATURE DEFINITION AND EXTRACTION 81 REVENGE OF THE ARTIFICIAL NEURAL NETWORKS 83 HUMAN INTERPRETATION OF ARTIFICIAL NEURAL NETWORKS 84 SO WHICH ALGORITHM IS BEST? 85 TRANSFER LEARNING 86 REINFORCEMENT LEARNING 88 BRAIN VERSUS ARTIFICIAL NEURAL NETWORKS 88 FUNDAMENTAL PRINCIPLES AND FUNDAMENTAL MISTAKES 90 SO … IT REALLY ISN’T ALL ABOUT THE ALGORITHM 91 IN SUMMARY – THERE REALLY IS SO MUCH MORE TO AI THAN THE ALGORITHMS 92 REFERENCES 93 Chapter 4 Know Where to Start – Select the Right Project 95 ◾ THE DOABILITY METHOD 96 INNOVATION AND EMERGING TECHNOLOGIES 96 A PORTFOLIO-BASED APPROACH 97 DOABILITY METHOD STEP 1 – TO AI OR NOT AI 97 THREE RECOMMENDATIONS FROM DOABILITY METHOD STEP 1 101 Contents ◾ vii DOABILITY METHOD STEP 1 – WORKED EXAMPLES 102 DOABILITY METHOD STEP 2 – PRIORITISING AI PROJECTS IN THE PORTFOLIO 106 IN SUMMARY – SUCCESS OR FAILURE WILL DEPEND ON SELECTING THE RIGHT PROJECT 107 REFERENCES 108 Chapter 5 Business Value and Impact 109 ◾ WHAT IS DIFFERENT ABOUT AI APPLICATIONS? 110 BUILDING BUSINESS CASES 110 STAKEHOLDERS 112 MEASURABILITY AND UNDERSTANDABILITY 117 IMPORTANCE OF ETHICS IN AI DEVELOPMENT 119 DELIVERING TRUSTWORTHY AI 122 FAIRNESS AND BIAS 123 EXPLAINABILITY 128 TRANSPARENCY 132 TACKLING THE WEAKNESS OF ML SYSTEMS 133 IN SUMMARY – THERE’S MORE TO VALUE THAN MONETARY RETURN 134 REFERENCES 135 Chapter 6 Ensuring It Works – How Do You Know? 137 ◾ MANAGING QUALITY OF TRADITIONAL SOFTWARE 137 MANAGING QUALITY OF AI APPLICATIONS 139 STATISTICAL ACCURACY 140 COST FUNCTIONS 146 MULTIPLE OUTCOMES 147 QUALITY METRICS FOR NATURAL LANGUAGE UNDERSTANDING 147 WHAT DOES THIS MEAN IN PRACTICE? 150 HOW ACCURATE DOES IT NEED TO BE? 154 WHERE DO YOU ASSESS ACCURACY AND BUSINESS IMPACT? 156 OPERATING WITHIN LIMITS 156 QUALITY ATTRIBUTES OF TRUSTWORTHY AI SYSTEMS 159 IN SUMMARY – IF THE AI ISN’T TRUSTWORTHY, PEOPLE WON’T TRUST IT 160 REFERENCES 161