Oil and Gas This Week | December 5 2025 | Ep 393

This week podcast. LeCore and Paige Wilson. This is the Who quickly? All right, you're listening to the Only Gas this week podcast brought to you by Cognite, AI for the industry. Optimize production and decrease operational risk with trustworthy AI across the oil field. This is the show for busy old pros who want to quickly keep their finger on the pulse of the industry. Thanks for joining us for episode three hundred and ninety-three.

We are sitting here at the shack in the back, we're outside of Odessa, Texas. If you're listening to us and you don't know where that is, it's right smack in the middle of the Permian. We have an all-star panel. We're getting ready to have some really cool discussions. And audience, by the way, and when I say audience, I mean the ones that are here live, not the one that's listened to this later remotely.

I want this to be interactive. If you have any questions, raise your hand. Paige will walk over to you with the microphone. You can engage with our panelists here. But I want to start.

With literary the definition panelists. When somebody says AI, what is AI? In which country? I say that'cause in Latin America and Argentina it's EI, right? So it's kinda interesting. But artificial intelligence, gosh, what would we think? The Skynet's been built and here comes the age of the robot and and taking over the world and The reality is that's not the case right now. Artificial intelligence is still grounded in

data foundations, exploratory, how we actually look at changing advancements in data and just in science in general and how that would actually apply in operations. I would say it's still very much um marketing, but at the same time it is advancing so quickly. that if we don't get s the ground and the foundation understood along with the data governance around that, it would impact our operations into some aspects that we're not prepared for.

AI to me is the growth, natural growth and development of what's been going on for what, twenty plus years. We're looking at how we're looking at our data, aggregation and development. So AI to me is not I it's not the artificial intelligence intelligence of our childhood.

But it is a growth and development process that's really been astounding. I would say two years ago, you if you talked to me about AI, I'd say, okay, that's a child's toy. Now AI is a serious tool that we're using in our daily lives. You're seeing people writing essays, you're seeing people writing code, you're seeing people doing advanced mathematics now. AI itself has become uh outside of its own definition, is what I would say.

Yeah, I'd say we've really moved a long way from the early days of just machine learning, where you were looking at building somewhat simplistic but high fidelity models based on very pointed data sets. And now we're looking at generative AI and we're looking at these much larger neural networks. And so what AI means today

Isn't really what it meant five years ago. And the value that we get from AI today looks nothing like it used to five years ago. Very exciting space. And it's a great question that I think if you ask a hundred people, you get two hundred definitions. And if you ask in a year, you'll get two hundred more. Yeah.

Yeah, definitely is evolving. I think it is way different than what we used to think of it. And again, it's just a really a a really strong tool for us to use. And I think the thing is that people think It's separate from the uh the person itself, but in a sense it is a tool for us to use. It's not gonna replace

really any of us. I mean it does replace certain aspects, but again, it does need the human aspect. And take in the past we thought of it as a s a scary thing to look at. But I think the the more we dive into it, the more we utilize it, the more we realize how much of a powerful tool it really is. So to me personally, AI is just another means of

developing models so that we can gain insights or make predictions, right? When I first started my career, I was developing data, empirical data and models, right? Just based on data. And then I migrated to physics based data, w whether C F D or finite element or whatever physics and then came along is AI was to me is just another way of developing models that gives you insights into the data that you have and give you the ability to make predictions.

And hopefully they're validated predictions. Yeah. The reason everybody's here and the reason our audience is interested in this, let's get to the root of the matter. How does AI make oil field operations better? I think the ability to take these large, massive data sets that we're producing constantly and turn them into value is the biggest thing that we're seeing right now. We're seeing production data sets.

being consumed and organized in ways that we've never seen before. I can say personally at Conico Phillips, we're taking s unstructured data sets and creating real value out of those. Where before we have these terabytes and terabytes of data built up and we didn't know what to do with them. I think this is giving us a path forward into approaching that. Before anybody else answers, I do want to make a quick comment on that'cause I've touched that world for a long time.

Unstructured data in our industry has been something from the very beginning. Even when it was paper mudlogs, right, in the forties and thirties, it's still unstructured data. And we've never gotten our hands around it. We tell others that we have, but we never have. And that's one of the things I'm seeing that I'm just finding amazing is finally we can take all this disparage data that's in different silos and different forms.

and actually put it together in a meaningful way. I think we're in our infancy with that. I think it's super powerful. And that's the I think the most interesting development right now is tying all those disparate data sets. If it's the equipment on site or if it's the wells nearby or if it's the benchmark, if you're getting all the wells at the same benchmark, being tie able to tie those different disparate data sets

into a meaningful theme to or to a story, to me is the most powerful and interesting thing that's going on today. Come on, I know the rest y'all have input on this one. This is a big question. I can go. So look, I think absolutely uh the ability to make better decisions because you're making decisions on more data. you're also going to be able to make quicker decisions, right? And time is of the essence in a lot of the decisions we're trying to make.

If you've got downtime, if you've got wells that are underperforming, every day that you delay making that decision and the lower the confidence in that decision, those are dollars that are wasted. But as well when you look at AI and the ability to deploy AI. So when you start talking about AI on the edge. You start talking about closed loop control and automation using AI, then you're doing things even with a human in the loop, you could never do, right?

micro adjustments and set point optimizations on the order of hundreds of times a day that even if you paid someone, they wouldn't be able or willing to make those kinds of optimizations. And so you're talking about better decisions, faster decisions, and the ability to act on those decisions in ways we never could in the past. I 100% agree. Absolutely. If you look at

the especially in the Permian Basin, right? We're sitting right here. If you look at a rule of pump by exception, right? With the amount of workover wells that we have out here, what type of AI route optimization tool could you use? What about AI in your well planning, just from a value in the time you spend doing that with external resource searching. There's a lot. However, I would still say that it still is gonna come down into the connection within the operational field, right?

I I hear it all the time and I I have people who are smarter than me in this. You can sit here and you can have a large language model and it could be sourcing and you could have an agent come in and say, figure this out for me, but are they gonna really understand that go look at the valve and tell me if it's open and closed? Are they really gonna stand understand whether or not it's open or it's closed? Do they understand the relationship of the data?

to the specific asset itself. And I think it's that element that only enhances artificial intelligence, to your point, right? We were talking about unstructured data, where that takes us, and then the value that ultimately outputs. I think currently a lot of the things that it's really improving on is being able to have a better understanding of what we're seeing or the things that we can't

necessarily see on our own. Or the things that we in the past have had to in a sense guess at. We always had in the past that the reservoir engineer or the geologist that was really had a good sense of what was going on, but they had so much of an understanding of it, but they can only see so much.

And I think a lot of things that AI is filling in the gap And some of that to where bringing more certainty in and uh and being able to com uh compute things that we couldn't do in the past, being able to try to understand things a little bit better, be give a better microscope in a sense to the scientific part of it that is our lives. So not just actually just producing the oil, but everything else that goes behind that.

