www.ifc.org/thoughtleadership NOTE 87 • JUL 2020 AI Investments Allow Emerging Markets to Develop and Expand Sophisticated Manufacturing Capabilities By Sumit Manchanda, Hassan Kaleem, and Sabine Schlorke As advances in machine learning, computer vision, and robotics help manufacturers around the world improve their processes and produce new and more complex products, artificial intelligence (AI) is becoming an integral tool of modern manufacturing, and one that is increasingly important to the industry’s future. By combining large volumes of data with the computing power to simulate human thinking, AI is increasing the efficiency, capacity, and complexity of factory floors, and is introducing robotics, the Internet of Things, and other cutting-edge innovations to manufacturing value chains across the globe. Artificial intelligence is a critical enabler of manufacturing complexity that is essential for companies to produce an expansive range of sophisticated products and dynamically engage with regional and global value chains. Firms and economies can increase both their manufacturing complexity and their market competitiveness by developing the foundational capabilities and know-how needed to adopt AI and other advanced technologies. Companies in both advanced and developing economies when machines and mechanized factories began to replace increase their levels of manufacturing complexity and hand production, and it remains true today as automation, contribute to economic growth and societal advances by cognitive computing, and high-speed data exchanges recast identifying and capitalizing on disruptive technologies. the methods of production and accelerate the development of Investments in artificial intelligence in particular will be new and more sophisticated products. critical to development, as they will allow emerging markets AI accelerates economic complexity and economic growth in to create and expand sophisticated manufacturing sectors, three specific aspects of manufacturing: products, processes, participate in increasingly interconnected value chains, and and value chains. For the purposes of this note, we follow compete in markets where speed and data are essential. the definition and description of basic, advanced, and Historically, there has been a strong relationship between autonomous artificial intelligence that were put forward in economic complexity, technical know-how, and economic EM Compass Note 69.1 That is, that artificial intelligence is growth. All countries that have been able to harness new the science and engineering of making machines intelligent. technologies have consequently been able to expand their In this note, the term AI refers to all computer systems that economies. It was true during the First Industrial Revolution, can continuously scan their environment, learn from it, and About the Authors Sumit Manchanda, Senior Private Sector Specialist, Finance, Competitiveness, and Innovation (FCI), IFC. Hassan Kaleem, Senior Economist, Global Manufacturing, Sector Economics and Development Impact, Economics and Private Sector Development, IFC. Sabine Schlorke, Manager, Global MAS, Manufacturing, Agribusiness and Services, IFC. 1 This publication may be reused for noncommercial purposes if the source is cited as IFC, a member of the World Bank Group. Drivers of production Structure of production Technology and innovation Scale Number of Human capital Complexity components in a product Global trade and investment Product Complexity Institutional framework Sustainable resources Number of tasks Production Number of activities it Demand enforcement in a process Complexity takes to deliver a through which a product to market in a product is made specific industry Process Value Chain Complexity Complexity • Technological changes impact the structure of production in all dimensions (complexity and scale) • Positive changes in product, process, and value chain complexity are important for industrialization FIGURE 1 Manufacturing Contributes to Economic Complexity Source: IFC Analysis. take action in response to what they sense, as well as to 1. Product complexity: AI enables companies to more human-defined objectives. efficiently manufacture sophisticated products such as automobiles, which contain a large number of complex What Is a Complex Manufacturing Economy? parts and components, all of which are separately produced and ultimately assembled into a single unit. Countries with a high degree of economic complexity are able to manufacture an expansive range of sophisticated 2. Process complexity: Today, by combining large products using advanced processes. They also have volumes of data with computing power, manufacturers dynamic relationships with multiple regional and global are using AI to simulate human cognitive abilities such value chains. These economies possess specialized know- as reasoning, language, perception, vision, and spatial how, and are generally where market leading firms and processing. AI is being used for predictive maintenance, conglomerates have cultivated sophisticated relationships assembly line inspections, and other tasks that range with a network of companies in the supply chain. Examples from the mundane to the cutting edge. of high-complexity countries include Germany, Japan, the 3. Value chain complexity: The real-world benefits of AI United States, and the Republic of Korea. were emphasized recently in a World Economic Forum On the distribution side, advanced economies possess survey of corporate executives that was conducted recognizable global brands; strong research and development, during the Covid-19 crisis. 