Adoption of artificial intelligence sounds kind of like the plot of a terrible movie –think “Three men and a cyborg baby”. But so many companies, particularly in manufacturing, struggle to apply the science of AI to their advantage. We can see AI applied in so many related ways. Not that I would advocate this, but perhaps your car is beeping at you because your eyes haven’t been on the road just now. Perhaps your car is letting you know you’ve left your lane. Or it’s even braking for you.
So why is it seemingly impossible to apply any of this amazing technology to the makingthat car, instead of driving it? People are notorious for being creatures of habit. Having familiarity with manufacturing processes that have taken decades to refine, on top of the many years it takes to personally learn those processes on the plant floor every day, can create a very skeptical and resistant community to which to sell a massive paradigm shift like applied AI.
If you can be tough enough to try to understand the challenges these people are tasked with -an elusive goal of trying to achieve 100% quality, 100% of time, with very imperfect information –you’ll learn that AI, despite its many advantages and insights, requires an engaged and informed community, not an army of mindless drones.
AI in manufacturing is an assistant to improve decisions regarding throughput, quality, and even safety. Having a system that can notify you before any unacceptable parts are produced can be an enormous benefit. Imagine being on the plant floor and not having to stop the production line, pulling suspect parts off the line, quarantining the parts, inspecting the parts manually, then choosing to scrap, finesse, or reprocess. Often our “boots on the ground” team members are financially incentivized on part quality or throughout metrics that result in a yield bonus. Avoiding these stoppage scenarios can even keep our team members from ergonomically compromising situations. Our team leaders and production managers are often tasked with improving summarized KPI metrics such as FTC or OEE utilization numbers. And certainly, at the top, overall utilization metrics are used to evaluate our department and general managers. Trying to move this needle by 1% without real information is like swimming in mud –you go through the motions and get nowhere.
Put on those workboots and safety goggles. Look at some bad parts with the team member. Show that operator the warnings that the system provided before that part was produced. Listen to the reactive solution, even if you’ll never have reason to apply it –what they are telling you is how to help them be proactive instead. Watch, count, document, summarize and deliver the estimated annual savings from a reduction in poor quality.
There are natural genetic mutations that give a species a huge advantage over their ancestors, but they can only survive in, and because of, a greater ecosystem. AI’s adoption in manufacturing is much the same way –all levels of an organization can reap the many benefits, but the key is identifying the practical application and a deep understanding of the overall processes.