And also one of the other aspects in in terms of like safety and things like that. being able to predict things that were gonna happen, like failures and or anything that was gonna happen from there and improve on those aspects as well, because we know one of the biggest concerns we have is safety and being able to see things that we wouldn't normally see as individuals in a better scope. I think that's one of the biggest things that's it's impacting right now. Yeah. So for me I remember

My journey started within AI machine learning in 2017 with G Oil and Gas. And so I was sitting in a meeting with the Chief Technology Officer for G Oil and Gas. And he basically said it, hey, roughly ninety percent of all our data is dark data. I'm saying, what is dark data? And then I finally realized that really we were collecting all this data, but we didn't know what to do with it, right? We have really no means of taking this vast amounts of data and and trying to bring some insights that

Would have an operational impact, performance impact, business impact. So for me, the ability to endow with Time series data, all this other data that is not historical, things that we're generating in real time. To be able to analyze this and make some insights and make some decisions, this is really the only way that I know of that we can actually take all this data that was dark. and show some light. And so that's where I believe the importance of AI.

Yeah, and I'll tell you something really interesting. So we have a predominantly an upstream crew here. I just recently met with a manager of one of the major refiners in the US And during a turnaround, which is when they have planned maintenance at a refinery.

They've reduced their lost time incidents with their contractors over 10% by using AI just to route the people and the traffic properly. No human could figure out what was the safest way for all these contractors to come in and leave the job site, but the AI could do it.

Such a simple thing, but such a huge impact. All right. So the next thing I want to talk about is let's just be open here. We're an industry that's very risk-adverse. We don't like new stuff because you introduce something new, it's a risk.

If we've done something on paper for 30 years and nothing's leaked, nobody's gotten hurt, nothing's blown up, we don't want to change that process. However, AI has come in. I would venture to say I've never seen, I've been in this industry over 25 years, I've never seen a new technology adopt it this

Which means people see the benefits from it. But from a culture point of view, I'm sure all of y'all have a story about struggling to get somebody to understand the culture of change and bringing AI in. Can we talk a little bit about that? Because Some people in the audience here work for smaller companies and they're probably wanting to get some AI projects going, but of course somebody says, we've never done that before. Can we talk a little bit about that?

I think the favorite term in this industry is fast follower, right? Everybody wants to be the second or the third person to do it. Once we see you do it well, we'll then we'll go ahead and we'll do it. I think AI has broke the mold in that and that it's popular and it's trendy. And it's because it is valuable, because it's so quickly and easily showing value.

So, my own personal story on that would be: I've been forced or not forced, but been pushed towards recently. Hey, where can we fit AI into our model? What where can we do it? And you're like, As you said, things are working. We don't really want to change things. But then you use it for the first time. Maybe use it in an email, maybe use it to make a slide, or maybe use it something like.

That wasn't so bad. So then you try it a little more and you do it for yourself and you open yourself up to these AI options. So now we're consuming data and we're building trends and we're seeing things where maybe one I so I think Ronnie mentioned it earlier, one geologist sees something, but two or three other geologists see something different.

You smash those together and you show it. You prove it. You prove value. And I think one of the reasons why it's easily adoptable is because it's easy to bring in. Because these AI, we're not training our own bots. We're not training on most of these come in pre-built, pre-trained, and you throw your data into it and see what comes out.

Now, two or three years ago, yeah, we'd throw the data in, it would come out kind of not so great. This year it's been phenomenal. The difference in last year to this year has been absolutely astounding. You're getting real answers in real time. So I can throw it.

I can show thro throw it a production set and say, show me what's going on here. It'll say this, this well and this well. Looks like they have issues, but if you were to do this or that improvement, I mean it's proving out so fast, it's real time. And it's an easier prop problem to or easier thing to sell because it is so fast and so powerful.

I think it's really astounding what we've seen in the last year. I think one of the biggest things too is I'm sorry, I didn't interrupt is that the I think one of the reasons this was, I guess in a sense easier to adopt is that it built on things that we were already doing in a sense.

We were getting in a sense along the lines of doing basic data science. Like the L L Ms and stuff like that. And things like that. We were in a sense our industry was kinda doing some of this stuff a little bit. And so we were skimming the surface on some of it.

And the other part of it I think that kinda helped us a lot was that other major industries adopted AI so well and showed such a great impact with it. You look at the medical industry, you look at things, even social media and things like that, like Google Maps, everything was so accurate.

that there was no arguing against it. There was no way of looking at it and saying, hey, there's we don't see the value in it because there was it was just right in front of you. And the other aspect was that it was available to us.

we could get our hands on it. The things that they were using AI on were on our phones and things like that. And so day to day we were actually utilizing it and we were using it in our own lives. And so seeing the impact happen like that, it was hard to argue that it was actually valuable.

And and there were so many of us that were getting into it ourselves. You look at like I said before, geologists and other engineers and things like that, even people that were just analysts starting to look at data and saying, we've got a problem with trying to figure out how to do more with this.

We make the data, but we're not really great at using it. And how can we improve on that? And answering those questions were starting to be answered by some of the AI AI. Again, the basis of being machine learning and everything else, but it built on it, some of the things we were already doing. just led us towards that. And I think it was just a natural progression.

We do get challenged a lot though because when we're talking about that's change management, right? And as you guys have been living and breathing it for this whole time. But some things it depends on if we're looking at our operations and we want to adopt AI. the easy things, the daily things first. Right. And we say, yes, we can by connecting your data and the unstructured data to your IT data, which is structured data.

But if we really look at it, still is gonna come down to can you make an impact on the business and the net present value? And that is really gonna be to where you can justify. The need in order to adopt artificial intelligence, right? So you not only will you get your people to buy in on it from an operation standpoint.

But then you'll be able to make an impact on your business overall, right? Whether or not that's just overall production output or throughput or O and M decrease, right? So there's a lot of things, but there is a balance when you start. to get change management and adoption driven by can you fix their everyday life by three hours in the morning? Or then okay, where am I gonna go with this investment and what could I really do with that? And imagine if you could imagine if you could connect.

your IT data, your structured data to your unstructured data. Right. And that's the next realm. Right. And that's really where you're gonna see an impact from your vendor and supply sele selection into any type of asset management programs. And we're talking around it a little bit. You say we're risk averse as an industry. The reality is we're not risk averse. We just are very good at managing risk. I say we're uncertainty averse.

Right. And so then the examples that we're giving here are it's how do you break down the barriers of what is this actually going to mean for me? So you're gonna start where there isn't a lot of risk tied to the outcome of that exploration. So decision making, managing things in the back office, making my morning cup of coffee more effective. And then when you take AI out into the field.

