Podcast: AI and the future of industrial maintenance

TW: To bring this to an industrial maintenance application, you had written some columns earlier last year about how AI is finding some toeholds in oil analysis. Actually, more than a toehold—you’ve seen some firms that are employing it increasingly for diagnostics. Not only because it’s accurate, but because, as you said, it processes information quickly, and you can get results back that much faster to the people seeking them.

MH: Yeah, exactly. And, you know, I’ve been—shoot—I actually started right out of college back in 1985. And during college, I worked on my senior thesis for Raytheon years ago. So really, I started industrial work back in, gosh, ’85. When we look at this, I started doing the whole lubrication oil stuff probably back around 2000, and it just took hold of me. Never really wanted to—it was just a job that was offered to me, so I started doing it. But I found it interesting enough, and I just kept on doing it, right?

So you go from lubrication technology to tribology, which is the study of friction, wear, and lubrication, all the way through to oil analysis. From oil analysis, I said, well, gosh, there are other things called failure analysis, which extends it even deeper. So I really started becoming an expert in failure, which is kind of a tongue-in-cheek thing, but truly I was really fascinated by it.

I found out over the past several years that when I talk to companies and people about the things they’re doing that are successful, what I always come back to immediately is: what didn’t work? What failed on you? But how do you go about inventing a better blend of coffee, or a better coating of paint, or a better engine? You have to try various ideas and go plus and minus, and various concentrations, and so forth and so on, in order to really hone in on exactly what the performance parameter you want is. And that could be a trial of several different iterations. But you can logically lay out what those combinations are to reduce development time dramatically. 

AI, synthetic thought, artificial thought can help us do that. So now, instead of waiting 10 years to develop something that can really be great—like a drug or a material—we could do this in a matter of minutes, and then go through the arduous task of physically doing it, finding out what the performance is, and seeing if that fits into our envelope of acceptability.

So when we look at what AI does in oil analysis, oil analysis is data from the past saying we see a certain trend emerging according to the conditions that this asset is experiencing and the way in which this lubricant is formulated. We can predict where it’s going to go. And we could have been—in, I don’t know, seventh-grade algebra—we understood what regression analysis was. This is exactly why you use algebra: because you can take a variable and chart it on an XY plot, or an XYZ plot if you get into that kind of thing, or go non-Euclidean—which we won’t talk about—but it’s even cooler. You take this crazy equation, and you can actually predict the future. I remember in seventh grade doing algebra and thinking, holy crap, this is powerful stuff. All the other kids were fooling around, and I’m like, guys, you don’t understand—you can actually predict stuff.

We’ve used that in oil analysis and diagnostics. We can look at it and say, okay, we understand what fails, but now what’s even better is we understand what the potential livelihood could be on this asset going forward based on the data we have right now. And we only came up with that through empirical evidence and, understanding that when something broke, we analyzed it. Why did it break? Oh, it looked like this oil had X amount of this and Y of that in it.

We take all that data, put it into an algorithm, and boom, I mean, right there—ChatGPT would do it in a heartbeat. In fact, I’ve written a bunch of algorithms for my company websites that will instantly do root cause analysis and corrective action plans for you. It’s actually using five different AI programs in a multi-layered applicability, all at once—which is how the brain works, so read about that too, it’s a whole other way of approaching synthetic thought. 

AI detects anomalies. Humans still own the root cause. Responsibility has not necessarily been automated. AI is an amplifier. You’ve got to think of it like that. It strengthens competent systems—and it accelerates weak ones into failure as well. So if you really want to think about it, the future belongs to those who can challenge the model, not just follow it. 

TW: Well, you mentioned people’s anxieties over AI in general. Is it going to cause job losses? Is it going to replace my job specifically? That’s everyone’s first question.

I think it’s interesting—the intersection of AI, compute power, and oil analysis. It’s one of those places where professionals may not realize exactly how good these programs and this software is at eliminating onerous work and getting to answers quickly. I guess what I’m getting at is that, at some point, this is going to be such an embedded part of the oil analysis process that, as you said, it’s going to be a companion. And if it isn’t already, it’s being normalized.

MH: Well, it is. In fact, I worked for two major oil analysis firms. I worked at ALS, and I worked at SGS—both great companies, very similar, for that matter. Both had lots of diagnosticians. And before I left SGS, I remember I was overseas working, and one particular lab manager showed me a diagnostic report. I read it and said, “this is actually pretty good, did so-and-so wrote this?” And he said, “No.” He said, “An AI program wrote that.” I’m like, really? This was two years ago—about two and a half years ago. And I said, really? He said, yeah, it’s something we have. I said, “this is good. This is really good. I said, how long did it take to generate?” He goes, maybe a second. Wow.

