of Traditional Artificial intelligence, which has grown over the past decade, has crunched numbers by looking for patterns and providing predictive analysis based on likely probabilities. Among many features, we introduce generative AI that provides gateways for: numerical AI predictions and observations open up possibilities for highly interactive verbal inquiries.
Related article: This year's top 8 AI use cases and what tech experts need to support them
Peter Zornio, Emerson's senior vice president and chief technology officer, said generative AI will help open up the previously very vague AI black box for a variety of enterprise functions. It says it could help bridge the gap between operational technology and information technology. I recently met Mr. Zornio in New York. There, he explained how generative AI and numerical AI represent opposite ends of a continuum. The two variations are based on numerical models and language-based models.
Although the technical underpinnings of the two AI variants are the same, the way they operate differs, he says. “Number-oriented production models are based on numerical data sets,” he explains. “Language models use datasets based on countless documents, images, and more.”
Now, he says, these two ends of AI are merging, opening up new territory for the behind-the-scenes aspects of traditional AI. “We're seeing the two used together,” he says, Zornio. “In an industrial environment, we use language-based models as a way to work with numerical-based models that we already have. So can you imagine an operator saying something like this? What should I do?”
This has a tremendous impact on productivity and time savings, he continues. “It's a natural way of interfacing. That's how he talks to experts with 30 years of experience at the company, right? He might ask Fred from engineering, 'What's going on?' Then Fred goes and looks at all the trends in production and eventually comes back and says to you, “Well, normally when this is happening, what's happening is the catalyst is getting dirty. and this is what you need to do.'' You should probably stop and play. ”
Also: GPT-4 Turbo takes back the title of “Best AI Model” from Anthropic’s Claude 3
Human talent is essential, and what Fred, the engineer, is doing is “using the model he's built in his head from running the place for 30 years,” Zornio said. Generative AI takes over that work, working with numbers-based AI to interact with computers in the same way professional engineers do, using scientific deduction. “You could also look at the past five years of operations and try to find scenarios where the exact same set of circumstances pattern matched a very similar production footprint; You'd say, “Well, guess what.'' Do you want to do it? Fred would think, 'Last time this happened, we did this.' ”
Finally, Zomio says the AI can “look at all the different scenarios, find out, look at the responses, and say, 'Here are the three actions that have given us the best results to solve the problem in the past.' He will tell you.”
This end-to-end AI approach “provides a great way to build a product support system. You can take all your documentation, all your interactions with your support reps, and incorporate them into your system to help you understand your product. Let them ask questions,’” Zornio says.
Also, just because AI takes on more jobs doesn't mean it will replace you.Here are 12 reasons to worry less
It has applications in every line of discrete and process manufacturing, from petrochemicals to automotive manufacturing. Think about the winemaking industry. Zornio points out that the winemaking industry will also benefit from end-to-end AI. Winemakers with well-sensed vineyards and storage tanks may ask questions like, “Why was this year's wine so much better than last year's?” AI uses “important indicators such as temperature, sugar content, grape acidity, fermentation period, etc. What is the condition of the soil, what is the moisture condition, how much sun is there, how much rain? You can check how much it costs.
AI can act as an assistant in a variety of ways and across a variety of industries, Zornio notes, and be a “great way to interact with and query the models you have.” “They might be more data-generated, meaning generated from numerical type of data, but you might also see scrubs like operator logbooks. Every time something happens, , because the operator will write it down. And if you enter them all, you can ask, “Where has this happened before in the operator's log?” or “What was done to solve the problem?”
Related article: Best AI chatbots: ChatGPT is not the only one worth trying
This also requires greater collaboration between two aspects of the company that have often been disconnected: operational technology teams and information technology teams. This collaboration starts with data. Zornio explains that his IT team and his OT team need to streamline data “in all formats from different manufacturers.” “Historically, there hasn't been a lot of love between these two organizations because operations people have built in their own systems to do all of this, and how they implement it. They have very different ideas about how to use it. Some smarter organizations are doing things like: We need to strengthen it further.”
Therefore, Zomio argues that: “We need to design architectures that allow us to pull data from the OT world to the IT world and back more seamlessly, especially when we talk about using AI systems that might be in the cloud.” OpenAI or other language-based AI models will be interfaced by everyone. ”
