This month’s Pipeline Technology Podcast episode sponsored by Pipeline & Gas Journal features Clint Bodungen discussing what ChatGPT is, the ins and outs of how it works, and different ways to utilize the booming technology. Clint encourages listeners to learn the technology and begin utilizing it in ways that benefit you.
In this month’s episode, you will learn about the efficiency of the technology, where it pulls its information from, and why it is not completely reliable to use without human supervision.
ChatGPT for Cyber Security Show Notes, Links, and Insider Terms:
- Clint Bodungen is a world-renowned industrial cybersecurity expert, public speaker, published author, and cybersecurity gamification pioneer. He is the lead author of Hacking Exposed: Industrial Control Systems as well as the upcoming ChatGPT for Cybersecurity Cookbook, and creator of the ThreatGEN® Red vs. Blue game-based cybersecurity simulation platform. He is a United States Air Force veteran, has been a cybersecurity professional for more than 25 years, and is an active part of the cybersecurity community, especially in ICS/OT (BEER-ISAC #046). Focusing primarily on ICS/OT cybersecurity since 2003, he has helped many of the world’s largest energy companies, worked for cybersecurity companies such as Symantec, Kaspersky Lab, and Industrial Defender, and has published multiple technical papers and training courses on ICS/OT cybersecurity vulnerability assessment, penetration testing, and risk management. Connect with Clint on LinkedIn
- ThreatGEN® Red vs. Blue is a revolutionary new game-based cybersecurity simulation and IR tabletop exercise platform .
- Pipeline & Gas Journal is the essential resource for technology, industry information, and analytical trends in the midstream oil and gas industry. For more information on how to become a subscriber, visit pgjonline.com/subscribe.
- ChatGPT is an AI chatbot that uses natural language processing to create humanlike conversational dialogue. The language model can respond to questions and compose various written content, including articles, social media posts, essays, code and emails.
- OpenAI is a technology company specializing in artificial intelligence research and development.
- Gamification is a method of using video game environments or gaming principles to simulate real-life events for training or education purposes.
- AI (Artificial Intelligence) is intelligence demonstrated by machines in contrast to the natural intelligence displayed by humans.
- Generative AI is a type of artificial intelligence technology that can produce various types of content, including text, imagery, audio and synthetic data.
- GitHub is a code hosting platform for version control and collaboration. It lets you and others work together on projects from anywhere.
- API (American Petroleum Institute) represents all segments of America’s natural gas and oil industry. API has developed more than 700 standards to enhance operational and environmental safety, efficiency, and sustainability.
- Stack Overflow is a question and answer website for programmers. It is the flagship site of the Stack Exchange Network. Stack Overflow is the largest, most trusted online community for developers to learn, share their programming knowledge, and build their careers.
- Visual Basic for Application is a human-readable and editable programming code that gets generated when you record a macro
ChatGPT for Cyber Security Full Episode Transcript:
Announcer: “The Pipeline Technology Podcast,” brought to you by “Pipeline and Gas Journal“, the decision making resource for pipeline and midstream professionals.
And now, your host, Russel Treat.
Russel Treat: Welcome to the Pipeline Technology Podcast, Episode 34. On this episode, our guest is Clint Bodungen with ThreatGEN. We’ll talk to Clint about some recent YouTube videos he’s been dropping about how to use ChatGPT and generative artificial intelligence to support and streamline your cybersecurity efforts.
Clint, welcome back to The Pipeliner’s Podcast.
Clint Bodungen: Thanks. Good to be back. I love coming on your show.
Russel: It’s been too long, man, and I’m glad to have you back. I have seen what you’ve been doing with ChatGPT. Before we go, maybe just take a minute and remind the listeners what you do and talk a little bit about what you do with ThreatGEN?
Clint: Sure. I am a career cybersecurity professional, started in the US Air Force in the ’90s. When I got out of there, I started doing cybersecurity there, moved into cybersecurity for my career. In the early 2000s, I got into industrial cybersecurity, and I’ve been working, consulting, developing, we’ll call it industrial cyber risk management since about 2003 or so.
I don’t know 2013 or so, we had this wacky idea to use computer gaming engines to be a medium for training and provide visual training and things like that. Obviously, I’ve been on the show here previously doing gamification type stuff. In 2017, started a company, said, “Hey, this can work,” and quit our jobs in 2019, and boom, we started to do this thing full time.
That’s what we do at ThreatGEN. We developed a tabletop exercise. It’s a gamified or game based cybersecurity simulation and tabletop exercise platform.
That’s what segues into what we’re doing here. What we’re talking about here today is that recently, the degenerative AI or the ChatGPT boom has occurred, and it’s a perfect opportunity to use that for cybersecurity education. That’s why I’m here.
Russel: I will tell you, Clint, that your episode on gamification of all the episodes we’ve done is one of the most listened to and it continues to have people find it and listen to it.
Clint: Wow. Number one, I’m honored. Number two, it’s good to hear that people are interested in that technology.
Russel: You’re doing some cool stuff. Look, I asked you to come on today to talk about ChatGPT. I saw some of the stuff you’re putting out on LinkedIn and sat and watched some of your videos and I found it really fascinating. I don’t have the opportunities to play with software like I did 20 plus years ago. Can you tell us a little bit about what you are doing with ChatGPT?
Clint: Maybe I should just explain really quickly for the audience that doesn’t know what ChatGPT is first, and then maybe talk about what I’m doing with it.