2 The executives said that design and innovation capabilities; client-oriented quality “their past investments in new technologies are paying controls; and a flexible network of outsourcing partners. On off now.” They emphasized, for example, how big data the buying and selling side, competition is based on brand and platforms and the Internet of Things (IoT) enabled quality, with many companies operating at the forefront of the them to quickly gather large quantities of information technology frontier. that helped to predict supply chain disruptions that would impact production. AI helped provide The Three Dimensions of Manufacturing instant visibility into the value chain and enabled Complexity quicker mitigation, which may have allowed some manufacturers to survive. AI accelerates manufacturing complexity in three ways: 2 This publication may be reused for noncommercial purposes if the source is cited as IFC, a member of the World Bank Group. Mounting Momentum for Smart Machines maturity and sophistication in their manufacturing subsectors. As would be expected, AI can be more According to market intelligence firm TrendForce, the demand economically justified in wealthier, complex manufacturing for global technology-based manufacturing applications economies where it is applied in large-scale industrial will increase to more than $320 billion in 2020, from applications. But equally important, there are many about $200 billion in 2019, and will grow at a compound instances around the globe where AI could have a profound annual rate of 12.5 percent.3 In a 2017 report published impact on the manufacturing processes and growth of by Infosys,4 75 percent of a sample of medium and large less complex economies. Already, AI is being used in U.S. companies said they had yet to reach their automation some of those countries for capital-intensive and labor- potential because of complexity and legacy issues. Seventy-six intensive processes such as monitoring and scheduling percent viewed AI as a key factor in transformation where pipeline maintenance and performing heat inspections in artificial intelligence agents could replace human cognitive cement kilns—critical and often dangerous tasks normally tasks. The Covid-19 pandemic has disrupted adoption of AI relegated to specially trained employees. in all regions as companies work to rebuild and countries rethink their industrial strategies in the wake of the crisis. But Pillar 1 Economies: Low-income countries and those in it is reasonable to assume that the global appetite for smart fragile and conflict-affected situations. technologies that can accelerate growth and complexity will Goal: Lay a foundation for industrial production in continue, and companies will weigh their future AI investment countries with a low-complexity manufacturing sector. decisions, in part, on how well AI performed during the crisis. Major manufacturers plan to monitor, record, and analyze Country classification: These countries generally have data across all stages of manufacturing.5 a small industrial base, lack economic diversity, have limited skills and technology intensity, and have low or In September 2019, the International Federation of no manufacturing exports. They are narrowly engaged Robotics predicted that industrial robot shipments would with global value chains, usually in agriculture, textiles, increase 12 percent annually from 2020 to 2022, on ready-made garments, light engineering, electronics average.6 Today, robots are used almost exclusively for assembly, footwear, and leather goods. They are also automation. But robotics is advancing rapidly, integrating typically characterized by low labor costs (this includes cutting-edge technologies that enhance automation and most countries in Sub-Saharan Africa). Low-income functionality. While most demand for commercial robots economies are generally less technologically advanced in has been in advanced economies and higher-income manufacturing and remain dependent on manual labor emerging markets, manufacturers in low-income countries and processes. They often lack the requisite capacity to are beginning to invest as costs decline. develop diversified manufacturing bases to broaden the As Industry 4.0 matures, technology companies will complexity of their economies, or to provide people with continue backward integration into existing manufacturing opportunities to gain the skills that can drive human functions, retrofitting machines and processes to make them development. Advanced technologies, however, can improve smarter, analytical, and increasingly data-oriented. At the these countries’ manufacturing processes, help them same time, established industrial companies will continue produce more sophisticated products, and enable them to innovate and incorporate more complex technologies into to engage with increasingly complex regional and global their production processes. Enterprises will continue to adopt markets and value chains. Advanced data