Asset management and production have gone through significant changes over the past decade, with digitalization at the forefront. Now, with AI in the mix, companies are figuring out where the technology can deliver the most value — and the manufacturers seeing the strongest results are taking a measured approach rather than trying to implement AI across the enterprise all at once.

Taking the pulse of the industry with the help of subject matter experts who work at some of the biggest providers of optimization tools offers a useful snapshot of where companies stand today.

In talking with Rahul Garg, Global VP of Industrial Machinery, Siemens Digital Industries Software, we went over some of the use cases for AI in industry. The companies incorporating AI into their digitalization and optimization strategies through a phased approach appear to be benefiting most. They are also better prepared to scale those applications into new areas of production.

Where AI is delivering the most value in industrial operations

The three big areas Rahul identified where he’s seen Siemens customers successfully implement AI as part of optimization are:

1. Asset performance and predictive maintenance. Using AI to monitor equipment, predict failures before they occur and prioritize maintenance issues — essentially anything that reduces unplanned downtime. Moving from reactive to proactive.
2. Closed loop optimization. Using AI to identify production bottlenecks, improve scheduling and optimize throughput, all to drive production efficiencies while taking into account change requirements, scheduling requirements, production equipment changes and even supply chain issues.
3. Training. To meet the need for skilled operators, AI is proving to be a valuable tool, making training materials more interactive and learning more engaging and faster. Not just on the shop floor, but also on the engineering side, it is a tool for quickly familiarizing someone with the processes and equipment.

AI is accelerating processes, especially for training. Simulation tools, like digital twins, make it easier to gain valuable information, and learning how to use those tools can now be done much more quickly. Skills and knowledge that used to take years to build through daily interactions with the equipment can now be learned through interactive tutorials using AI. 

Giving engineers and operators access to better, more accurate information helps them make faster and more informed decisions.

How manufacturers move AI from pilot to production

Rahul went on to explain that at a recent roundtable event in Amsterdam, he brought up AI with about 15 Siemens customers. He’d expected a few would be dabbling in AI, but the answers he got very pleasantly surprised him. “They were all deep into AI. They’ve been using it, trying to expand it and driving more adoption,” Rahul explained. “They’re moving from the initial piloting, prototyping phase to the production and industrialization phase.”

Why human oversight still matters

A concern about possible hallucinations creeping into the process illustrates why keeping the human-in-the-loop is an important part of optimizing and scaling, just like the human is needed to oversee, monitor and use AI. AI should be seen as a tool, not a replacement of the human. “

AI is a decision support system, but it’s not a decision-making system,” Rahul said. 

Optimizing and adopting at scale remains the hurdle many companies are still figuring out how to clear. A recent report released by Rockwell Automation, Scaling MES Across the Enterprise, puts some stats to the scaling challenge. Particularly in Chapter 3 of the report, their findings indicate that manufacturers expect 42% of manufacturing processes to become AI-supported within the next year. However, they also note that AI is scaling faster than the systems required to support it, with barriers to AI readiness listed as:

• Data security
• Cost
• ROI uncertainty

“Manufacturers are advancing AI strategies while still working to strengthen the system environment that supports secure, scalable and high-value AI use … Among those reporting that their data creates meaningful value, 38% point to AI and machine learning use cases, with quality (36%) and process optimization (35%) close behind.”

Closing the gap toward AI-led process and production optimization will require long-term planning, effective change management, and skills training for a workforce that can use AI as an effective helpmate.