Where does AI actually belong in oilfields and IT?
You are under pressure to put AI into operations, but in an oilfield or an IT and SCADA environment the wrong deployment can weaken safety or replace a working system for no gain. Rick Carlisle’s answer is to use AI where it adds speed and another layer of safety, not where it substitutes for proven physical work or existing safeguards. Carlisle is the guest on The Artificial Intelligence Podcast.
His framing starts with the word itself: “We throw the term ‘artificial’ on it, but the biology hasn’t changed.” The intelligence is not new, and neither are the physics, chemistry or hydraulics that decide whether a tank overflows, a pump fails or a pipeline is sized correctly. What is new is speed. Work that took a week can now be done in seconds. That is the A part. The I part is still the physics.
Can AI replace field workers or safety systems?
No. Carlisle’s position is that you still need a human in the loop, especially where physical work and safety decisions meet. AI can help optimize how a rig is run, but it will not run the rig. It will not turn a wrench, fix plumbing, change a fuse, wire a panel, put a meter on something, change a packer on a rod pump, or put a transformer and pole in the ground. Someone still has to stand on the platform and handle the pipe.
The host’s formulation is: “I’d rather use AI to add more layers of safety than to replace the existing layers.” That is the test for any deployment. If the project trades away an existing safety layer for efficiency, it is pointed the wrong way. If it lets a worker stay off an icy road at three in the morning because the data can be checked remotely, it is pointed the right way.
What does AI catch in alarm and monitoring data?
It catches repeated patterns and trends that humans miss when they are reviewing logs one morning at a time. Carlisle’s routine was to look at historical alarm logs every day and ask why the same alarm kept appearing. A Power BI app counted how often each alarm fired over the last week, month or six months. Drilling into the data showed whether six alarms landed on one day or one or two arrived every other day. Then the I part could be applied: a valve was sticking, a tank was filling while a pump was not pumping down, or a pump needed maintenance.
AI can watch that pattern 24 hours a day. The same logic applies to pump trends, flow rates, pressure and tank level slopes. If a tank normally rises a foot an hour and the pump is running below historical flow, the model can flag it before an operator is called out. The physical fix still belongs to a person.
Why are data centers reviving nuclear power?
Because AI data centers need enormous amounts of power, and the grid and pipeline constraints are forcing a rethink. Carlisle points to West Texas becoming a data center hub because the oil and natural gas are there. Gas that would otherwise be sold overseas can be used on site, with a power plant built next to it. The transmission buildout is part of the same shift.
Nuclear is part of that shift. Microsoft is paying to bring reactors at Three Mile Island back online for its data centers. That would have been a much harder public argument a few years ago. The demand from AI has changed the conversation faster than most people expected.
What to do before you put AI into an oilfield or IT operation
Start with where it helps and where it makes the job easier. Carlisle’s first question for anyone running IT or SCADA operations is exactly that. The A part is faster, faster, faster. If a use case does not speed up a calculation or reduce a risk, it is not the first place to deploy.
Put safety ahead of efficiency. Use AI to add layers of safety, not to replace existing layers. Remote monitoring that keeps a worker off a dangerous road at three in the morning is a stronger first project than trimming a headcount number.
Feed it historical alarm and monitoring data. Build the simple view first if you need to, as Carlisle did with a Power BI app that counted repeat alarms and let him drill into the trend. Then let AI watch the same data continuously and flag the valve, pump or tank pattern before it becomes a callout.
Keep a human in the loop for physical work and final judgment. AI can recommend pump maintenance or tell an operator the tank has room until morning. It will not change the packer, wire the panel or fix the plumbing.
Use physics and hydraulics models for sizing decisions. When a smaller operator plans wells, an AI model with physics and hydraulics built in can answer how big a battery or pipeline needs to be for expected production. The physics is still physics. AI just gets to the answer faster.
The point is not to make AI the operator. It is to make the operator faster, safer and better informed while the physical work stays with people.
Original show notes
What does AI actually change in heavy industry? Jonathan Green talks with Rick Carlisle — a lifelong technologist who started coding in COBOL — about AI in IT and the oil fields: using AI to add layers of safety rather than replace them, catching patterns in alarm and monitoring data, and why the data-center boom is even reviving nuclear power.