analytics and technologies that are relevant to the stage of development artificial intelligence have enormous potential to propel of the economies in which they operate, and AI—as it manufacturing forward in these economies. becomes less expensive and more commonplace throughout AI adoption and use: AI in Pillar 1 countries is mostly the value chain—will inevitably spread to countries at every limited to digitalization of production data with IoT, stage of complexity and will play an increasingly important including in account payments and inventory management role in the industrialization process. As Figure 2 illustrates, systems. At the consumer level, mobile technology and over the past two centuries, every industrial revolution has financial services companies such as Ant Financial in East been distinguished by a new technology that has driven Asia, M-Shwari in East Africa, M-Kajy in Madagascar, manufacturers to a more complex economic stage. and MoMo Kash in Cote d’Ivoire are harnessing AI applications to better predict customer default probabilities, An AI Solution for Every Level of Complexity increasing their confidence in credit scores and enabling Manufacturing economies can be categorized into three them to expand financial services to unserved, underserved, broad pillars of complexity depending on the level of and unbanked populations, while facilitating industrywide 3 This publication may be reused for noncommercial purposes if the source is cited as IFC, a member of the World Bank Group. First Second Third Fourth Industrial Industrial Industrial Industrial Revolution Revolution Revolution Revolution through the introduction through the introduction through the use through the use of mechanical of a division of labor of electronic of cyber-physical production facilities and mass production and IT systems systems with the help of water with the help of that further Degree of complexity and steam power electrical energy automate production First programmable logic controller First mechanical loom, First assembly line Cincinnati (PLC), Modicon 1784 slaughterhouses, 1870 084, 1969 Time 1800 1900 2000 Today FIGURE 2 Four Stages of Industrial Revolutions Leading to Industry 4.0 Source: Pouliquen, Emmanuel, Hassan Kaleem, and Sabine Schlorke. 2018. “IFC Manufacturing Sector Deep Dive: Unlocking the Value of Manufacturing for Development.” International Finance Corporation, World Bank Group, slide 11 (internal document). financial efficiencies such as digital wage payments. Country classification: These countries have broad and sophisticated industrial bases where technology, education, Pillar 2 Economies: Emerging markets. and skills traverse sectors to drive growth via collective know- Goal: Expand and diversify the manufacturing base in how and resilient industry networks. Pillar 3 economies are countries with a mid-complexity manufacturing sector. characterized by their global competitiveness in multiple value chains and their high levels of technological sophistication Country classification: These countries have an evolving (e.g., Germany, Japan, United States, and China). industrial base that is becoming more diversified and competitive, the technology skills of their workforces are AI adoption and use: AI is adopted faster and has improving, and they have developed some global value more impact opportunities in Pillar 3 countries where chain relationships (e.g., Brazil, Turkey, India, Serbia, applications are used for planning, designing, mocking Thailand, and Greece). up, prototyping, testing, fine-tuning, producing, and post-design product and process improvements. What AI adoption and use: Pillar 2 countries adopt and use distinguishes Pillar 3 countries is the volume and artificial intelligence algorithms more broadly, including for capability of companies that can invent, access, and asset performance management, smart image recognition, utilize cutting-edge technologies to manufacture a range process and quality control, and product engineering, as well of sophisticated products such as autonomous cars, smart as optimization of resources and supply chain management. robotic applications, and passenger jets. Many Pillar 3 These countries are involved in all industrial sectors and manufacturers are capable of optimizing their supply interact with companies that span the range of complexity. In chains, production processes, inventory-management some cases, Pillar 2 overlaps with Pillar 3 and, consequently, systems, and transportation logistics. these economies often adopt advanced AI applications. Pillar 3 Economies: More advanced markets. How AI Can Accelerate Complexity Goal: Support more complex manufacturing in countries Investment in artificial intelligence tooling can be costly and with a higher-complexity manufacturing sector. therefore constitutes an impediment to adoption. But AI is 4 This publication may be reused for noncommercial purposes if the source is cited as IFC, a member of the World Bank Group. being readily adopted for a slew of industrial applications Machine makers have also developed such algorithms for ranging from robotic assembly to high-speed communications manufacturing applications, where, for example, image to air filtration systems for sterile manufacturing. And AI recognition is part of a quality-control