ChatGPT: Using Generative AI to Allow Our Plants and Factories to Talk to Us

When the newest version of ChatGPT dropped at the end of November of 2022, people (including myself) were astonished at its language capabilities.  ChatGPT can build an elaborate and surprisingly accurate narrative about almost any subject when presented with a number of facts by the user.  Love letters, business plans, game designs, even Python code to perform a mathematical function—all within reach of ChatGPT with a well-constructed prompt from a human.  Yes, it makes mistakes and it has limitations, but when it does get things right it is simply remarkable.

Now that the “whoa” moment is over and more reasoned thinking has time to set in, experts in the field have started to make connections between tools like ChatGPT and its practical applications in business.  Some of these are trivial and silly—like using amazing technology to post cat videos.  But other more sensible voices are beginning to define categories of usage that could truly advance many aspects of our economy.

My own experiments with ChatGPT and others involved finding out what it did best with the lowest error rates and high-quality responses.  There is definitely a “better prompts lead to better responses” dynamic when using these tools.  But what they do very, very well, and consistently is to generate a nicely readable narrative from a few cogent facts.  That got me thinking: “Where does the ability to speak to people in written form from a tiny number of facts come into play in business?”

I pondered this question while on a plant tour in middle America.  At one point in the tour, we stopped at a particular machine that gave the production team fits.  “We are always attending to and troubleshooting this machine and the machines it is connected to get it to run at nameplate capacity” the operations exec told me.  “If only we could get it to talk to us every day to tell us how it’s feeling —we would get along so much better”

I sat in deep thought back at the hotel later that afternoon.

Machines talking to the people who run them…Whole factories telling us how well they are working…these are futuristic concepts in the automation domain, but something that is easily within our grasp with generative AI today.  Think about it—a few salient facts from IoT devices on the factory floor or at the plant, fed into ChatGPT with a few rules thrown in about the context of the assets—and you have a plant that “speaks” to you about sophisticated conditions in real-time, perhaps even making suggestions as to make it run better in the moment.  Imagine if that conversation was broadcast to everyone in the organization from top to bottom, as frequently as needed throughout the day.  This “last mile” between the factory and the decision-makers represents the next great connection to be made where ineffective static reports and dashboards dominate the landscape today.

I am not suggesting that this use case is the only value language models bring to business.  What I am saying, however, is that this is a use case that could help us make our physical production sites better, faster, and cheaper by giving decision-makers a much more complete summation of how our physical assets are performing—in real-time.  Your operations, with their new data superpowers, want to talk to you.

Let them be heard.