Key Takeaways: • Use AI to add layers of safety — don't trade existing safety for efficiency • AI shines at spotting patterns in alarm/monitoring data that humans miss • The data-center power demand is reviving nuclear • AI's real value is augmentation and speed, not replacing proven systems • "Artificial" is a label — the underlying reality (and the safety stakes) hasn't changed
Notable Quotes: "We throw the term 'artificial' on it, but the biology hasn't changed." — Rick Carlisle "I'd rather use AI to add more layers of safety than to replace the existing layers." — Jonathan Green
Connect with Rick Carlisle: LinkedIn: https://www.linkedin.com/in/rickcde
Enjoyed this? Follow The Artificial Intelligence Podcast and share it with someone in IT or heavy industry.
Auto-generated transcript, 5,153 words. Timestamps link to the moment in the episode.
So, IT in the world of AI and oil fields with today's amazing special guest, Rick Carlisle. Now, Rick, I'm excited to have you here because you've basically were born into technology and coding and understanding how things are built from COBOL. Even before then, you started playing around with programming as early as five. Can you tell me a little bit about that journey and maybe how things have changed across your journey, how people have approached programming differently in technology from when you started to now?
Sure, so my father was one of the founders of the computer industry. He sold the first mainframe to the Census Bureau, helped create Fortran, Cobol. He used to laugh about the fact that when he was teaching Fortran to these kids, they would call him professor. ~ So I mean, Five years old, he's teaching me Fortran. I wrote this magnificent program. It printed out the multiplication table. So I've been, before it was ever called IT, I was in it. And then as the years progressed,
you went from, writing down the code line by line on a coding paper. And then you go punch it into the punch cards. And then you take it and you load the deck and you run your program. And we got into Cobol with Grace Hopper. Now we actually write things in English language or whatever. And then it gets compiled in the machine code. Now it's not punch cards anymore. And then Pasco and Basic and C and C++ and Python and whatever all is out there
now at this point. I mean, the things that you used to have to code by hand, like write this into this register, you know, this address, write this to this port, you know, you want to send something out to the printer. Well, ~ stream of bytes to this particular port address, you know, then we had DLLs then .NET then, you know, now wherever we're at now, you know, now you've got AI that, that, you know, code this for me and there it goes.
It's, you know, I say from my perspective, I'm looking at from the first mainframes that were built and now my mobile phone has more power than the lunar lander. Do you think that the way we approach technology has really changed dramatically in last few years? that AI, what we call AI is here, there's this kind of big excitement and this hype around what these tools can do, but they mostly solve problems that we've already been solving just in a different way.
Yeah, absolutely. I'm not a big fan of the term artificial intelligence. It's not artificial. It's the intelligence we've had for decades that, know, physics hasn't changed, chemistry hasn't changed, biology hasn't changed. We throw the term artificial on it. All that we're doing now is doing the things that we've done for decades. We just do it faster, you know. what could take you a week to figure out, debug your program. you whatever it is, now it's done in seconds. But the underlying intelligence,
know, what's here, hasn't changed at all. Yeah, kind of seems like we're trying to outsource our intelligence. It's like, let the AI make do decisions for me. I don't read maps that let the GPS tell me where to go. And we're starting to see the shift where people use AI to solve problems that we used to just ask someone or do by ourselves. Now we're asking AI simple math problems or directions or which restaurant to go Like these are problems I thought were already solved,
like asking for a recipe. We've had cookbooks for, it feels like a thousands of years. ~ but I'm really curious about as well is like, you've worked on some really big projects and handle and been in charge of really large oil fields and had a lot of these experiences. Have you seen do large oil companies and gas companies, are they slow or fast to adapt to new technology? Are they always chasing the new next thing? Like a lot of people seem to be these days.
It all depends on the size of the company. mean, the bigger the company is, the more inertia there is in all the infrastructure, know, and I mean, internally employee infrastructure, right? You've got reservoir engineers, you've got geologists, you've got performance engineers that are looking to how big do I need to build the facility? Well, the geologist and the reservoir engineers tell me, Hey, if I put a hole in the ground here, I'm going to get this much fluid out of it.