You're not going to start with safety critical elements necessarily. You're not going to start with your prime wells. You'll start maybe where you'll still see a measurable impact. through applied AI, not necessarily where you'll see a measurable negative impact if it doesn't go well. And so to me, risk management is always at the heart of the industry. And so what it's up to us as an industry is to take the uncertainty element.

Move past it through case studies, through diligence, through a structured approach so that we know we're getting the impact that we believe we should be able to get from AI. I'm a borrow your uncertainty phrase. That is a perfect way to describe it. It really is. And this a natural segue, we didn't rehearse this.

Can we talk a little bit about some of y'all's pilot projects, some of the where you first got started? Tell me a little bit about what it was and then how did you convince your company to let you move forward with it?

Yeah. All right. So let me take that. But I do want to answer the other question, right? As a former professor, right, in in terms of oil and gas, I guess the majority of the people were engineers. And so I would say in the very beginning, If you're an engineer, you're very reluctant to adopt AI because throughout four years or six years of university life, if you're an engineer,

You're gonna trust physics. And so if you're gonna come along and say, no, I'm not gonna trust physics, now I'm gonna trust data, you're gonna have to prove it to me. And so I would say the whole boom of AI really started about three, four years ago, where I think everyone's was really starting to say, I think there's something within AI, and I think it gives us the ability

To analyze vast amounts of data, but then at the same time, I can bring some physics into it to help me explain. But that's my take on it. So I would say. In the last three, four years there there's been examples

And of course, ChatGPT was one of them where the benefits of artificial intelligence, machine learning, generative AI came in, where it was actually solving real world problems and bringing some business sense to it. So Now to answer your present question, my first experience, again, I was forced into AI, but I was quite reluctant.

I was reporting to a general manager and he says, and of course at that time I was managing a team of computational scientists. So everything that we were doing were was physics based, whether CFD on element. And he said to me, Hey Ed, I need you to lead this effort in what is called machine learning or analytics. I said, I know nothing about it, but anyway, I learned about it. And so from that, I was able to lead my team.

Into projects where we're actually starting to apply machine learning, not AI, just machine learning, to help us make predictions in terms of asset monitoring and things like that. And once the confidence And the predictions were validated. Then I think the engineers, once they see that your predictions are validated.

then the adoption started and we felt that there was something there. So that was really my first experience with AI. I was really forced into it. I'm really a physics guy, but now I'm a data science guy.

But that comes back into doing the stuff easy, right? Show me the stuff I do every day with AI. So if I'm sitting here, I'm in the field, I've been in the field for 20 plus years or even 10 plus years, right? I know I need to look at that this data, this document, this time series, this temperature sensor. And I need to trust that what the AI is looking at is what I would look at in my everyday life. Absolutely. And that is where that trust and the adoption process really happens. So if we can

build the foundation to where they trust that where the data is coming from, then they'll find the they'll find the value to it. I'm with you though, man. I join I everything we do is AI and I'm like trying to write emails with Jim and I. It's not working so great, but I'm still old school on that.

I would say for us, actually it's management's been pushing AI because it's the easy button, right? Everybody wants to see this easy button, but who pushes back? I think you nailed it as the engineers. Those of us who are buried in the fact.

And the figures and the I'll call it the data because it's a separate fact though. We're looking at actual figures. So we do a lot of forecasting AI right now. And it can becomes a finer and finer razor that we shave it. Like how good does the AI have to be able to replace an engineer? Or do at least

help out an engineer. I think a lot of the mentality is once the AI gets so good, we're out of a job. I don't think that's ever gonna be the case. AI is not here to replace our jobs, but it's to help us do better jobs at what we do. I think really the issue is it's those who are in the ground on the ground doing the work are the ones that push back the most. I think management sees it as.

this golden easy button, which the truth is it's not quite there. I don't know if it ever will get there. It needs I I think you mentioned earlier, it needs a human to book end it. We need that logic philosopher thinker. to say, Hey AI, I want you to solve this set of problems specifically and then we step forward from there step by step. I think that's how these projects end up going.

And have fun with it. I'm gonna step in real quick. Have fun with it. One of the things we just did for a client, and just because we're on the OGGN podcast, is we created a podcast for Shift Change. So it's an a a Gen Tech agent working in the background, looking at all of the workover that happened over the weekend, any failures, any route things that need to be changed, and they're drinking their coffee, driving into work and listening to the podcast on what is their shift change.

How fun is that? How creative is that with AI? I don't know about y'all, but that's one of my current favorite things is using AI for summaries. If you miss part of a meeting or even if you're in the meeting and you haven't quite paid atten attention as much as you should have. AI is great at summarizing or it can summarize code or it can summarize a long story or a book or whatever. I think they're using it on the legal side now to summarize big long dictations or whatever. So

A guy's ability to summarize is phenomenal right now. Yeah, the copilot summaries I use All the time. And we make sure that all the action items go to the people that didn't join the meeting and we say, Hey, AI said that was your job. So not who am I to say no? Absolutely.

No, back to that idea of the philosopher and the editor and you know, how AI can really move the needle for that person with everything in between. So you don't have to be an expert in a lot of the mundane aspects of solving those problems. As long as you can articulate well what the problem is that needs to be solved. And then when you see the output or the data product that comes from AI and you say, okay, that's pretty close, but here's how you get it to the last mile.

Right. And so then that human comes back in the loop, gets it across the finish line, operationalizes it. And then through exploration and a bit of play and having fun with it, it starts to become more ubiquitous with the daily life and then it starts to get a lot more traction.

Going back to like your question about how I got into doing it into AI with oil and gas. I'm gonna try and summarize this as much I can'cause I can really talk long about this. But my background is was just mathematics, pure mathematics. And I actually got into the level industry as a former math teacher over a dozen years ago. And just because I was very analytical and they asked us, you know, what I

got hired on by this amazing reservoir manager who was just really brilliant, but he knew that we had really we were really good at producing data and creating data, but not good at managing it. And he was like, let me bring in somebody that's really good at managing data. And so when I I initially got into it, I realized it quickly that was true.

that the industry and this was again, like I said, over a dozen years ago, we can produce data, we can make data, we couldn't manage the data. And we weren't really great at managing tons of data. And so I knew that eventually there was going to be a kind of revolution when it came to this kind of thing. And for me, I was always trying to break down the barriers of talking across the table and everything else with us talking about what we did with our data and everything else.

And then so when it came I ended up working for concho, they put me on the forefront and said Ronnie, everybody's talking about this was about six years ago. He's like, everybody's talking about doing AI and machine learning. We gotta do AI and machine learning. I said, we're not ready. And they're like, what are you talking about? We can bring people in. We'll hire some group we'll hire some companies and we'll we're gonna do some data science. Let's do data science.