process. In the food can accelerate complexity by empowering companies to processing industry, Domino’s Pizza has integrated an manufacture more sophisticated products that are sought image-recognition video control system driven by artificial in more affluent markets. Another benefit is that the intelligence that checks whether pizzas meet the company’s development of AI may be encouraging greater skills building quality standards before they are delivered to customers.12 and education achievement by labor forces eager for more Process and quality improvement in more advanced sophisticated and higher-paying jobs. countries. The Advanced Manufacturing Research Centre’s Industrial robots in emerging markets. Some industrial Factory 2025 demonstrated that computing devices fitted robots can be seen as potential opportunities for AI in computer numerical-control machines could collect applications. This is the case with some robots equipped power consumption data, run it through an AI algorithm, for image recognition-oriented tasks. The vast majority and analyze production variations against the production of industrial robots were shipped to more advanced cycle to achieve efficiency gains and cost savings. The manufacturing subsectors in emerging economies in ramifications for production processes are myriad, ranging Asia between 2015 and 2017. Robots are mostly used from improved product quality, increased savings on in the automotive, electrical, and electronics industries, repairs and warranties, and a reduction in production although applications are deployed in other sectors, mainly downtime, all of which are efficiencies that can bolster in handling.7 There has been very limited adoption of a company’s market share and profits. Taking these industrial robots in lower-income and fragile and conflict- factors into account and weighing the costs and benefits affected countries, where resource-intensive manufacturing is an important exercise for companies contemplating an is the main focus. investment in AI applications. Asset performance management. Data-driven maintenance Resource and supply chain optimization in more advanced decisions are a cost-effective way to predict and prevent countries. The increasing importance of global value chains breakdowns in machinery and production. Effective has driven demand for data-hungry applications that maintenance practices are critical to an efficient can sense, control, monitor, analyze, and independently manufacturing value chain. Oniqua Enterprise Analytics maintain not only machinery but also the processes— estimates that 40 percent of scheduled machinery and from raw material extraction to successful product plant maintenance costs are spent on assets with negligible delivery—that companies rely on. The focus on value chain failure impact.8 Up to 30 percent of maintenance activities optimization with AI is largely due to: (1) the increasingly are carried out too frequently, and up to 45 percent of all important revenue-generating role that services play for maintenance efforts are ineffective, according to T. A. manufacturers; and (2) the increasing dependence on Cook.9 Data-driven AI solutions leverage historical data efficient global value chains.13 Historically, the gap between and correlate manufacturing breakdowns with critical revenues from services and revenues from direct sales process parameters to create rules that allow manufacturers of products has been blurry. An example of a company to operate more reliably and with less downtime. that has successfully bridged that gap is IBM, which has evolved from a “box” manufacturer into a high value- Smart image recognition in advanced countries. Smart added and complex services company. Communication image recognition, or the use of AI in machine vision, and data exchange have been an essential part of this has many potential manufacturing applications, such as transformation. In most industrial environments, detecting product defects by conducting pixel-to-pixel communication within and between industrial sites has comparisons. The global machine vision market is not highly been based almost exclusively on wired networks due to a concentrated, and the key players are American, Japanese, need for reliability. and Chinese companies. But with the advent of AI, other companies are emerging in this space, including Facebook Remote maintenance, network, and enterprise and Alibaba, which have made acquisitions in machine communication in emerging markets. Time-critical process vision firms.10 The California-based company Similarity optimization inside factories of different tiers of suppliers specializes in Automated Image Anomaly Detection, which can reduce inefficiencies, support zero-defect manufacturing, uses vast amounts of satellite imagery data to generate rapid increase worker satisfaction, and improve safety. In the most awareness of critical anomalies on the ground.11 sophisticated cases, this could include: remote maintenance and control that may use connected cameras and possibly 3D 5 This publication may be reused for noncommercial purposes if the source is cited as IFC, a member of the World Bank Group. Our vision is to unlock the value of manufacturing for development to strengthen economic complexity. To achieve this, we will develop a