I got to put that fluid somewhere. How big does the tank need to be? How big does the pipeline need to be? You know, and then you've got all the folks behind that. The, the management teams, the engineers, the automation analysts. All those people are there, you know, they're writing the PLC programs. They don't particularly want to have their job go away. Right. Then you've got the smaller folks that don't have that inertia built in. They'll readily pick up,
you know, Hey, it takes me a week to figure this stuff out. You know, this much fluid comes from this part of the field. How big a tank, all that stuff. For them, they don't have an entire geology department, an entire reservoir engineer department. They may bring in consultants or whatever, but they're the ones that are quick to pick it up and say, hey, don't, know, physics is physics. If there's an AI model with physics and hydraulics built into it,
and I can say, hey, I expect to produce 5,000 barrels a day, you know, from these three wells, how big does my battery need to be? How big do my pipes need to be? The physics is all there. So they're the folks that are picking up AI. I recently attended a conference and one of the things they talked about was simply asset management. Where are the trucks? service companies that say, have a customer that needs a vacuum truck or
it needs a pump truck or needs a load of needs a load of caliche to build a road. before AI and it was one of the interesting applications that I saw. A guy in the field gets on his phone. He calls the local rep. The local rep gets on his phone. He puts in a ticket. It goes into AI. It says, I have a truck three miles down the road. I can get it to there, get it to you in 10 minutes.
And you click on the button and you say, hey, works for me. I agree. And then a work ticket goes out to the truck. The dispatcher gets a note to say, we're dispatching the truck. The invoice gets processed and there were no people involved other than I need a truck. Hey, we got a truck. Truck will be there. Invoice is going to go out. So these like these small service firms, even some of the larger service firms at this point,
you know, it's like, hey, all that happened with a guy on his phone took 10 minutes. How long would that have taken before? Yeah, so it's really good the way it can help with agility or kind of repetitive processes. But how do you feel when it comes to things like of oil fields or security? Like an AI really fill most roles right now where it can't handle the skater and watch the fields and it's going to detect if something's off track or do
we still need human in the loop? No. you still need human in a loop. There are certainly applications in one that I advocate for is you have historical alarm logs, historical data. And this is like one of my basic tasks at Joe Mill is every morning. Look at the alarm logs. Hey, I'm getting this same alarm over and over again. ~ we need to get the maintenance folks out there. Right. We recently, you know, what, what I will call pseudo AI,
we built a power BI app. So where I would go through the alarm logs and I'm going to say, Hey, you I've seen this alarm too many times. Now I pull up my Power BI app and it says, here's how many times you've got this alarm in the last week, last month, last six months. I can drill down and see, okay, did I get six of them on one day and then I only get one every other couple of weeks?
Or, you know, am I getting one or two every other day? And then I can look at that and apply the I part. You know, and say, Hey, op supervisor, you know, construction rep guy. This valve is sticking. I'm getting high level alarms on a vessel or a tank. The pump is running. It's not pumping down the way it is. pull up the historical trend. look at the slope with the trend. AI could certainly do that, feed it the data.
And again, back to the basic parts of the eye, physics is physics, right? I mean, if the tank is filling up, but the production has an increase, then there's something going on. The pump's not pumping like it used to. We need pump maintenance. Those are certainly areas where AI could... could do, you know, what, what used to take me a few hours now with the power BI app takes me 15 minutes where AI could look at that 24 hours a day and say,
It's my eye going into the A. Yeah, we're noticing that now we have this sudden massive need for more power, more power, more power for all of these AI data centers and AI here and AI there. Do you think that's going to lead to a shift in the perception of the oil industry and that this is going to be a long term shift that data centers are the future? Absolutely. It's happening now. West Texas is becoming a data center hub because we have all this oil,
we have all this natural gas. Yeah, we'd like to sell it on the open market, know, sell LNG, send it off to Europe or Asia or whatever. But we're pipeline constrained in a lot of cases. But if you put a data center and you build a power plant and you take that gas straight away, you know, and all the electrical infrastructure that goes with it. Encore just announced they're bringing in what I think know what they call it ultra high voltage like 760 kV
or something huge huge power lines because of the data centers it's not going to go away I mean you know it's like when the automobile came out know stables for horses weren't the problem. Gas stations were the problem. You know, and again, you know, we talked about the migration from technology to technology, you know, the horse and buggy to the automobile to the, you know, the dirt road was fine for the horse and buggy. You need a paved road for the automobile.