We're not ready for that. And so the thing was is And e my mys I myself only knew a little bit about it, right? And coming into it I was more read as much as I could on it and understood it. But again, just having a mathematics background and not really a heavy computer science taught myself Some coding and things like that. But I realized that we needed to clean our data up first. And so that for me getting into it, the initial thing was data engineering.

and learning about data and really getting into the nitty gritty and realizing that we had to over overcome so many obstacles before we could get to the point of where we could really use AI. for to its full benefit. The fact that AI needs a ton of data in order to work efficiently and correctly and everything else.

But we can have a ton of data, but it's not standardized. It's all over the place. You can't bring things together and companies aren't talking to each other. Not even companies, but inside of companies. You even have individual parts. Drilling's not talking to reservoir, not talking to production.

Because they don't want to share what they have in data. And they don't agree that this is right or wrong. You're saying our production numbers are right, yours are wrong, whatever it's else. Because nobody could agree on Or they can't share. Like the data's just all different formats, all different structures. And AI is really good at bringing crossing those bridges too.

Right. No, definitely. And so the problem for me that got me into looking into it was how do we get to that point and trying to break down a lot of these barriers initially and getting people in the conversation. So we started Bringing in professors, honestly. I brought I I hired professors to come into Contro and talk about what is AI, what is data science, and teaching our engineers and saying, before we can say hire a data scientist and an engineer goes, here's my data scientist.

I do data science. Here's your answer. Okay, this is the answer the data scientists gave me. So this is must be the answer. And you're like, okay, does that make sense? No, it doesn't. You gotta have somebody at least a a working knowledge of it and an understanding of it. And so that's where my approach to it was let's all have an at least an understanding of what we're talking about. And that way when somebody comes to the table that actually will do for you that

you know, AI or m data science or engineering or anything that you need done that you have an understanding of what's going on and it's not just some black box that we look at. Because no, as a in as an industry, we hate the black box approach. We really do. We don't like things we don't understand. And that's why it's so hard for us to adopt things we don't understand. And so to get over that hurdle, we had to first understand what it was.

to be to understand what is data science, what is machine learning, what is all these things that are coming up that we knew were coming on the horizon and were gonna have an impact on our industry and how can we understand it before it gets here.

Yeah, I agree. And that black box and demystifying the black box is such an important part of the change management element of applied AI, in my opinion, because you're right, no engineer wants to just listen to something else tell them how to do their jobs. if they can't interrogate how it's come up with that decision.

And a lot of the way that data is parsed and you start to believe in a single source of truth because you believe in the way that those data pipelines are interconnected and you make sure that if an asset has a slightly different name, so it's, you know, A or well underscore A or Hey, I like A, and that all comes into one place together. You have to trust that all of these wirings are actually valid before you start to trust the output.

And then once you trust the output, you can start to trust the AI models. But until you do that, you're living in this world of a black box and it's great when it works, but people don't trust it to work all the time. And when it doesn't work, that's the one reason they need never to look at it again. And I'm gonna add sorry, I'll add a real quick to the black box because that's a that is just an industry discussion in general, right? In in and really short, I think it's also open.

To do AI in your company, you need to have partners that are Interoperable. Right. They it's ultimately it's your data. It's no longer hi, this is my offering and this is where I go. If you really want AI to work between IT, OT and ET, you have to have that interoperability of a platform. in order to enhance and make sure that the hallucinations of AI are not exceeding. And I think that's a big thing where we used to be this

Oh, it's this and you can't touch that and it's this. So you need to really work with your providers and your partners to make sure that they have that same philosophy as you. Well, okay. So my personal experience, I remember giving a talk in front of at NLV. I was invited by SP to give a talk and I was there to give a talk about machine learning AI, but really was there to to really talk about upskilling, right? Upskilling engineers, anyone who wanted

to least be involved in AI. And I was talking about the magic of developing these machine learning models. And then I get a hand raised in the audience and he was pretty bright. He says, it's a black box. How are you going to validate it? I said, we're going to have to bring in some physics to validate it. So this whole concept of a black cock. and then explaining the outcome can be done now. There's such thing as called explainable AI.

And it's all based on game theory. And so based on the algorithm that you develop, you can bring in explainable AI that will tell you which particular input parameters are really influencing your output. So at that point it no longer becomes a black box. Right. The only way that you're really gonna have to understand it is to really go into the algorithm and look at the mathematics and look at the weights that are doing all the intelligence stuff.

But you can explain predictions through explainable AI. So at least that way you know what is influencing to what level the output. And that gives people a little bit more confidence. Right, in terms of the prediction. Yeah, audiences here live. I'm gonna ask one more question to the panel. We're getting close to wind this thing down. After they ask this question, I'm gonna open up the questions for the audiences. So start thinking of stuff that you'd like to ask. There's no dumb questions here.

All right, my next question for the panel is All of y'all have been involved in AI projects to various degrees. And listening to us here in person and on the podcast itself is gonna be a lot of companies that maybe are just thinking about starting the first AI project.

Or maybe what happened where somebody tells you you have to do an AI project, whether you know what that what it's for or not, right? Can we talk a little about some of the gotchas? What is the one or two things that y'all know from experience that if you're gonna start an AI project? That you need to look out for. Like what's one or two things that you've seen over and over again that is something that may be simple that listeners need to be aware of?

I think some of the major obvious ones were hallucination that was brought up, but one of the other ones is it largely it what is it what is it it reinforces your statement. Tell me why these wells are bad. Tell me why these wells are good. Same set of wells. It'll come up with two very different results. Occasionally you ask it the same question for the same set of walls and you'll get very different answers. I think it's important, like the professor said, to bring

physics in to test your data and always query your data. Never trust AI absolutely and never take whatever you receive from your AI agent and just go straight forward with it. Always ask questions and always doubt. Now, AI has come leaps and bounds, as I mentioned earlier, in the last year has gotten phenomenally better, but we do still get hallucinations. And it's important to remember AI is not an expert.

AI is really good at pattern matching. It's a really good at taking in the knowledge based data that you give it and giving it back. I think there was an example Earlier this year where uh was it ChatGTP was regurgitating everything off Reddit. So everything on Reddit, whatever the opinion or idea was, it was taken as gospel.

So I think you need to always and it should be the same thing when anybody gives you an answer. You should always use your own logic and always read through what it is you're given and query the data. But I think AI is a strong tool that just needs to be used properly. You can use a hammer for a wrench if you want, but it's not gonna turn out well.

So I think that's the biggest thing about AI. It's a strong tool and understand what it is you're working with. For me that's simple. I can easily develop any kind of model, whether it's physics based. physics informed AI, machine learning. But the one thing that always gets me is every time I go in front of my senior VP, now this is at Champion Act. He basically said, What's the business case? What is the ROI?

And that's the one thing I never really think about, right? Because my mindset is always about developing the model, making sure that the predictions that I'm making, the metrics that I'm using are at the highest level as possible to give me confidence. But the one always gotcha is Where's the business case? Where's the revenue? How is that going to affect my EBIDET? And so I think when you want to first start a project, you should take roughly about two months to really plan it out.