portfolio approach that incorporates… Laying the foundation Supporting more advanced Helping countries enhance and 1 for industrial production 2 manufacturing 3 deepen manufacturing complexity • Building commercially viable resource- • Continuing the build-up of • Strengthening inter-industry linkages based industries for local consumption foundational manufacturing to sustain global competitiveness and higher value added exports capabilities • Supporting leading domestic/regional • Strengthening basic supply chains • Developing products and solutions firms to build a global footprint mainly targeting local and regional focusing on local and regional markets • Enhancing “Servicification,” R&D demand [anchor companies] • Helping countries become more leadership and branding • Supporting production and assembly competitive by entering multiple GVCs of simple products in GVCs Technology: IFC’s approach is to work with clients to take advantage of disruptive technologies to accelerate manufacturing complexity across all pillars FIGURE 3 A Framework for the Manufacturing Sector, from Low-Income Countries to Advanced Economies: The Three Pillars of Manufacturing Complexity Source: IFC. virtual reality applications; connected goods that can create 4G LTE, which is low-cost, reliable, and flexible, has aided new value-added services, including real-time monitoring physical security and cybersecurity protection.15 of fluid levels in engines; seamless intra/inter-enterprise AI-based virtual reality is being applied in creative ways to communication, for example, the widespread use of tracking improve productivity and complexity in manufacturing. At devices such as RFID stickers or connected sensors that can Ericsson’s factory in Tallinn, Estonia, the company uses AI- monitor assets distributed over large areas; and the efficient based augmented reality to help predict and troubleshoot coordination of cross-value chain activities and optimization breakdowns that could interrupt production, idle workers, of logistical flow. and increase costs. Using AI, the company, which has an New cellular network technologies. With the advent of established data and quality culture, can reduce the cost of 5G, seamless real-time data communication between a breakdown by as much as half. Generally speaking, AI a manufacturer and its value chain partners allows for technology provides incrementally increasing benefits for more comprehensive and precise tracking of deliveries and companies that grow their data-based learning methods. usage of products, not to mention quicker identification Therefore, it is critical for manufacturing companies in and response to problems and failures. 5G will impact Pillars 1 and 2 to establish data and quality cultures based manufactured products that need to exchange massive on conventional approaches before embarking on AI-based amounts of data in real-time with the rest of the world. techniques. Over the years, the progressive introduction of 2.5G and An Approach for Each Pillar 3G mobile communication systems on plant floors has helped open more options in mobile Internet for digitalized Pillar 1 Economies: In fragile and conflict-affected communications. But it is the more recent preponderance of situations and low-income countries, it is necessary to remote video surveillance, which requires a massive amount build commercially viable, resource-based industries that of data transmission broadband,14 that augers for more manufacture products for local consumption and for higher robust networks. Machine communication aims for lower value-added exports. Strengthening local supply chains and complexity, less power usage, deeper coverage, and higher building capacity to produce and assemble low-complexity device density. As the volume of data grows with data-heavy products as part of global value chains is key. AI adoption applications, so will the need for higher data transmission is limited, but it is important for new investments in rates for such applications. In oil, gas, and water plants, resource-based industries to involve the best available 6 This publication may be reused for noncommercial purposes if the source is cited as IFC, a member of the World Bank Group. process technologies, including AI. It is also important with self-driving vehicles, a very sophisticated example for these countries to build technical foundations in data of AI-based robotization and driving automation. A usage, capture, and statistical analysis. large automotive parts manufacturer that is an IFC client is developing an AI-based virtual-simulation program The Dangote Group, for example, uses cement-loading in collaboration with a German start-up to accelerate robots. Sophisticated cement companies have kiln control development of the company’s advanced driver assistance systems with rules-based programs that optimize yield, systems and automated driving functions. The simulation reduce thermal and electricity consumption, and improve program creates a realistic traffic environment that enables process quality and reliability.16 This conserves fossil fuels, new driver assistance products to be tested virtually. Up to minimizes CO2 emissions, and encourages a sustainable 8,000 kilometers per hour of testing can be performed with industrialization approach. Chemical