Well, now we have trucking, you know, now we need freeways, super highways. that analogy of transportation isn't any different for the energy markets. I mean, we get it out of the ground, you know, it has to go somewhere for processing. there's pipelines, the, you know, you need a power plant, need electrical transmission lines. In my mind, it's the same analogy, right? You went from horse, you know, And now look what we got, know, Katy Freeway in Houston is 12 lanes wide,
16 lane wide, something like that. Same thing with the electrical transmission. We have to put more down, more has to go through to support the new technology. So a lot of people don't really understand what it takes to keep an oil field running. What type of technology, what type of workers, kind of ~ processes are, what is the AI hype crowd really missing about how the industry actually operates day to day? I think there's a lot of hype about...
certain kinds of analysis that you can do. I mean, I gave the example of an alarm analysis. I think they're missing the bigger picture. The eye part, you know, you can't impart, you know, what my 40 plus years of experience into a model. You can't, you can't run a rig. I mean, you can't turn a wrench with AI. You still have to have guys putting holes in the ground. Yes, you can optimize how they put the holes in the ground. with AI, but there's still somebody
standing on the rig platform throwing the, you know, throwing the chain around the pipe. You can't AI that you can't use AI to wire up a panel. You can't use AI to put a meter on something. You can't use AI to change the packer on a rod pump. I think micro is real big on all this. know, at the end of the day, AI won't turn a wrench. It won't fix the plumbing. won't change a fuse. Maybe it can help
you determine how big a transformer should be quicker than normal, but it can't put the transformer in. It can't put the pole in to run the electrical lines. There's a huge part of AI, you know, that you simply can't apply it. So for the most part, as far as keeping an oil field running, there's not a lot of application for AI other than, like I said before, speed up some things that used to take longer to figure out. But somebody at three o'clock in the
morning gets to call out and gets in their truck. You know, one of the things I talk about with skater is does the guy need to get in the truck? If he can turn out his laptop at home and he can get on skater and he can see the data and he can see what's going on. Hey, my pump's not running. My tank's too high. It's three o'clock in the morning. If I got room in the tank till I can get there at six,
you know, there's just no application for AI. in that part of the auto build. You think we're going to experience like a generational gap in a lot of other industries? Well, absolutely. There's no doubt about it. When I was at Chevron, you know, one of the things that I was, mean, I was a mentor for a lot of the young engineers coming up. And it's not, it's not that they don't know a lot. It's they don't know how to approach problems.
And as folks like me age out, there's a gap in the middle. There's a ~ lot of... You can call it mid-level. And it's not their fault, but they weren't exposed to some of the stuff. They were put in the pipeline. my first job as a manager, then I move up to management, then I move up to management and you're versed in the processes and the procedures of the corporation. But the guys out in the field, there aren't enough new people coming in to pick
up where Bob's instance calls us as gray hairs. You keep the gray hairs for as long as you possibly can keep the gray hairs because that point in the middle. There were a number of years where oil and gas was frowned upon. We didn't get a lot of college graduates 10 years ago that wanted to go into oil and gas. We've got a lot of them now, and there's a lot of us old timers, but that part in the middle is missing.
Yeah, I'm seeing it. think that when I was in high school, I'm in my forties now. Everyone was like, you have to go to college. You have to go to college. There's nothing else. And a lot of the trades kind of got ignored. And that's why we don't have enough plumbers, electricians, builders, all these jobs kind of got looked down on. And I'm not really sure why now it's, it's kind of shifted back again, because it's like, wait, we need infrastructure. things don't build themselves.
AI can't build its own data centers. So there is exactly like you're talking about this shift. And I'm really, I do know we've kind of gone from a really anti-energy. We kind of thought everything would be solar by now to suddenly nuclear is popular again because these data centers need so much power. And we're kind of seeing what happens when we don't have enough internal oil resources. start to, we have to import and things happen. Do you think that this mindset will continue or people will
turn against oil again in the short term or in the long term. of course. I mean, you know, you can't call me a climate denier. I mean, the climate is changing. Whether or not that's entirely due because we were burning fossil fuels or it's due because of the per-annual, know, whatever solar, whatever solar period there is. I mean, we've been hotter, we've been colder. I'm not a big fan of the, my God, that's the warmest year on record. Well, how far does the record go back?
You know, there's been around quite a while. It was a lot hotter earlier. It was a lot colder later. But the bottom line is you go back to before fossil fuels. What was civilization like? How did you heat your house? Where did the work day come from? Well, because it got dark at night, we had to stop. You know, it's simply progressed from there. Energy density is the, the, what drove fossil fuels. I can go chop down a tree and I can put
the wood into the fireplace and it'll heat my home. But it's a lot of wood and it's a lot of trees. I can throw a barrel of oil in there, you know, a few thousand cubic feet of natural gas. I don't have to chop down any trees. you without the energy density, you don't get the factories. ~ you know, you don't get the fact that you walk to your, you you go into a room and you flip the switch on and the light turns on.