And understand that what is the business case? Is it going to be a revenue? Efficiency, what is a ROI? And then ensuring that I get buy-in not only from the executives, but also from the subject matter experts, because without their buy-in and without the business case, that's always got me. That's always been my gotcha. Never the technical side. I I think we're all gonna say the same thing on this panel for this question uh because

That is correct. And then Vince, what you said was absolutely correct. AI is a tool, but you have to be able to know and trust that the data it's pulling is actually the data that you need, right? And that's going to start with access to my data, right? I can use my data. uh and I can connect my data to the correct asset or anything I'm looking at. And then lastly, I'm going to trust that the data because of the tools I give my AI agent.

You treat it as an I was treat as an an additional intern, right? You have an intern come in, you need to train and prompt your our AI. To actually, okay, what are the tools I'm gonna give it to use? And what exactly is that AI agentic, especially agentic now, right? What is that role?

of that intern. So you need to be able to set these governance and these standards in order to trust that the data it's pulling is going to be accurate. And then you need to also be able to go back and like you said, run a query and say, where did you find that data? I need to know that. And then ultimately that's like the beginning, but the end is where does that business value come into play? I still have to have a business case under that.

Yeah, and not just to repeat what's already been said. I think the one thing that I would caution people with is to say that there is no silver bullet in the industry and AI will not be a silver bullet either. Right. So it comes back to that business case and knowing exactly what are you trying to do with AI because if management says use AI to improve Ibita twenty percent.

Fine, nice. Thanks for the challenge. We'll do what we can, right? But it's gonna be incremental steps because you've got to have data quality, data fidelity, and a lot of other elements that work in your favor before you start making such meaningful changes. And from my perspective, I you just brought up physics informed AI. I love to use this example when I talk about pure AI, especially if I'm on stage, I'll say an AI model is gonna do a really wonderful job defining the stage.

based on the steps that I've taken. But what if I take three steps forward? Will pure AI tell me I'm about to fall on my face or not? And that's where physics informed AI comes in and you start to define the boundaries of the science problem that you really have on your hands. And then with AI, you're doing a much better job of defining the high fidelity operating envelope.

that you're usually seeing. And so that blend allows you to not only get really good and really fast at making normal decisions, but with the physics informed elements, you're starting to get more predictive around things that you don't normally run into. Right. And so I would say don't treat it as a silver bullet. Make sure you have really good quality data and also be reasonable with the way that you're gonna apply it if you're gonna start looking at predictive analytics.

Wow, it's hard to follow all that. I guess it everything. I think maybe one last thing to simplify some things. Um The idea that AI being called artificial intelligence

is modeled supposedly after human intelligence. And so the idea that you're only going to get out of it what you put into it. It's not going to You're not gonna be able to give it the parameters of this table and say, where is what's the location that we're at or things like that. It's not gonna answer any questions for things that it doesn't have any fair information for, right? And so it doesn't it's not some kind of magic eight ball that

that solves all your problems for you, it literally can only answer what is actually put into it. If you have if you don't have quality data, if you don't have a valid question to ask it, a valid business question, if you don't have the things that you need for it initially,

You may not be ready to get into AI, right? So you may need to wait a little bit. If you're just starting out on some small company and you have very little data to start off with, maybe AI is not your answer at this time. Right now it's just to work on just physics based. type stuff. And then later on, once you start developing more and having more things you can actually look at and having something more of substance to go with, then maybe AI would be the thing to go to.

And I think that people think that it like you said, it's not a magic bullet. It's not gonna fix everything for you. It will only answer the questions it has the abilities to answer for you. And that off of the things, the knowledge it already has. It is, after all, a machine learning thing and it is we talk about the fact that it is it is something that it learns off of things that it has previously experienced. Just like

human intelligence. I couldn't answer you anything off of visiting Africa. I've never been to Africa. So I couldn't tell you anything off of that. You could ask me about it. I could make up something because you asked me a question about it. And I could say, Oh yeah, Africa's it's dry right now. I don't know. That's not that's vague. But that's the thing. AI would answer some question like that with a vague answer because you asked it a question.

N not based on things that it has and because it may not have the knowledge for that. And so depending on what you feed into it is what you're gonna get out of it. And to add to something you said, Ronnie, too, is like uh in the smaller organizations, you don't necessarily have to eat the elephant all in one bite. Right. And so then it's looking at how can I apply AI to get some meaningful business value today? And then so maybe your foray into AI is

edge deployed AI for single well optimization. So advanced analytics deployed directly on a rod pump to do a lot more than you could do with traditional technologies, have advanced analytics and advanced control. making hundreds of adjustments a day to make sure your wells are kept running, but also producing more oil and at a lower total cost of operations. Yeah, I'll maybe I guess end that question to

tie everybody into that. The last thing that we need to look at is there's the one business case and one business value. There's the data foundation and making sure we have correct data. But there's also the ultimate time to value and scale.

Right. And this is no longer a multi year depending on the size of your company process. So I think you have to keep at the forefront that If I look at all of my operations, not just one business case to maybe prove out the concept of AI, w what else can I do and how quickly can I scale that?

uh in order to get multiple business cases across all of my divisions. I think Vency it was you said it, the well to the production to the GL I mean, all of that. So I think there's we still have to keep in the forefront that Technology moves quickly now, right? You can't m you're gonna make mistakes. Everybody does. But how quickly can we scale that value? Yeah.

I just want to add one one other thing because uh this actually happened to me six months ago and the issue of being able to scale your solution. was something that I didn't think of. And so the gotcha for me was we were able to develop a really nice AI solution that actually detected methane leakage from start to finish.

And I'm always quite interested in in terms of latency, in terms of how quickly my AI model ran. And when I timed it, it it ran one second. And I said, Wow, one second, that's great. And it worked fine when I was looking at just forty sensors out in the field. And data coming in at fifteen I I think every fifteen minutes. But then when I try to scale that to six thousand sensors out in the field for this particular application.

It just crashed the entire enterprise. And so then I I meant that I had to go back and look at the latency of the model. And this is an AI model. Instead of running one second, I had to really modify the code so that instead of running one second, now it's running point one second. That gave us the ability to go from 40 sensors to 6,000 sensors.

So as a small company, you have to ensure that whatever solution that you're coming up from an AI in SLB is very good at this actually cognite is very good at. being able to scale your solution. is so important. and so you always have to keep that in mind. I think that's an easy slope to fall down. We started with a couple hundred thousand files and we're like, Oh, this is good. It took

minutes. Then we add a few terabytes of data and you're like, oh, what did we just do? Yeah. So it's easy to slide down that slope and not realize what you're doing. We did a whole database we're just gonna categorize and organize and

Wait a minute. We just flipped the switch on something much bigger than we thought we're getting into. So those are easy to slide down. Those are easy holes to fall in. All right. In person audience, raise your hand if you have a question for our panelists. Paige is gonna walk around with the microphone.

