As new developments are made in Artificial Intelligence, conversations often turn to concerns about potential threats to humanity—and rightly so. At the same time, however, it often seems too little attention is paid to the possible benefits for manufacturing and the supply chain, where there is potential to change the way we work by improving processes ranging from material requirements planning and capacity planning, to reducing a plant’s energy consumption as part of Green initiatives.

 

You may have seen a back-and-forth exchange in recent weeks between Tesla and SpaceX CEO Elon Musk and Facebook CEO Mark Zuckerberg. It began at the annual meeting of the National Governors Association, where Musk said that even though we may not fully grasp the “real threats in front of us”, we all need to be “quite concerned” about unregulated AI research. He later doubled down on earlier warnings about the threat posed by AI.

 

“I have exposure to the most cutting-edge AI and I think people should be really concerned about it. I keep sounding the alarm bell about it, but until people see robots running down the street killing people, they don’t know how to react, because it seems so ethereal,” Musk said. “AI is the rare case where we need to be proactive in regulation, instead of reactive, because by the time we are reactive in AI regulation, it’s too late. AI is a fundamental existential risk for human civilization.”

 

Later, during a “live” Facebook discussion from his backyard, Zuckerberg said he doesn’t understand people who are “naysayers” and try to “drum up” doomsday scenarios.

 

“I actually think it’s pretty irresponsible,” said Zuckerberg, who also added that he believes people “can build things and the world gets better.”

 

With that type of exchange in mind, I was interested to read an opinion piece on manufacturingglobal.com. There, Jane Zavalishina, CEO of Yandex Data Factory, points out that while people focus on talking and emoting robots when discussing AI, there is also a so-called “narrow” AI, which is very good at specifically defined, constrained tasks, where it can deal with uncertainty better than humans, such as in evaluating a manufacturing process over time.

 

In the real industrial world, nothing is ever fully known, Zavalishina says. For example, the exact composition of raw materials vary, as do external conditions, and multiple fluctuations occur in every process. What’s more, classic science can only describe the physical or chemical processes happening inside the equipment with a degree of certainty, she writes.

 

“Process control is never precisely accurate: the operations happen within a permitted range and a great deal of variability, which results in inefficiencies,” Zavalishina observes. “This can come in the form of the excessive use of a certain raw material, suboptimal energy consumption levels, scrap due to defected production, or poorly planned logistics. These inefficiencies are as equally as bad for environment as they are for the bottom line.”

 

What makes AI especially attractive, Zavalishina says, is that manufacturers can use AI to drive improvements—such as consistently reducing variability—without changing the process itself or requiring significant capital investment. That’s because AI enables more precise decision-making for each individual process iteration, having learned from previous cycles, she writes.

 

For example, Zavalishina says AI can precisely model the expected output of a process, thereby allowing more precise identification of raw material requirements. It also can recommend the best operating parameters to reduce energy consumption without affecting throughput, she continues. Another example is to use AI to detect hidden product defects early on to prevent further processing bad lots.

 

What do you think of “narrow” AI and its ability to help solve manufacturing challenges or those in the supply chain? What types of “What-if?” analysis would benefit from the use of AI?