Well, that's cool. But how big does a solar farm have to be to provide the same amount of energy as a single gas fired? gas-fired turbine in a power plant. to how big the solar panels do. How many wind farms do you need? And when it gets dark, need batteries. When the wind doesn't blow, you need batteries. Every bit of renewable energy, I don't like to call it renewable, every bit of alternative energy we can get is a good thing.
But it's just not, it's just not realistic to think that demand for oil and gas is going to go down anytime soon. I think the last prediction I saw was it's going to continue to increase until at least 2050. You know, so. I might've gotten a little off topic there, but it's kind of one of my pet peeves that, know, hey, leave it in the ground. We don't need it. We don't need it. Like, yeah, okay. Fine. You know, that rubber puts soul on your shoe.
~ you know, where do you think that comes from? Yeah, so I totally understand where you're coming from. I think that just in my lifetime, we've seen my whole life, everyone's been like, no more nuclear power. It's the worst. we need it for AI. Nuclear power is cool again. It's the fastest I've ever seen a shift in a perspective about something. And ~ I've just always been so curious about that because I was born 1980 right after we stopped at the nuclear power plants.
And I always thought Three Mile Island was this huge disaster. like Chernobyl. And then I looked it up recently. I was like, am I missing something? Why do they always call it incident? And the reason they have to call it incidents because nobody died and nobody even got hurt. It's like zero injuries is an incident. Well, yeah. That's wild. And now suddenly people are rethinking it and it's like so much of its perception. Well, you know, and also in that regard,
I mean, I was a nuclear reactor operator in the Navy. So. I come at it from a different perspective. be a reactor operator in the Navy, the training you go through is equivalent to, ~ you could call it an associates. It gets close to being a bachelor's degree in about a year's worth of training. But we learn it all from the physics, from the ground up. This is what happens at the atomic level. This is how much energy is released.
This is how much radiation is released. This is, if you put, if you take, if you have beta radiation and you take a piece of paper, you block it. If you have alpha radiation, you put on some clothes and you block it. You know, it takes so much steel, so much water, so much of this, so much that, so much thickness and you cut it down. And then, you know, uh, we were doing maintenance in the shipyard and I was laying. Literally.
Lay on top of the reactor core. Vixen about position indicator. And I got less radiation than I would with the chest X ray. Or if I fly a lot. Three Mile Island, we know exactly what happened. I was briefed completely on the sequence of events, everything that happened there. Things that haven't been released to the public that I can't speak to, but nothing happened outside. Look at Fukushima. know, people are moving back. And that was an incredible disaster. Chernobyl was poor design,
bad decision making. They ignored warnings and alarms and said, ~ we're just gonna keep doing our testing here. But what was the end result? What's the result today? know, it was bad then, but nobody builds Chernobyl reactors anymore. Well, Russia has a few still out there, but you know, we know how to build them. We know how to run them. We learned lessons, you know, from three mile line. And look, Microsoft is just paying. to bring the reactors three mile on them
back online for their data centers. Yeah, it seems like suddenly, now that we need more power, all the things that were naughty before are okay again. And are you seeing, when you post about artificial intelligence on LinkedIn and talk about it more and more, what kind of perception or reception do you get from people inside your industry? It's in my industry, not so much. Again, for the reasons we talked about, what I do get feedback in is the general applications of AI. The large language models,
they're great for search engines. You know, mean, I like when I go to Google, you know, and I say, you know, ask a question, I get a nice, you know, nice language model answer. tells me, hey, if you want to do some more research, go look over here. You know, I think that's fantastic. But in my mind and the feedback I typically get is about agentic models. You know, what you can do with an agent. One of my posts was addressed to,
for example, detecting breast cancer, right? So to detect breast cancer, you get an X-ray or a sonogram or whatever it is, but you get an image, right? And it's just an image. It's pixels with characteristics. What's the gray scale? Maybe it's color and you have RGB associated with each. an RGB number associated with each pixel, but it's just an image, right? But you can feed those images thousands upon thousands into an agentic model and say, Hey, this group of pixels with these characteristics,
these are breast cancer. These are not. So you can have an agentic model that says that you feed an image in and it scans the whole model and it looks for those particular characteristics of maybe it's only four or five pixels square with certain grayscale factors or RGB numbers associated with them. But it says, well, I've been told that's breast cancer. So that model, that agentic agent says, hey, I think I detected breast cancer. It doesn't know what to do after that.