Anyone. My question for you guys is across your day to day, what are some of your favorite tools and in which area do you apply them? I'm gonna have to say we have an entire value division and they have a tool called Minoa. And it is phenomenal at helping you plug in real life scenarios, experience and what you're looking at from the numbers to to help find really that business value.

And they come from the industry it themselves. So we validate a lot off of their expertise. But I probably s I would probably say that one. I was gonna say did aggregation. I started off back in the old file explorer days where you had to look up each file by itself and then you get into these better systems and softwares they can find maybe by a well name or maybe by area, but AI finds stuff that I didn't even realize would be tied to something. It'd be

like a permit or some sort of work over just oddball files just all over the place trying to aggregate all everything I know about a certain region or project or something like that. So the ability to really scoop through all that big pile of data and come out with useful things has been really awesome for me just recently.

I I'm personally using a lot more generative AI than I thought I would at in the business level. I it was a fancy, fun tool to play around with my kids to make images and things like that in its infancy. But now I'm seeing it's becoming part and parcel of my daily activities, right? So being able to use embedded generative AI

to help with daily sales motions or discussions or PowerPoint, what have you. But also now that it's getting deeper embedded into physics-based simulators and actual what we would call hardcore industry products.

makes it a lot more accessible. And so now I'm able to go back and do things that I hadn't done in five years or ten years. And even though the products have significantly matured, I have I don't have as much of a learning curve that I have to overcome because I can use generative AI to help me uh along that way.

Yeah, I have to agree with generative AI. I think for us and a lot of us that deal with data in the industry, there's so many overlapping solutions and things like that. Oh overlapping data, data coming in from different SCADA solutions and everything else coming in from the field. And so sometimes trying to initially the old way to do that was trying to

just sit down and figure out individual solutions, try different codings and things like that, trying different if different platforms, Python, SQL, and just nail it out until you got it right. And then sometimes you can ask generative AI and it'll come up with something you hadn't thought of.

Sometimes it's not correct. You try it out and it doesn't work, but sometimes it gives you a solution that is a little bit out of the box and you're like, wow, that actually is something I hadn't thought of. And so it's helped out a lot in those aspects. I have to agree with that, right? Because there was a specific example that I utilized generative AI last summer when I was teaching a class, summer class.

And at the University of Houston. And as an engineer, if you're a process engineer, one of the things that you always want to do is try to optimize some type of output parameter. It might be yield. And you might have multiple input parameters, right? And so how can I develop a really efficient experiment that allows me to minimize the number of runs, but maximize the number of in terms of output?

And and I actually taught a call um a class in the design of experiments, but I also teach a class in machine learning. So I thought, wow, wouldn't it be great if I can combine both design of experiments with machine learning? And so lo and behold, I used. generative AI to help me not only develop examples and generate the code, right, in a matter of

hours, which otherwise would have taken me days. But now what I was able to do through this generative AI is take two separate concepts called design of experiments and machine learning together. I was able to put them together So that now If one wants to develop or run, let's say, a computational

design of experiment, I can bring that aspect up with it and then combine it with my machine learning models to give me a much quicker answer. So I took two concepts, two separate concepts, machine learning and design them experiments and I was able to put them together and taught that as a gradual level course. So I think that's the tool that I really go to. Awesome. Any more questions?

Yeah, so my question is, of course, AI is beneficial to our lives and our businesses, but is there any areas in your life where it's we're not gonna cross this line? Do you have any no-nos whenever it comes to AI that you're not gonna use? Or go to a four in why?

I think some of the most stark examples I've seen had been very defined areas of like mathematics, physics, chemistry. So they showed some proofs of postulates of like mathematic postulates and AI just made stuff up. I think they've seen the same thing in the legal realm. I think when it comes to those very highly technical human knowledge requirement areas, I wouldn't say it's a no-no and maybe AI will find its way there, but right now it's just.

I think you you nailed it earlier when you said it's the intern, right? That's the new intern of shops willing to work hard. You give it a job, it's gonna knock it out, but let's not make AI an expert or let's not trust its expertise. I think for me, that's where the no no zone would be for AI. It's not out there to define the limits of human knowledge because it's just not there. I don't think so. Yeah, you gotta give it you ha there's so much. Like it's not just having

that foundation, you have to have the foundation plus the data governance that comes across with that. And so if you're I'll go back into saying it, you have to have the tools You have to tell the AI what it n is, who it is, what it needs to do and focus on, but you also have to trust that it's pulling where it needs to pull from. I I think if you go in and just say, Hi, I have all these systems of record.

And I'm gonna trust that you're just gonna go connect all my data and tell me what my output is, then I would be a no no. Yeah, I think we all have those friends in our lives that are confidently wrong and AI is Really good at being confidently wrong. That's a great way to put it. Yeah, I think uh the key is just define the boundaries because yeah, you wouldn't trust it to run every aspect of your lives.

At the same time, ten years ago, very few of us probably would have said we would have trusted a car to drive itself for us. And now you see that picking up steam. So it's about the technology maturing alongside the way that we use it, that's really the key here. So there are plenty of places I wouldn't unleash AI today that I'd probably expect to see it in five years' time, but it's just about being smart in its application.

Yeah, so for me, uh anything that would require actually require me to do something unethical or immoral, that's the boundary for me. So I will not write code that is biased. Or will force a unethical reaction or ask me to do something that is unethical. That's really my boundary.

I think ultimately we're not to a point in AI the ultimate scare where it it replaces humans. It's nowhere near that. And so I think really the idea is that there's no way that it will replace the human aspect. You've got to have it there to govern it the whole time. And the idea that it's not gonna replace the human aspect in any part of our industry, right? It'll do certain things. It improves on certain things. But it's not

All in all, you gotta have a person sitting there that monitors it and everything else, like the intern, that holds its hand and understands sometimes it's gonna say dumb stuff. Sometimes it doesn't make sense. Sometimes it doesn't understand the question you asked it. And that's basically it. I think it's a lot of it is that. And so

where you you have to actually have the SME. You gotta have the person that is extremely knowledgeable and knows what they're doing and knows what they're talking about, that can sit there and know the difference between what is given is right or wrong. as well. And just having that human aspect has to be there. And emails too.

A great panel and discussion thus far. My question is going back to data privacy and confidentiality. So how you envision AI working on that part, meaning competitive advantage or working for a particular small company or for one company versus the other one? Do you think that that's something that needs to be considered? And if so, what would be the vision? What's the conversation out there to bridge that part?