Now, I’ve done diagnostic reports as well. The most anyone can typically do in a day is about 300. After that, you just go brain-dead. It’s difficult. You have to pore through data. You might be able to do a simple one in less than a minute, but sometimes you have something more sophisticated. That could take you five minutes to research exactly what the right response would be, because you’re dealing with everything from a piece of farm equipment to something over in a paper mill—and anything in between, anything from a gearbox to a hydraulic system to a bearing. There’s all kinds of stuff. Engines alone are crazy. So to have a system do that for you that quickly—and that good—yeah.

And I remember calling my boss at the time and saying, “you want some cost savings? I bet you could eliminate about 237 jobs within a day like that.” He’s like, oh no, we don’t want to do that. He said, this thing will do it? He’s like, you serious? I’m like, oh yeah. He goes, well, what are you going to do with all these people? I said, “what are you going to do with them? You’ve got technically sound people. I’d put them in factories and do audits. Actually have them step up and say, ‘I tell you what, I’ve got data here—let’s figure out the whole entire system and how to take this one step up,’ ” right?

But the idea is this: I use this as an example. My great-grandfather came over from Ireland, and he was a blacksmith. He opened up a blacksmith shop in New London, Connecticut, and did horseshoes—until he didn’t. Because people started having cars. So he started making automobile springs. That’s what you do. You just pivot. 

I’m sure there’s somebody out there who still makes horseshoes, because people still have horses just like there are still people out there—I think there’s only one company, maybe two, in the whole country now that makes typewriter ribbon. Back in the day, there were plenty. Now there might be only one or two, but they’re still around. And there are still people who make photography film—not many—but the companies that decided to say, okay, we see the market change, we’re still going to go in that direction as well.

Some people are reticent to stay in the past, and they’re conservative. And I understand that, because change is scary. You don’t have control of change. Once again, we go back to the beginning: comfort, control, and convenience. When there’s change, you don’t necessarily have control. That’s very unnerving if you don’t have the confidence to deal with it, understand it, embrace it, and use it.

Change is a chance, right? But if we go about it methodically, with evidence, change can be embraced. And AI is just another form of change we have to embrace because, guess what? It’s not going away.

TW: Yeah. I’ve got a good friend who’s a CIO for a commercial real estate company, speaking of examples, there are two that stand out to me in his career—it’s about a 30-year career now. The first was when he went from paper leasing to digital leasing using iPads. That was a huge change. 

But for what we’re talking about—AI—a different change was the ability to combine a portable thermal camera with drones. Because every year, they used to have to hire helicopter crews and thermal imaging professionals to take those helicopters up, go over the properties, and do thermal assessments to find mold spots they missed or leaks in the roof. It was an annual inspection mandated by the terms of the lease, but it was a very expensive and time-consuming task.

Now, the combination of drones and thermal imagers means you can send them up almost whenever you want to check the properties. You feel bad for the helicopter pilot who has one less job option out there. However, the savings in expense—and the environmental savings from reduced fuel use and noise pollution—it’s a no-brainer. You make the shift.

MH: Yeah. And it’s crazy, too. And here’s something I’ve often challenged. So CRC Press contracted me to write a seven-book series. And it all came about from my first book they published for me, which is called Synthesizing Materials in Microgravity, where we’ve actually built various things in space—like alloys, crystals, polymers, and stuff like that. No one had ever written a book about this.

I said, you know, I’ve been collecting information on this for years because it fascinated me that you can do things up there you can’t do here due to the constraints of gravity. And gravity has a huge influence on chemical reactions—people don’t realize that. So they loved it, they published it. They’ve been selling tons of them. And they said, “Do you have any other ideas?” I said, “Other ideas?” I said, “I actually have an idea for a seven-book series on the microgravity environment.”

The second one we just published was on the effects of microgravity on terrestrial-borne systems. When you put a living creature up in space without gravity, what happens to the cells? Once again, this book is built on all the evidence we’ve gathered from China, Russia, India, and definitely the United States on what happens to living things up there—good, bad, and indifferent.

Now I’ve got five more to write on building stuff up there, terraforming, and all kinds of other things I’m researching. That’ll be pushed out over the next few years. Gives me something to do, right, at three o’clock in the morning.

