AI Opportunities and Obstacles in Manufacturing

Where to start

According to Lomasky, there are some key areas of manufacturing that can benefit from the power of AI, including: increasing productivity via robots that can work on a packaging line 24/7 to increase output and efficiency; analyzing consumer data to create personalized products, packaging, and branding; improving quality control via AI-powered vision systems to quickly detect defects and flaws; and connecting machines and systems to enable real-time analysis that optimizes production and decreases downtime.

To that end, there are specific use cases for AI in manufacturing such as generating step-by-step work instructions for technicians performing assembly and inspection processes, analyzing machine data to optimize OEE or reduce waste, or assisting design engineers to create new product concepts using 3D models and prototypes.

Applied correctly, AI has the potential to disrupt—in a good way—the current approach to product design, production, packaging, and distribution. It also has the potential to destroy an organization if not approached or managed correctly.

“There’s a lot of risk with AI, so you have to use it responsibly,” Lomasky said. “My recommendation is to start with a policy. It doesn’t have to be anything super formal or even anything that robust, a page or two will do, but write down on paper ways that are acceptable in your company to use AI, or ways your staff should never use AI.”

While there are hazards associated with unintended bias, as AI tools can make assumptions based on the information it consumes, the bigger risk is the potential for a data breach.

“There’s risk of confidential data getting into the wrong hands,” explained Lomasky. “Cybersecurity is extremely important, especially if you’re talking about putting in machinery data or recipe data or anything that is proprietary, you want to make sure any AI platforms you interact with are protected with a really strong password, have multi-factor authentication if they offer it, and make sure any access privileges are using the rule of least privilege, that is the least amount of access someone might need to get their job done.”

CPGs should consider implementing these AI governing polices.PMMI

You also want to properly vet any AI platforms to ensure they have controls around how they use the data or the prompts put into to train the AI model, Lomasky said. This ensures none of your data ends up in someone else’s AI response.  Much of this can be solved by having the right AI governing policies in place around acceptable use, data collection, ethical use, prompting, and outputs.

AI in the future factory

It’s still very early in the AI game. “If you think about a maturity curve, we are not crawling yet,” Lomasky noted. Therefore, there’s a lot of room for AI to evolve in manufacturing.

Some futuristic applications include superior data analytics that will uncover new kinds of insights previously unimaginable, as well as empowering machinery to do prescriptive maintenance in that it will self-heal, tweak a recipe, or stop a job all on its own.

Lomasky has big visions for the future of AI. “We talk about the concept of predictive maintenance to predict [machine] failure, but what if it could use AI to take corrective action to keep the line running?”

The good news: People are still important to the process. Highly-skilled individuals will still be needed to train AI, validate what it’s doing, and adjust to ensure accurate outputs.