That's definitely been a concern of ours at Conico Phillips. One of the first things we did was we locked out all AI. We weren't allowed to use any AI. We're allowed to use some open source stuff to start off. real small scale, but you have to think all of these ASA IIs are based on training sets. And any data you dump in becomes part of that training set. So if I teach it something about my wells or my proprietary data, if you use that same software

Technically in the beginning you would have been able to see my data. You would have been able to query and access my data. So something we've done as a larger company is we've locked it out. We have it, we use it internally, but it's entirely you take the data the training set that it's caven with and it is cut from the outside. So as a smaller operator,

I imagine that's how the process is these days. I think in the early kind of Wild West days of AI, it was a little more open and there were some lawsuits and stuff. We're not there anymore. So in proprietary data and data privacy, I think it's something we have to maintain at this point. Just because it is accessing that open data set and if I were to allow that to be unleashed to the public, they could look at our private data, and that's not something we find to be acceptable.

So at Cognite, we actually have agreements to where between two operators that we actually do share data, we actually create An integration into the data foundation to where they're completely separate company documents, but we go through and we Take the metadata, it's completely secured. between the two companies. And so even with AI and the sharing, you can actually lock down to Vince's point, you can lock down who sees what part of that data.

You can do that with Incognite and you can share with your customers or with anybody else what data you want them to see because you're obligated. You can do this also with like regulatory data and everything. We actually just showcased a use case with another partner of ours to where everything was built from the data foundation and the secure and the data governance that came into place. But we had to solve a problem.

Two agentic agents talking back and forth to each other to get the data that they needed that was open and allowed. in order to solve a root cause analysis problem. So that's out there now, right? It just is gonna come again to your data governance within your own company and where you're gonna take that.

Yeah, it comes back, Vince, to what you said and what we were talking about earlier with the uncertainty aversion, right? And so our first reaction was also one of, hey, hold on a second. Let's make sure we understand what we're talking about before we really start talking about it. And then by taking that step back, we were able to get more comfortable with the idea of where we're gonna start.

with something that's maybe foundational and then we're gonna put it behind a DMZ and then we're going to train it, we're gonna opinionate it, and then we're gonna deploy it in a safe, secure way. It's no different than all cloud technologies, right? I think we underappreciate the complexity of using the cloud and the fact that all of our data is stored in combined ways.

But we trust that the cloud providers are doing their part, you know, to ensure our data privacy and our data continuancy, right? So there's a lot that comes into it. It part and parcel with any digital technology these days, it has to be secure. It has to be maintained. If it's not someone I would want to talk to, right? And AI is no different than every other cloud-based technology in that regard.

As an industry, when it comes to anything along with our data, I think we're very secure with it. I think that a lot of the stuff that we set up is is one of those things that yeah, unless you are you're gonna go out there and outsource to something that is public and you put your pri proprietary

data out there on some public service or anything like that, then you've done that yourself, right? You've actually done that yourself putting your data out there. So I think that a lot of security's already been set up for it. So I don't think it's as much of an issue as Maybe in the past it might have been when people were trying to do things with AI initially, but I think it now it's very really well established.

I really don't have my much to add to it. Actually reinforce what Linda said. You have to lock down your data. And that data locking down means whether the data from company to company or even internally, right, based on your persona or the person's logging in. He or she should have access to data that is pertinent to their job. And so locking down data based on persona internally and also across companies is key and it's a must.

It's a must. Otherwise you lose credibility quite quickly. And once credibility's gone, it's really hard to get it back. Mark, I gotta tell you before my question, we've been drinking each time someone says data.

We may need an OGGM Uber at the end of this thing. So my question is related, it's along the thread of the small company experience versus the large company experience. It could be initially said that as such a powerful tool at the hands of the public may be a bit of an equalizer, maybe level the playing field. Do you guys foresee a tool such as AI?

widening the gap and continuing to make the gap between small companies or small producers and large producers? Or do you see it as a tool that could potentially close that gap or continue closing that gap? Which way do you think that tool may go?

I'm gonna leave that to you guys. You guys have more experience in that. But I would say it's an equalizer because a lot of these tools, even with us, they're third party, right? Like we do we develop some internally, but a lot of this stuff is external solutions.

So my initial impression is it's an equalizer, but I'll leave that to you guys to answer. Yeah, uh it comes down to this concept that we talked about a little bit, uh the fast follower concept and who's gonna be third through the door or fourth through the door.

The companies that are more agile and a little bit more willing to be enthusiastic about these new technologies are gonna catch up a lot faster by using them. The ones that are really laggards, they're gonna get the divide, not because of their size, but just because of their willingness to look to new ways to solve classical problems.

I heard a CEO that everyone would recognize the name, but I'm not gonna share it, but here in the US say AI isn't gonna be the thing that differentiates us. It's gonna be the thing that allows the industry to continue to exist. I said that's a really strong statement. But also, probably not too wrong. If you look at the efficiencies that we have to really achieve to stay relevant for the next four decades.

AI is the thing that's gonna help us really do that. So I think it comes down to the willingness to adopt is gonna be the defining factor between those that close the gap versus those that are widened by the gap. Right. And then looking at how we apply it from a value basis, which will allow you to continue on that journey. Because if you just apply a bunch of AI and none of it adds any value, good luck getting it back in again. It'll be that once bitten, twice shy kind of analogy.

So yeah, that would be my recommendation. And then to end to that is are you going to scale it and how quickly are you going to scale it? And how quickly can you scale it? Because again, you can go in from a one use case AI, okay, I've got this one problem on these 10 wells or this compressor station. I'm gonna go ahead and take this out of the box. But you can't scale that long term. So that to me is like the summary of what everybody said as far as

You could start with AI as a small company. You can start with AI as a big company, but it's really what's the long term plan on how you adopt it across your operations. And I do believe it is it can absolutely be an equalizer. It's just gonna come at the pace that that you need it to. We have one

They've scaled across sixty petrochemical sites globally, forty eight languages. They've got forty use cases of artificial intelligence, agentic AIs running. So it really just depends on the pace that works for your company. I don't know, for me it's double edged sword, right? Because I think as a smaller company at startup

You're quite nimble, quite flexible. You can make decisions quite quickly. You don't have layers of management to approve your your uh your decisions so you can move quickly. And so from that aspect I think it will equalize

the playing field. But on the other side, unlike a large company that has tremendous amount of resources in terms of money, people that they can throw at a problem, I don't know. I think the advantage would be to the larger companies that have the resources, have the money, have the wherewithal, have the talent that a smaller company may not

So to me again it's a double edged sword. So you don't think that having AI tools might actually lower the barrier to entry to be able to scale up business value? It might. It might. Again, it it it it depends on the strategy and the plans and approach, right, that that now that they take. So it may. So that's why I'm torn, right, in terms of answering either one way or another, because

Again, if you have a smaller company that's very talented and you're backed up by venture capital and you have millions of dollars, then I think it will equalize the playing field. But if you don't have that, then I think it's gonna be a little tough.

