AI use cases in manufacturing

A lot of hype surrounds the benefits artificial intelligence (AI) will bring to manufacturing, but frequently, few specifics are offered. That changed recently when I had an opportunity to chat with Kirstie Tiernan, a principal at BDO Digital, who serves as the Digital Go-to-Market leader and expert on AI.

Tiernan outlines three use cases where AI can have a positive impact: back-office, preventive maintenance, and safety.

“In back-office finance, everyone focuses on ‘how do I create more capacity with fewer resources?’ ‘How do we close the books faster?’” Tiernan explains. “Instead of processing accounts payable invoices by hand, we can have our people look for duplicates and optimize contracts, because we now have technology to do manual tasks much more efficiently.”

On the factory floor, all the Internet of Things (IoT) data collection offers an AI opportunity. Tiernan says, “Predictive maintenance was a typical use case with manufacturers able to predict when a machine was going to go down. Now, not only can we predict when a machine is going to go down, we can also generate a plan to fix it.”

The more data points you have, and the more you understand correlations, the more you can increase the accuracy of AI models and predictions. Tiernan describes the next step. “We know when a machine is going to go down and how to fix it. Could we renegotiate our service contracts in the future?”

Generative AI can look at many different contracts and, taking in all the purchasing information, determine how you would renegotiate a contract. “The more examples of similar contracts you can feed it helps it understand what your goals are,” Tiernan says. “AI uses a bit more strategic analysis and all that historical data, plus, if you add any economic or supply chain data, it can make a much more nuanced decision than we’ve been able to do in the past.”

Another application for AI is in safety compliance. A security video feed compared to an AI model can detect a forklift moving toward a person and sound an alarm. Tiernan says, “Generative AI could constantly analyze the video stream, and even though you haven’t told it an action is high risk, it may be able to identify a high-risk situation it hasn’t been trained for, because you could be feeding it online articles, case studies, litigation, all the things that have happened to manufacturers regarding safety issues, and it could start developing its own knowledge around how it would identify a risk within your plant.”

Tiernan advises, “Existing software is going to be upgraded with AI functionality, but the real need is going to be having people in your organization who understand what AI is and keeping up with what it can do, who can then translate that into business use cases for you.” – Eric