companies typically virtual simulation, while only about 10,000 test kilometers incorporate similar technologies in their offerings. Thus, per month can be driven by a real vehicle. a few process-heavy Pillar 1 industries already use some AI applications. But because of the low cost of labor in Another IFC client in the automotive sector is adapting AI- most Pillar 1 countries, the economic calculus for making integrated sensors in air filtration systems in its paint shops an expensive investment in AI is very different than in to predict and analyze dust particles in the air and create an expensive labor market. On the other hand, engineers cleaner and more sterile manufacturing environments for and technicians may cost much less in low-income production of sensitive products. That technology promises countries, significantly reducing the cost of developing and to have multiple applications for manufacturers in a range implementing advanced applications. of industries that require such production environments. At a major original equipment manufacturer, an innovative Pillar 2 & 3 Economies: In recent years, more advanced “dust particle analysis technology” has been deployed as a developing countries have embraced sophisticated pilot project in the automaker’s paint shop. The application manufacturing applications, including some powered by AI can forecast and identify instances when there will be an and machine learning. Mexico, for example, added 6,334 increase in dust particles in the air that can mar a car’s industrial robots in 2017, largely to service its automobile painted finish. It can then fine-tune filter replacement based industry. AI adoption by Pillar 2 countries, particularly on a series of factors such as historical levels of airborne in established applications such as asset performance dust by season, or by monitoring trends in prolonged dry management, smart image recognition, process and quality periods. The algorithm monitors 160 factors related to control, and product engineering, as well as supply chain the application of paint and can make highly accurate management, is encouraging. predictions about the quality of the paint process. For example, IFC is exploring a partnership with a textile This AI solution can also be applied in production facilities manufacturer that uses computer vision and AI to detect of Pillar 2 and 3 countries to series production as the defects on its production line. The technology will allow database expands, capturing more and more sensitive the company to reduce waste and shrink its environmental information and enabling manufacturers to produce footprint. more complex products. Robots and smart factory floor In areas such as resource efficiency, DataProphet,17 a automation are other indicators of industrial complexity. business consulting firm in Cape Town, specializes in AI for According to the International Federation of Robotics, in manufacturing by improving process efficiencies through 2017, robot sales increased by 30 percent to a new peak optimization of process variables.18 It has helped a major for the fifth year in a row. Five major markets—all of them engine-block manufacturer attain zero-percent external among the world’s most diverse and complex economies— scrap, and helped an international car manufacturer reduce represented 73 percent of the total sales volume in 2017: stud-welding defects by 75 percent.19 These examples China, Japan, Korea, the United States, and Germany. 20 are indicative of various AI efforts in emerging markets. However, the reality is that most manufacturers in emerging Conclusion markets are still using traditional data analysis methods. Similar to previous industrial revolutions, when Developed countries: The most advanced industrial innovative machines and technologies replaced countries, which are also part of the Pillar 3 group of conventional methods of production and spurred the economies described above, are the primary users of AI invention of sophisticated new products, artificial across many manufacturing sectors. For example, the intelligence has the potential to transform today’s automotive industry has triggered tremendous interest manufacturing. The most complex economies are 7 This publication may be reused for noncommercial purposes if the source is cited as IFC, a member of the World Bank Group. predictably the earliest and biggest adopters of AI This will require additional investment to cultivate and technologies, as they already had the foundations and strengthen sustainable and socially sound manufacturing well-established tech centers in place, where some of the cultures. In Pillar 1 economies, manufacturers need to earliest AI applications were created. These economies build more sustainable and efficient industrial sectors possess a wealth of capital and an abundance of data and minimize the negative impacts of pollution, CO2 for machines to analyze and adapt into the algorithmic emissions, and weak labor standards by adopting advanced patterns that are economically scalable for AI. technologies that bolster complexity. Yet AI is not confined to the world’s biggest and most For years—and particularly in emerging economies— complex economies. As cloud computing capacity expands, the biggest obstacle to adopting AI was the extravagant global data volumes balloon, and processing power cost. Measured against inexpensive workers in low-wage becomes more affordable, cutting-edge applications have economies, the investment made little sense. Now, however, been winding their way through global value chains and as the price of implementing AI falls and data analytics planting seeds in every pillar of complexity. increasingly become the language of global value chains, companies and governments are reevaluating their options In less complex economies, companies can gradually and rethinking their policies. acquire and adapt AI to address unique market needs. 