All right. We have one more question, then we're gonna start winding this down. All right. I hope y'all ready for this one because it's a mind boggling question right here. Loaded intro, very short question. Last year, my buddy Jerry Solas, he's a motivational speaker. I went over to his house and he was showing me his whole system. He used Chat GPT to create his own AI system to work for him and create

A massive platform for himself. He is no, he is a millionaire now. Actually, he's only 26 years old, which is amazing. And he created a system for me as a photographer. On this is on podcasts. I'm not gonna throw it all out there. But this system was gonna make me very wealthy too. And it was gonna be like an app. My thing is this. The question is, where does it stop? If there is a line, forgive me for wasting your time, but if there isn't a line, where does it actually stop?

because we're good people, at least I can assume we're good people in this room. But where does it actually stop? Where if we create our own, you're gonna see it in the future. You're gonna see local businesses, all you feel businesses, people creating their own system that's gonna work for them.

But where does it stop where employee doesn't go in there and says, hey, AI, you're only going to respond to me and you're going to give me this money, you're going to do this, yada, yada, yada. Because Jared is a very intelligent man. He knows the right questions to ask AI. I know. That's why he's a millionaire now. And I'm not. But where does that line stop where AI is, hey, no, we can't do that. Security reasons, privacy reasons, or whatever. Is there a line for AI and not just ourselves?

I think that's a good question and it leads to any technology, right? Consider the computer or consider the cell phone. Like cell phones, I don't know if you realize that the ability for something to communicate data has gone everywhere. It's in everything we do with it's in toilets, it's in dishwashers, it's

Where do we stop putting the ability for a device to communicate its location and its tech its condition? I think it's a complicated question that I think it's a moral question we should all ask ourselves. What shouldn't be able to think for itself at that point? I think

It drops to a moral or ethical thing. Like at some point it should not morally we shouldn't have thinking cups or something like that. I that's a silly example, but I'm struggling for analogy. I think it's just a technological question, I think. I don't know, maybe I can add some insight. My husband's an artist and he did the same thing and he's not a millionaire yet. Darn it. We'll work on that. I think he needs to pick a different AI tool.

But he makes his own commercials. It's all AI and there's multiple tools in that type of social media setting that are just so set in stone. I think that bus is already sailed with a bunch of social media people. And I'm that's awesome for your friend that he's found the right the right crack on that. I think from an industry perspective, it's extremely regulated. Coming back into Brandon's statement before, right? The cybersecurity, the data governance.

All of that, I think, at this point is overtly conservative. because of that question itself, right? And where we actually put AI into our everyday lives in the industry. Maybe that'll help, but it's not totally the same playing field, but it's definitely a thought on it. Yeah, and I think if it's concerning our industry, the oil and gas industry, we're very guarded. I think the oil and gas industry is very guarded against anything new.

So if it's concerning that I don't I think that I don't think we have that kind of issue as of right now. I don't think seeing it on the horizon. But anything outside of that, I think it's it's up to us.

just to to regulate that, right? As we do anything. You can even think of in medicine or anything else like that. There's gotta be some moral implications that are put in, right? Yes, there were some moral things in history, immoral things in history that we did that we learned a lot from in medicine.

Which were really bad, but the thing is there's gotta be some moral applications as to w how far can you go with something and then versus the science that you learn behind it. How far do we want to push that boundary to get to that point? And that's It's really where humans step in to set that moral boundary. Yeah, I don't know. I don't really have a good answer for you. I apologize. But I don't know. I think I don't know if there is a point a no return for AI.

But I think this is where government can come in and really set the regulations that we need to follow. So I think this is where government can play a key role in in ensuring that that we don't go beyond a point of no return. where some ethical, unethical tool or decision is made based on AI, but I don't see where it ends. I just don't see it.

Yeah, I think hopefully AI can filter out the train that just passed by so no one on the podcast can hear that. But you said something I was gonna say, Ronnie, which was in the medical field, right? And you start dealing with medical decision making. and how patients will be treated or aggregated or worse. Those are the kinds of things that I would say give me pause when I think about applied AI today.

to the point of what your husband's dealing with, there was a company out of Europe I think that just trademarked an AI actor or actress. And that actor and actress has an agent now. And so you can hire a completely artificial actor. to be in movies. That's gonna end singers and songwriters and it isn't absolutely gonna disrupt a lot of industries, not just in oil and gas and the way we do our daily business. So we're already seeing the tentacles reach out.

To the point of regulation, I'm sure there is going to eventually be a good application of regulation, but what we've seen over the last twenty years is regulation doesn't keep up with the pace of change in technology for the most part. I'm not necessarily putting all my eggs into that basket personally.

I think it is up to the individual industries to a point to help regulate ourselves. And we're doing that with security. We've done that with the cloud. There were a lot of things that the oil and gas industry could have done very wrong. And I think we can pat ourselves on the back for the most part that we haven't done a lot of those things in the wrong way. We haven't seen refineries. have unregulated AI lead to loss of life.

We haven't seen massive disruptions in the midstream companies' ability to deliver oil and gas to market or upstream companies really have to shudder because they applied AI in the wrong way. And this comes back to risk mitigation in general. But hopefully that partially answers your extremely complex question, maybe the source of more time spent after the podcast. Yeah, speaking of the podcast, we'll see if my editors who do use AI, see if you can get that train sound out.

And if not, audience, you'll hear the train. And if you don't hear the train, that means yeah, the AI did get the sound of your job. There you go. That means it didn't want you to answer that question.

We need to wind the podcast down. We're not gonna wind down what's going on today, but I just want to give a huge congratulations and thank you to Linda with Cognite, Brandon with SLB, Ed with Champion X. Of Conico Phillips and then Ronnie also Coniko Phillips. Thank you for your time today, ladies and gentlemen. This was absolutely fantastic. Appreciate you. Round of applause. Thank you.

All right, so we're gonna wind down the podcast. This is the point where I usually try to get you to sign up for stuff. There's no reason to do that. We went on for an hour, which is longer than most. Audience, let us know what you think. This is the first time All and Gas This Week has ever done a panel. AI is very pertinent to the industry right now. It's on top of everybody's mind. There's a lot of misconceptions out there. Cognite does sponsor this podcast.

However, we did not do this because Cognite sponsors this podcast. We did this because we want to educate you, our listeners. So hopefully you gain some information. This let us know what you think of this type of format where All and Gas this Week every now and then does something different than cover the news.

Paige you ready to shut this thing down? Yep. Yep. So remember folks, do great work, pay it forward, and we will see you next time. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thanks for listening to OGGN, the world's largest and most listened to podcast network for the oil and energy industry. OGGN dot com to learn about all our other shows. And don't forget to sign up for our weekly This show has been a production of the Oil and Gas Global Movement.