A 2016 U.S. Department of Commerce survey of 80 U.S. manufacturers and vendors reported that smart manufacturing would provide $57 billion in annual cost reductions. This represents an approximate 3.2% reduction in the shop-floor cost of production.
The promise of smart manufacturing is significant. Artificial intelligence is already being used to improve safety, quality, maintenance, scheduling, product design. But many companies do not have the necessary culture in place to benefit.
A Weak Problem-Solving Culture Hinders AI
Smart manufacturing depends critically on information governance: rules concerning the collection, flow and analysis of performance information, most often in digital form.
If your company isn’t already good at these things—if it doesn’t already possess a culture of curiosity, effective data gathering and use of data in decision making and problem solving—it won’t suddenly get good at these things upon installing AI.
A past client of Rick’s experienced a good bit of unplanned downtime on its extruders. It turned out that the organization possessed a considerable data regarding the extent of that unplanned downtime from week to week, which had been manually gathered by the operators.
Nonetheless, managers and engineers didn’t seem much interested in tackling the unplanned downtime problem. They didn’t trust the data being gathered by the operators. And so, the downtime problem continued. Mostly, leaders just shook their heads and said, “Ain’t it awful.” That client lacked a culture of good problem-solving.
... As Does a Fix-the-Blame Culture
On the other hand, organizations that presently have a culture of using performance metrics simply as a way of monitoring and “holding workers accountable” will also have trouble implementing AI effectively. Those companies will use AI simply to create resentment and fear among workers even more quickly than they once did. Sadly, some proponents of AI see this use as an appropriate deployment of the technology.
An article we read had this to say about the promise of AI: “After a company deposits enough big data from the workflows, a manager can query the average time it takes for a field worker to perform routine maintenance. When a particular worker goes out to complete the task, the manager can tell them, ‘The average time to complete this is two hours, and you’ve been taking three hours.’
If AI is used to enhance a “fix the blame” culture, it will fail in all instances. All to say, a mismatch between culture and technology will lead to a failure of the technology.
We’ve Been Here Before
We’ve already seen this with Enterprise Resource Planning (ERP) initiatives over the past thirty years. Many millions of dollars have been spent on that technology, which made similar promises of large increases in productivity achieved through better business process orchestration within and across business functions. Studies show that somewhere between 55% and 75% of ERP implementations failed to meet expectations or failed outright.
Several years ago, Brandon managed a large machining plant that made critical pressure control equipment. The company had made a large investment in technology that promised an end to operating inefficiency and was touted as an easy integration with current systems.
Brandon was concerned that his team had just started using data to better understand the plant’s operations and to solve their problems. Fundamental data analysis tools weren’t well understood and were little used. Other basic tactics like bill of materials (BoM) and router accuracy audits were not deployed to ensure ERP systems data integrity.
The software vendor promised a turnkey solution to those challenges. Checks were written for the hardware, subscription services for its use were signed into agreement and, once the company plugged everything in and turned it on, the company would jump from a 1990’s machine shop to the modern century with the turn of a key.
The reality was far from that.
Because a culture of good data gathering and analysis wasn’t already in place, there was no disciplined approach for formal systems management and problem solving. When the system turned on, the plant was treated to a display of red, yellow and green blinking dashboards that made little sense to operators. They grew frustrated with the ongoing false alarms the system generated due to bad information fed into the new system.
Over time, everyone simply quit using the new technology. One by one, the system’s screens faded to “off” and were eventually disconnected and packed into cardboard bins where they were forklifted to some remote location in the warehouse to become a dusty monument to a failed project.
Management’s illusions were replaced with the realization that such technology fails when it is implemented on a weak foundation of data management, gathering, and problem-solving fundamentals. Company leaders eventually accepted the fact that the company was first going to have to take the journey into problem-solving proficiency the old-fashioned way.
AI Works Best in A Strong Problem-Solving Culture
Ironically, companies who seem to need AI the least are probably in the best position to take advantage of it, because they are already so good at identifying and addressing problems. Ron once worked for a candy company that was having serious scrap problems. Ron led his associates in an effort to get better data. As they gathered more information, the best approaches became clearer. They were able to identify the sources of the problem and correct them. In other words, a culture of good data governance provided a foundation for effective problem-solving without AI. Nonetheless, it’s exactly this culture that will be most successful in implementing and deploying AI.
Such a culture is established only through an investment of time, financial resources and, most importantly, the attention of leadership. Rick’s client, mentioned above, could have taken a number of steps to improve its capacity to solve the unplanned downtime problem. Leaders could have had conversations with operators about the manner in which downtime data was collected and how it could be made more effective.
Leaders could have had conversations with operators about their experience with unplanned downtime and what hypotheses they had with respect to the issue.
Leaders could have spent more time on the plant floor observing operations to develop their own hypotheses, which they’d then share and discuss with operators. The company could have trained leaders and operators in good data-gathering and problem-solving skills and made sure that their use was reinforced. My client did none of these things.
As Brandon’s story illustrates, there will be plenty of vendors of AI technology who will tell companies like my former client that their “plug and play” technology will overcome those cultural shortcomings. They’re wrong. Before you invest in AI, you need to invest in creating a strong problem-solving culture.Rick Bohan, principal, Chagrin River Consulting LLC, has more than 25 years of experience in designing and implementing performance improvement initiatives in a variety of industrial and service sectors. He is also co-author of People Make the Difference: Prescriptions and Profiles for High Performance.
Ron Jacques is a 35-year veteran within the lean, manufacturing and consulting arenas. He is a Certified Lean Practitioner who has delivered hundreds of kaizen and transformational solutions to clients and companies within the Pharma, Medical Device, Automotive, Food/Beverage, Electronics, Military Defense, Personal Care, Consumer Durables and Capital Equipment industries.
Brandon Davis is a 24-year distressed asset facilities turnaround leader. Recently, he helped lead an Industry Week Best Plants-winning team at NOV Texas Oil Tools. He has led positive lean and cultural change in a variety of core manufacturing industries that service consumer goods, steel production, wire and cable, automotive, oil and gas drilling equipment machine shops and assembly operations globally. He is also a U.S. Army veteran and holds an MBA in Global Business Management.