11050_Peter_Zemsky_AI_Business_School_7_Manufacturing_FINAL_V2

[Driving Business Value from AI, Manufacturing, Peter Zemsky, Eli Lilly Chaired Professor of Strategy and Innovation] >>Peter: Industries focused on manufacturing and other large-scale materials processing are at a pivotal point for the deployment of AI. On the one hand, there's clearly a rich set of applications. These range from quality assurance of final output using advanced image processing to process optimization, making use of vast quantities of IoT data that's flowing from each piece of equipment. However, while the sector has seen much experimentation with proof of concepts, the move to fully scaled AI solutions has been slower than other sectors like retail and financial services. One reason is that the cost focus and tight margins typical of the sector have made it harder to build unambiguous business cases for scaling up. Part of the answer is to understand that initial deployments of AI build experience and capabilities that will enable a series of value-creating moves. However, one still needs results to build momentum. It's important to focus on applications where there is a clear path to substantial value creation. Consider, for example, mining operations. Mines typically involved heavy capital expenditures. In addition to sophisticated and expensive equipment, the mineral rights and site preparation can involve significant capital outlays. Hence, this is a setting where efficiency can have a big impact on value creation. Another key feature of the industry environment is volatile commodity prices that regularly depress margins, and reinforce the search for production efficiencies. Moreover, worker well-being is a critical concern. Mining remains a hazardous activity, typically conducted in remote areas where labor is expensive. Regulation in many countries reinforces a focus on reliable environmental operations and worker safety. Hence, we see some very clear high stakes value drivers that can be the basis of effective data-analytic strategies. First, driving operational efficiencies, given those high capital expenditures, and assuring worker well-being. Moreover, the rough conditions mean that equipment maintenance costs are a key driver of operating expenses. So how can data and AI-enhanced digital solutions drive value creation in the sector, and help mining companies better adapt to their challenging environment? Interestingly, mining is one of the very earliest examples of the deployment of autonomous vehicles, with industry leader, Rio Tinto, a notable pioneer in using self-driving trucks to move ore on its sprawling properties. Why is this a great use case? On the one hand, labor is expensive in these remote locations. On the other, autonomous trucks are operating on dedicated paths, where there are no risks from interacting with the diversity of human-operated vehicles. What is striking today, though, is the emergence of a Connected Mine solution that well demonstrates the value of applying the latest in data analytics across complex process operations. And much of the focus here is on enhancing and protecting workers, rather than replacing them. One leading example comes out of a collaboration between the mine operator, Freeport-McMoran, and Accenture. The initial location was Freeport's open-pit copper mine in Arizona, whose financial performance was under pressure during a period of depressed copper prices. Now, there was no issue with the quantity of data coming from sensors off the vast number of vehicles and equipment across the mine's extraction, transportation, and processing activities. The issue was that the data was neither integrated nor widely available. Major impediments to driving value creation from data analytics. Having each data stream displayed on separate systems in a central control room meant it was only available to experts, a barrier to generating insights and influencing decisions and actions in the field in a timely way. Supervisors were still steering operations based largely on offline data, rather than real time. In order to unlock the value from analytics, the teams created an integrated app that pulled all of the data sources together to create a Connected Mine. With rugged versions of smartphones and tablets, supervisors in the field can now be made more effective. Supervisors get real time notifications about equipment operators that are showing signs of fatigue, which is, of course, a key source of safety problems. Coordination can be improved as equipment breakdowns in one part of the mine can be addressed by reoptimization of activities across the mine with the new production plans communicated out through the mobile devices. There are, of course, important costs to developing and rolling out these integrated systems. A strong ROI depends on finding multiple avenues for value creation. In the case of the Connected Mine, delivering data integration can also support more effective management of the mine operations from department heads up to the corporate. Such rich data can provide new insights into the drivers of, say, machine utilization. Ready access to the data also allowed management to measure the impact of new efficiency initiatives, and operators can see the real time impact of their decisions in the current shift. Just as importantly, mining data can be leveraged for analytics to improve yields, the amount of valuable ore extracted from a given quantity of rock. As many mines age, yields come under pressure, providing further returns from effective data analytics. While the Connected Mine is focused on data integration and dissemination of insights, these advances are a critical compliment to more advanced AI techniques. Well-structured data is essential for machine learning. Just as importantly, the widely distributed mobile devices provide a great channel for analytic insights to influence the real time operations of the mine. Many early AI initiatives across sectors, actually, have struggled with the challenge of integrating insights into operations. After the successful application of the Connected Mine concept in Arizona, the partners have extended their collaboration for another five years to scale the solution globally. The case of the Connected Mine well illustrates key lessons for manufacturing and process industries. It is critical to start with applications that have clear value drivers, where you can demonstrate impact. Here, improved efficiency, better yields, and enhanced worker safety are high-powered levers for value creation. Other mining applications, like AI models to optimize the use of explosives in the mine, or models to decide where to locate mines in the first place, will come. But they pose greater challenges as initial use cases. Unlocking value depends not only on picking the right use cases, but also understanding the keys to effective execution. For me, the keyword is integration. First, the integration of the huge volumes of data from plant and equipment. Rio Tinto's operations reportedly produced 2.4 terabytes of data every minute. Without integration of these streams, data analytics will fall short of its amazing potential. Second, integration of the analytic insights into the organization and its operations. The proverbial marrying of man and machine. The power of the Connected Mine is that it focuses on these two key challenges, laying the foundations for additional AI value creation. Finally, all of this implies that your AI strategy needs to be well-integrated into your broader digital road map.