But here’s the concept. The more research I do into what happens to biological systems up there, the more I realize we shouldn’t be going up there. Let’s send drones. That’s a better idea. I remember I was on a STEM committee with a very famous astronaut who’s also a senator—who’s been in the news lately. He and I got into a one-on-one discussion. 

I said, “You ever think about being on a lunar base?” He’s like, “Oh, absolutely.” I said, “How about a Mars base?” He’s like, “Definitely.” I said, “You’d live on Mars the rest of your life?” He goes, “Definitely.” I said, “Let me get this straight, Senator. You’d wear the same clothes, eat the same food, sleep in the same rack, not get paid anything, and do a job where if you don’t do it, everybody dies.” He’s like, “Yeah.” I said, “Man, that sounds like communism. “When you put it like that,” he said. 

And I said, “Dude, I’d give you five months—and you’re a AAA human being, the best of the best. You’re an astronaut, a Navy pilot, a senator. In my mind, you’re the ultimate male, and you’re a nice guy, too. You wouldn’t last six months up there. I don’t see Joe Blow lasting more than three days without wanting to shoot somebody—or kill himself.” 

Human beings are not going to be good in that environment. We are born capitalists, even though we like to mess stuff up. We just are. I said, “We don’t belong up there. But there are great things to be had up there. We can mine asteroids. Heck, that’d be great. We could do amazing things in space and build stuff. But, God forbid, don’t put people there for any extended period of time. It’s not going to end well. It just won’t.”

But we love our Star Trek, our Star Wars, The Expanse, The Outer Limits—all these science fiction stories that glorify space or terrify us. And really, if we just focus on taking care of things here and making them really good using the tools we have available, a lot of problems go away. And I truly believe artificial intelligence will help us do that—whether it’s analyzing oil, vibration data, thermography data, manufacturing prowess—everything in between. We can use this the same way we use calculators today. And there’s no reason not to. There’s just no reason not to.

TW: Well, let me get us out of here on this question. In your column on AI, you mentioned wage expansion opportunities and wage contraction opportunities. It’s a better way of saying winners and losers—because it’s not really winners and losers—but it is what’s likely to happen in the future, right? You already touched on a couple of examples of wage contraction potentially. Where do you see the opportunities, especially for our maintenance and asset managers in the audience, for wage expansion? What should they be focusing on if they want to grow?

MH: It’s interesting, too. If nothing else, it can be something as simple as this. They say, “Well, robots are going to take over my job.” And I say, well, why not build robots? I mean, if you’re an engineer, why not build a robot? Maintain a robot? We’ll have robots to maintain robots. We’ll build a robot to maintain a robot, then. How about that? 

This reminds me of a project I worked on years ago at former GE Plastics. My job as an application engineer—this was back in the early ’90s—was to go in with materials, but also with a value-added proposition to help them do other things that would help improve overall productivity. And one of the things was automation.

So one of my tasks on this project was to figure out how to automate something. GE Plastics had come up with a really high-temperature adhesive used on range tops. You know those glass cooktops that sit on top of the heating coils? At this particular company, they had a couple dozen people that would have these pieces of glass on this device that would pick it up, and would take this high-temp glue and squirt it our manually around the periphery of this glass piece, flip it over, set it on the frame, let it cure, and move on. And they said, “Why can’t this be automated?” And I said, “Well, that’s exactly right.”

So I got in there, and they said, “what we need you to do is map out the points,” which means paying attention to the person doing the job so we could emulate it. I visited the plant, spent about half a day mapping out this process, measuring and timing and all this other stuff so we could automate it, basically duplicate what they were doing.

First thing in the morning, eight o’clock, the plant manager was very excited to get me in there because, you know, you’re from GE. They put me with one particular line worker. She was five-foot-nothing. I remember she was wearing a yellow hard hat with a blonde ponytail sticking out the back—really kind of cool. This was in Kentucky, and she talked like she was from Kentucky, which was kind of endearing.

She said, “So you’re going to be paying attention to me all day?” I said, “Sure.” I told her I’d be taking copious notes. She said, “Oh, let’s go grab a cup of coffee—it’s break time. We get 15 minutes.” So she brings me to the break room, and we sit down with coffee. She said, “So what is it that you’re actually doing?” 
I said, “Well, I’m mapping out what you do for your job.”
She goes, “Oh. Why?” 
I said, “Because we’re going to put equipment in here to help you out.”
“Really? What’s the equipment going to do?”
“Well, it’s going to pick up the glass, squirt the glue on it, and put it back on the frame.”
She said, “Well, what am I going to do?”
And I said, “Uh, well, you don’t want to do this the rest of your life anyway, do you?”

