[Honeywell]
“Unfortunately, in the industries we’re serving, there are more people retiring and leaving the industry than there are new people coming in,” Urso explained in an interview at the Honeywell Users Group in Dallas. “So there’s a net reduction of people, and there’s also a big demographic shift, particularly in developed regions where a lot of retirements are happening.” As a result, he emphasizes, “We have to find ways to preserve and transfer that knowledge to the next generation.”
Closing the expertise gap in industrial sectors
Jason Urso
When seasoned operators walk through the plant, sensing the hum of machinery, they often instinctively know when something is amiss. But training greener employees to pick up those skills in short order is elusive. “So how do we help these new people accelerate their time to expertise?” asks Urso, Chief Technology Officer at Honeywell Process Solutions. “If we know that the 30-year expert is making decisions based on this very sophisticated knowledge graph in their mind, what if a computer could do that?”
He explains that the intuition of seasoned industrial maintenance technicians, reliability engineers, and the like is “all human intellect.” These professionals develop keen intuition for diagnosing equipment issues through years of experience, combining their cultivated senses with modern diagnostic tools to detect subtle shifts in pump whines, off-gases from distillation columns, and other telltale signs of potential problems.
Urso provides an example: “For example, what if a two-year employee goes and looks at a piece of equipment and says, ‘It’s vibrating, it’s making a noise. I’m not sure what that is, but I can enter that into my system.’ They can input, ‘On this particular pump, it’s vibrating and making a loud clunking noise.'” An AI-enabled system could then reach into the repository and inform the worker, “‘Actually, that has happened before, and it’s probably this. Take these couple of steps,'” Urso said. “AI is wonderful at doing that type of pattern matching if you build up the data set the same way a human built up their mental data set.”
AI as a knowledge preservation tool
Honeywell aims to use AI to capture the knowledge of seasoned experts and make it accessible to junior workers. “We want to build a massive knowledge repository from experienced workers so that enterprise knowledge isn’t lost when someone retires,” Urso said. “AI excels at pattern matching when the data set mirrors how humans build their mental models.” One of the themes of the HUG event was that Honeywell’s AI strategy wasn’t about selling AI for its own sake, but offering AI tools based on user feedback to help solve specific problems.
“We have to find ways to preserve and transfer that knowledge to the next generation.”
While in the longer run, deep learning tools trained on vast datasets of cleaned sensor data, expert human knowledge, and other sources could redefine the concept of smart facilities, “for now, the benefit is about how we make the humans as good as they can possibly be,” Urso said. “How do we make someone who is more junior have access to the same level of expertise that a 30-year veteran has? That’s where I’m really excited.”
Honeywell is expanding its offerings
Honeywell is putting these ideas into practice with several initiatives. For example, they are developing a Qualcomm-backed AI-enabled Multi-Modal Intelligent Agent for mobile devices, set to launch in early 2025. This agent will allow workers in distribution centers and retail industries to interact naturally with handheld devices through voice, pictures, and barcodes, functioning as a digital resource for the modern workforce.
Honeywell has also announced a strategic collaboration with Chevron to develop AI-assisted solutions for refining processes. This partnership aims to create a new generation of AI-assisted alarm management solutions, providing operators with guided actions to respond to alarms and operational events. The goal is to reduce lost profit opportunities and process safety incidents by integrating AI into Honeywell’s Experion distributed control system.
AI as a cognitive partner, not a replacement
While the integration of AI into industrial processes brings significant new capabilities, Urso emphasizes that it’s not about replacing human workers but enhancing their capabilities. “I think it’s a great opportunity for us to apply AI to help improve the skill of humans. I’m not quite on board with it displacing humans yet, not in our industry anyway,” he stated. “There are still humans needed for safety reasons and to fix things. Until robots come and are doing all that stuff, humans will still be there.”
Tapping industrial data promises recursive benefits as connected devices grow
The vast amounts of data generated by industrial operations present a significant opportunity for AI to drive innovation. “With those archives of data, being able to identify correlations of things to deliver new insights, I think that’s the power of AI,” Urso explains. “We use that archive of data, draw an insight based on past history, and present that to a human in the present, saying, ‘I’ve seen this happen before. This is what the problem is. This is the action you need to take.'”
By unlocking the knowledge already existing in data repositories, AI can help achieve “the next generation of benefits” in the process industry. Urso believes that leveraging historical data can lead to improved decision-making and operational efficiency.
However, effectively tapping industrial data is not without its challenges. “First of all, data is stored across many, many systems. There are no correlations between the data. So overcoming that problem is a big challenge,” Urso points out.
“AI is wonderful at doing that type of pattern matching if you build up the data set the same way a human built up their mental data set.”
“In any data science project, when we’re working either internally or with a customer, the first step is combing through the data and making sure the data itself is even valid,” he explains. “Because a lot of times there’s data that’s not clean.”
Urso acknowledges the complexity of preparing data for AI applications. “Figuring out how to clean that out so you could do a proper analysis can be a pretty enormous task,” he says. “From my perspective, and what we’ve tried to do, is to automate that process so that we can get to a clean set of data as quickly as we can.”
“AI is set to transform industrial work by bridging the expertise gap caused by retiring workers and a shortage of experienced professionals,” Urso believes. “Our goal is to make the humans as good as they can possibly be, ensuring that industries continue to operate safely, efficiently, and innovatively in the face of evolving challenges.”