1 Strusani, Davide and Georges Vivien Houngbonon. 2019. “The Role of Artificial Intelligence in Supporting Development in Emerging Markets.” EM Compass Note 69, IFC, July 2019, pp. 1-2. That note defines AI as “the science and engineering of making machines intelligent, especially intelligent computer programs.” This definition is also guided by the AI100 Panel at Stanford University, which defined intelligence as “that quality that enables an entity to function appropriately and with foresight in its environment.” See “One Hundred Year Study on Artificial Intelligence (AI100).” 2016. Stanford University. https://ai100.stanford.edu/. See also Meltzer, Joshua, 2018. “The Impact of Artificial Intelligence on International Trade.” 2018. Brookings. https://www.brookings.edu/research/the-impact-of-artificial-intelligence-on-international-trade/; Nilsson, Nils. 2010. “The Quest for Artificial Intelligence: A History of Ideas and Achievements.” Cambridge University Press; OECD. 2019. “Recommendation of the Council on Artificial Intelligence.” OECD Legal 0449 as adopted on May 21, 2019. https://legalinstruments.oecd.org/en/instruments/OECD-LEGAL-0449; and PwC. 2018. “The Macroeconomic Impact of Artificial Intelligence.” February 2018. https://www.pwc.co.uk/economic-services/assets/macroeconomic-impact-of-ai- technical-report-feb-18.pdf. 2 Doherty, Sean, and Kimberley Botwright. 2020. “What Past Disruptions Can Teach As About Reviving Supply Chains After COVID-19.” World Economic Forum. https://www.weforum.org/agenda/2020/03/covid-19-coronavirus-lessons-past-supply-chain-disruptions/. 3 ResearchandMarkets.com. 2019. “Smart Manufacturing Market, 2025 - Global Market Is Expected to Grow by $320 Billion, Driven by a Compounded Growth of 13.5%.” www.prnewswire.com/news-releases/smart-manufacturing-market-2025---global-market-is-expected-to-grow-by-320-billion- driven-by-a-compounded-growth-of-13-5-300907605.html. 4 Infosys. 2019. “Amplifying Human Potential – Towards Purposeful Artificial Intelligence.” https://www.infosys.com/aimaturity/Documents/amplifying- human-potential-CEO-report.pdf. 5 Walker, Jon. 2019. “Machine Learning in Manufacturing – Present and Future Use-Cases.” August 13, 2019. https://emerj.com/ai-sector-overviews/ machine-learning-in-manufacturing/. 6 IFR.org. 2018. “Executive Summary World Robotics 2018 Industrial Robots.” https://ifr.org/downloads/press2018/Executive_Summary_WR_2018_ Industrial_Roots.pdf. 7 Rosen, Rebecca J. 2011. “Unimate: The Story of George Devol and the First Robotic Arm.” August 16, 2011. https://www.theatlantic.com/ technology/archive/2011/08/unimate-the-story-of-george-devol-and-the-first-robotic-arm/243716/y-320-billion-driven-by-a-compounded-growth- of-13-5-300907605.html. 8 Walker, Jon. 2019. “Machine Learning in Manufacturing – Present and Future.” 9 T.A. Cook. 2013. “Maintenance Efficiency Report 2013”, August 2013. https://www.tacook.com/en/expertise/.com/ai-sector-overviews/machine- learning-in-manufacturing/ 10 Walker, Jon. 2019. “Machine Learning in Manufacturing – Present and Future Use-Cases.” August 13, 2019. https://emerj.com/ai-sector-overviews/ machine-learning-in-manufacturing/. 11 See https://simularity.com/solutions/. 12 Fantozzi, J. 2019. “Domino’s Using AI Cameras to Ensure Pizzas Are Cooked Correctly.” Nation’s Restaurant News. https://www.nrn.com/quick- service/domino-s-using-ai-cameras-ensure-pizzas-are-cooked-correctly. 13 5G Infrastructure Public Private Partnership. 2015. “5G and Factories of the Future – White Paper.” https://5g-ppp.eu/wp-content/uploads/2014/02/5G- PPP-White-Paper-on-Factories-of-the-Future-Vertical-Sector.pdf. 14 Lou, David Zhe, Jan Holler, Cliff Whitehead, Sari Germanos, Michael Hilgner, and Wei Qiu. 2018. “Industrial Networking Enabling IIOT Communication.” https://www.iiconsortium.org/pdf/Industrial_Networking_Enabling_IIoT_Communication_2018_08_29.pdf. 15 Al-Saeed, Mohammed A., Soliman Al-Walaie, Kahlid Y. Alusail, and Turki K. Al-Anezi. No year. “How to Use 4G LTE Wireless Technology to Secure Industrial Automation and Control Systems.” https://automation.isa.org/4g-lte-wireless-technology-secure-industrial-automation-process-control- systems/. 16 Galbraith, Christina. 2015. “Artificial Intelligence Catches Fire in Ethiopia.” August 25, 2015. https://techonomy.com/2015/08/artificial-intelligence- catches-fire-in-ethiopia/. 17 See https://dataprophet.com/. 18 Ibid. 19 Ibid. 20 Ifr.org. 2018. “Executive Summary World Robotics 2018 Industrial Robots.” https://ifr.org/downloads/press2018/Executive_Summary_WR_2018_ Industrial_Robots.pdf. 8 This publication may be reused for noncommercial purposes if the source is cited as IFC, a member of the World Bank Group.