That was me being a stupid engineer with advanced degrees, and, you know, this person just got out of high school and landed a manufacturing job. And she says, “I love my job. I like what I do. I like the people I work with. I don’t have to worry about anything. On Friday, I go home with my boyfriend, go to the lake, have beers with our friends. Monday morning, I start all over again. It’s nice. I like my job. I know how to do it really well.”

And then she looks at me and says, “But what you’re going to do is take my job away from me.” I looked at her and said “I don’t know what to say, I’m just doing my job.”  I finished the project, flew home the next day, and went into my office talking to my boss, and I said, “I have a bit of an existential issue.”
He goes, “Oh Jesus, what now, Holloway?”
I said, “We’re taking away people’s jobs.”
He said, “Yeah, they’ll find other jobs.”
I said, “But they like their jobs.”
He said, “They’ll like their new jobs.”
I said, “Will they?”
He goes, “They don’t have a choice. It’s called progress. If they like putting glue on glass, they can work at some other place.”
I said, “They really can’t, because it’s the only show in town.”
He said, “Well, maybe they have to change.”
I thought, “They don’t want to change.”
And that’s when he said, “Mike, we all have to change.”

It’s change, man. We have to embrace it. It’s going to happen. Either you challenge it—and you’re not going to win, because you can’t win against time—or you embrace it and figure out how to make change work to your advantage.

You didn’t want to stay in sixth grade all your life, did you? You had to go to seventh grade. Some people want to stay in college forever, but eventually you have to graduate. You have to change. You have to develop. You have to grow.

TW: Part of what your comment says is that the skilled trades are actually fairly immune to a lot of this, because the skilled trades don’t have a whole lot of repetitive tasks. There is some busy work, but the percentage of repetitive tasks is lower in the skilled trades.

MH: And to that point, anytime I train millwrights, mechanics, and technicians, I say, do you guys enjoy repairing the same machine the same way week after week? And all of them say no. I say, would you guys be more interested in having varied work that’s challenging, that employs things like diagnostics, things like problem-solving, things like really sophisticated repair, as opposed to the same old plug-and-chug bearing? They say, yeah, absolutely. That’s the thing they enjoy most about their jobs—the one-offs, the challenges that we didn’t seeing what’s coming, not understanding what’s going to happen next. Yeah, that makes it interesting.

Listen, you know, changing a tire every day is boring. And they don’t like that. And these guys—and some women too, mostly men—but they want something challenging, something better. And that’s why I think this opens up the opportunity to reduce the repetitive crap and say, let’s get better. Because there’s something we know we can always do: we can always make everything we have better than the way it is now. But maintainability and reliability are at the forefront of that. That’s the thing that makes it happen. If you’re not maintaining, if you don’t have a reliability mindset, your projects, your products, your services are going to fail.

So why not embrace tools that are going to help you get there? You know, I guess you could find somebody who makes custom furniture and the only thing he or she uses is a knife and a hammer. Okay, it’d probably be one heck of a piece of furniture. But, you know, I might want to go to the person who’s got a pretty good workshop laid out and can really build me a nice piece, and who employs the latest tech. I mean, that would be interesting, I feel.

TW: No, it’s almost a paradox, where the more AI can reduce these repetitive tasks, the more people will be challenged to expand what they know by doing these one-off, non-common jobs.

MH: You know, there’s T.S. Eliot, and I sometimes like to quote these various thinkers, and T.S. was always one of my favorites. He wrote this one thing—it was a eulogy for one of his friends—and it’s a long eulogy, so I won’t say the whole thing. But it ends with:

“They constantly try to escape from the darkness outside and within
By dreaming of systems so perfect, that no one will need to be good.
But the man that is, shadows the man that pretends to be.”

I love that quote, because that speaks to exactly where we’re at right now.

TW: With that, let’s close on that thought. I’ll find the T.S. Eliot poem (note: "The Rock") and link it in the notes too, for people who want to read the whole thing.

MH: There you go. I think I’ve got it verbatim. I’m pretty sure I did. I’m usually pretty good at that kind of thing.

TW: Mike, thank you so much for being on the podcast today. For people interested in more of Mike’s work, you can check out his websites and also his articles on Plant Services. Thanks again, Mike.

MH: Thank you.