Generative AI in Manufacturing

Meeting the future of manufacturing, today

Man sitting at a desk working on a computer generating code

Artificial intelligence (AI) is everywhere, but experts say we’re just scratching the surface of the value it can bring. Its growth potential, risks, and future can’t be ignored. One section of AI making more commercial debuts is generative AI. From writing to coding to analysis, generative AI is here to stay. 

To better understand what it is, how it can be used, and its challenges, we spoke with Jason Lagneaux, Research and Development Manager of Services and Customer Experience at Intralox.

What Is Generative AI? 

Before identifying generative AI’s potential, it’s important to first break down what generative AI is. At its simplest, generative AI produces novel content such as text, images, videos, and code using generative models. These models are trained on a wide variety of materials, and then find patterns to create something new from the underlying patterns. 

“It’s a subset of AI that takes billions of data points to train a model,” explains Lagneaux. “In this case, we’re talking about text, and text-based generative AI takes a data set of hundreds of billions of words to create a model. This resulting model is effectively the patterns in words. Once it’s trained on the data, you then have a base model to start creating.” This base model is paired with a user’s initial input or prompt that’s used to seed a new creation. 

The Potential of Generative AI in Manufacturing 

Text-based generative AI has a near-infinite number of practical applications, especially when you consider the models these tools can access. From recipes to poetry, text-based generative AI can achieve a lot. But what manufacturing applications are possible? 

I have no doubt Generative AI will cause both positive and negative disruption. The biggest unknown is the rate of change coming. If the change happens too fast, it could cause more negative disruption than desirable.

Jason Lagneaux
Jason Lagneaux
Research and Development Manager of Services and Customer Experience at Intralox

“I think this is what most companies’ stakeholders are trying to uncover right now,” says Lagneaux. “I can see how generative AI can and will be valuable for broad productivity gains across all roles of manufactures, like for translations, writing assistance, troubleshooting, building enhanced learning tools, and assisting engineers program automation controllers.” 

This may mean more completed work, increased productivity, optimized processes, and greater capacity for teams. In summary, generative AI offers the potential of actual time and money savings for cross-functional teams and individuals. 

The Risks of Adopting New Technology 

In this post-internet era, technological advancements paired with global media share new technologies with the average person faster than in previous generations. This can be exciting, and in the world of AI—especially generative AI—excitement is heightened because every interaction with the tool could improve it. It’s an environment, an ecosystem, where we all can contribute to and watch something improve. But what if you don’t want to share your intellectual property (IP)? When is it for the greater good and when are you helping your competitors?

Most people who use generative AI tools apply them in both personal and professional settings. Lagneaux uses them daily. Though AI isn’t new by any means—you may be familiar with website chatbots and social media algorithms—generative AI that anyone with an account can access is indeed new. What risks and challenges does this pose to companies?

Did you know? In a June 2023 report by KPMG, 78% of industrial manufacturing executives surveyed selected generative AI as "the top emerging technology (compared to 67% in March)."

“With most things, and especially in business, it’s critical to analyze the risk vs. reward of any new technology,” says Lagneaux. “Since we’re still in the early stages of working with and discovering how generative AI can work for us, we always learn as a team to uncover any risks. One challenge is people’s perception of it: whether AI makes people uncomfortable or not, and how we work around or with that. Because of the large potential for disruption, we must research these tools. But we must do it in a way that’s grounded on morals, values our people, and reflects our business philosophy.”

Unique challenges are to be expected, and the risks are certainly enough to demand close attention and care. “To truly leverage generative AI at scale you have to trust tech companies with your intellectual property even more,” explains Lagneaux. “There’s the risk that IP accidentally gets shared. We know these base models will primarily be offered by a small number of tech companies, some of them relatively new and unproven. The challenge is balancing the new risks with the new—hopefully large—rewards.” 

Generative AI in Conveyance Automation 

Lagneaux discussed some of the potential he sees generative AI bringing to work environments. In particular, what it can offer the world of conveyance automation. 

“Many controls engineers who normally work on PLCs, where they write ladder logic, are already using generative AI tools,” says Lagneaux. Since this work is text based, generative AI tools make it easier. “It’s hard to imagine not using a tool like this in the future,” Lagneaux continues, “because you can’t compete otherwise; you may not be as fast or precise as your coworkers.” 

He compares how generative AI can help lead the way for data analysis to “having a personal data scientist assistant at hand, all the time, for anything. And to process your data in a fraction of the time.” 

Evaluating the authenticity of claims around generative AI is the toughest challenge right now. Emails from vendors flood in, all claiming to leverage AI, but discerning what’s real requires a deep technical understanding and drilling down to foundational questions.

Jason Lagneaux
Jason Lagneaux
Research and Development Manager of Services and Customer Experience at Intralox

And though there’s a lot of potential for text-based generative AI in conveyance automation, it’s important to remember that generative AI is multimodal. “Generative AI is and can be very visual,” says Lagneaux. “There hasn’t been much generative AI in the CAD space yet, but it will come with time.  I can’t wait to see how the major CAD vendors incorporate AI to eliminate the drudgery. Maybe one day soon we can fine-tune base models in our CAD systems to assist and accelerate line layouts with our best layout IP that we’ve developed over the years.” 

Navigating Generative AI and the Future 

In this evolving landscape of AI technologies, information is leading the charge. Knowing—or finding out—what’s real or not can be the make-or-break point. 

“In today's landscape, everyone’s grappling with distinguishing what’s real and valuable in the realm of generative AI,” says Lagneaux. “It’s not about chasing shiny objects; it’s about finding genuine value worth taking a risk on.”

News & Insights