So again, the, the I part, right? It can scan all these things almost instantaneously. Right. And then it can kick it up to another agent. The other agent says, ~ I know what to do. If we think there's breast cancer schedule, blood tests, you know, do more testing, maybe a biopsy, you know, whatever. So that agent says, I recommend these things happen. And then those things happen and then you feed it to the next level, right? You feed it to the level of the
cancer specialist who says, okay, now I have the result of these tests. have the image. And then that one says, well, I recommend this course of treatment. Let's do surgery and remove the tissue. Let's do chemotherapy. Let's do whatever the therapeutic recommendations are. So you have these hierarchies of agentic agents. that each is specialized at what it does. And at the end of the day, once you feed it to the chain, you get a diagnosis, you get a doctor that looks at it all and says,
well, yeah, this all looks good. I agree with the recommendation or I don't. That's to me is where the power of AI is. Yeah, that's amazing. I think that it's going to be an exciting future. And that kind of leads me to my final question, which is like, for listeners running an IT or skater operations at an oil and gas company, and they're trying to figure out where AI belongs in their stack, what's the first question they should be asking? that's a very good question.
I think the first question I should be asking is where does it help? Where does it make my job easier? Again, back to the A part. The A part is faster, faster, faster. If, if you can, if there could be an agent built into skater that did exactly what I was talking about before, right? I've got to, I got called out. got a high tank level. So just three o'clock in the morning, the roads out, it's the middle of winter.
Is it safer for me to drive the road and save the plant or am I going to get killed trying to get out there in the first place? an agent that said, you know, AI that said, let me, let me look at the pump trend. Okay. The pump is running flow rates way below where it should be historically. Pressure is good, flow rate's bad. Let me go analyze the trend, look at the slope. What's our normal slope look like on the tank level when
the pump is running? AI could certainly do that. And then tell the operator, you your pump is bad and then you feed it a little more information in the model. Well, how long, what's the normal fill rate for the tank? Does it go up foot an hour and I have a 24 foot tank and I'm at 16 feet? Well, shoot that, that's a problem. know, it's a 24 foot tank and I'm at two feet. ~ well, that I probably don't need to get
out on the road three o'clock in the morning. So that's the I part. So the a part easily, you could build that model and you can have that model run, ~ maybe intercept the call out ahead of time, not call out unless there's a reason that somebody needs to get on the road. That's where I see AI being integrated with SCADA process control. It's. When I make my presentations, the first thing I talk about with Skata is safety. If you could look at something remotely
and not have to drive out there and take a look at it, you're safer. So anything where the application of AI makes things safer, think that's the first priority to the extent that AI can speed up things. Like I talked about before, we're going to put six wells in. How many batteries do we need? How much, how big does the pipeline need to be? Well, for the first well, we only need this, but for the sixth well or the 10th well,
we need this. Well, you can't just change out a pipeline. You know, if the pipeline size for 10 wells, when that first well runs, maybe we change the characteristics of the pump. Maybe we build that into the plan. When I put the 10th well in, we upsize the pump. Those are the kinds of things that a person needs to do the calculations. Okay. Calculate it for one pump and calculate it for three wells, six wells, eight wells, throw that into a physics model with AI.
bang here you go you need two seven hundred and fifty barrel tanks today gonna need six you know you're gonna need six and twelve weeks but your pipes gonna have to be twelve inches in order to handle when you finally get to the end of the line That's where I see AI in the oil and gas industry. Like I said, it's not so much the I part. The I part is simple. It's physics. It's hydraulics. You know, the A part is the speed.
Hopefully that makes sense. It does. think that's really helpful for a lot of people out there trying to figure out what they can and cannot do and kind of how to get into this world. And I love that you brought up safety first, because a lot of times it feels like we want to trade safety for efficiency. I definitely agree with you that I'd rather use AI to add more layers of safety than to replace the existing layers. So this has been absolutely amazing,
Rick. I appreciate you giving us so much of your time for today's amazing episode of the Artificial Intelligence Podcast. Thank you so much for being here. She's quite welcome.
More on ai in manufacturing, energy and infrastructure