Russel: Yeah, perfect. That’s a great idea.
Clint: ChatGPT is a product related brand created by a company called OpenAI. What it is is a way for the average consumer to utilize generative AI, in this particular context, a large language model generative AI.
Now what generative AI is, is if you’ve ever used autocorrect or more specifically, maybe autocomplete on your phone and you’re texting and it automatically predicts what words that should be next, that’s generative AI. Specifically, it’s a large language model. What it is, it simply predicts what word it thinks you should be saying next or using next. In a broader context, a large language model generative AI application is, instead of just the next word, it can predict entire sentences and paragraphs and context. What it does is seemingly it creates conversations. That’s why there’s things called chatbots. ChatGPT is a chatbot that can emulate human language, and in many ways, human tasks that are language based.
It’s very accurate because of the way it does it. It has a huge database, data set of information, which has been trained on the entirety of the Internet up through 2021. Whatever question, whatever prompt you give it, it very quickly analyzes its entire database and makes predictions on what you’re talking about, and how it should respond. You can ask it questions, say, “What is a cybersecurity policy?” It will reference its entire…Ironically, it’s only like eight gigs of data out there that the entire Internet in text that it references. It very quickly references that. It finds out through context what a cybersecurity policy is, or maybe what cybersecurity is.
It looks through all conversations and all public domain text or information that is out there on the Internet on conversations about cybersecurity policies, templates, everything that’s out there. It starts to, word by word, sentence by sentence, generate a predicted response. It is so contextually accurate. That’s what ChatGPT is and how it works. A lot of people think that it’s like, “Man, it can think, it can reason.” No, it just analyzes and predicts a reasonable, feasible, accurate response based on its entire database of information, which is everything. That’s what ChatGPT is.
Russel: I think that the use of the term artificial intelligent is a little bit of a misnomer.
Clint: Yeah, I agree.
Russel: This kind of technology has been around for a very long time. For those of us that do it in math, trying to understand math problems, this is nothing new. What’s new about this is they’re doing it with language.
Clint: Exactly. For you math people out there, if you’ve ever played those little puzzles that said, “Here’s a math pattern. This and this plus this times this equals this,” and they’ll show you several of them, and you have to find the pattern and predict what the next one is, that’s exactly what this is, just in a language context.
Russel: It’s just doing it in language instead of math. That’s it. The science of that, once you understand it, is not that complicated. The part of this that’s complicated is what it does to have this big data set and how it uses the big data set. That’s the bit that’s complicated.
I know a lot of people are using ChatGPT to author articles, particularly author first drafts of articles. Because the Internet rewards you for generating content, ChatGPT can radically impact the workload for people that are copywriters writing blogs and articles for magazines and all that kind of stuff.
Russel: How are you using it?
Clint: I’m using it specifically in the context of cybersecurity in a couple of different ways. As I mentioned before, it has the ability to create natural language processing conversations based on and around cybersecurity.
A lot of people don’t realize that, number one, it can write code. I’m using it to write Python scripts, shell scripts, and write code. 90 percent of the time, the code works out of the box, out of the chat box. It just works. It writes code. It’s a developer. It’s my sidekick. It’s my copilot. In fact, there’s a code assistance tool in GitHub and in a lot of other coding applications where it’s called Copilot. In fact, Microsoft is coming out with a cybersecurity product using the ChatGPT-ish application called Cybersecurity Copilot. I use it as an assistant to help me write code.
I use it to help me generate first drafts of articles, of policies, of documents, things like that. That’s something we should talk about here in a bit is about copyright and plagiarism issues, because there’s some miscommunication and misunderstandings there. I’m using it in those contexts. ChatGPT can also generate formatted reports. In the ChatGPT web user interface, you can use it to generate tables, you can use it to generate headers and nicely formatted things that can translate or be copied directly into word documents. However, I’m using it beyond that.
The real power of, let’s say…I’m going to put quotes around, every time I say ChatGPT from here on out, it’s in quotes, because ChatGPT is a generic term, like Kleenex, for a broader use case using other tools like the API.
Don’t get me wrong, I use the ChatGPT web interface daily, hourly. However, I also use the API. I’m writing scripts and I’m writing code to build applications on top of the API. This is where the deeper context of cybersecurity comes in. To be able to have it, you can have it access the Internet. You can have it analyze large documents. There’s tons that you can do using the API that play into the world of cybersecurity in terms of policies. Let’s say I don’t know anything about API 1164. I can literally have it ingest that document, give me summaries.
I can chat with it, I can ask it questions about API 1164, and it’ll give me the answers. What if I take my own cybersecurity policy and 1164, combine them, feed it to ChatGPT, and say, “Hey, how does my policy align with 1164?” It’ll tell me where it does and doesn’t and will spit out a report. That’s the kind of analysis that’s useful in cybersecurity.
Russel: You just said a good gob there. I was going to say a mouthful, but that was way more than a mouthful.
Clint: That was my summary. We can break it down after this. That was the summary.
Russel: Let’s start by talking about this from the policy and documentation standpoint. What you’re telling me is I could take my cybersecurity policy and procedure documents, load them into ChatGPT, and then say, “ChatGPT, compare this to the requirements of 1164”?
Russel: It’ll spit me out answers?
Clint: Yes, absolutely.
Russel: That’s handy right there, let me tell you.
Russel: How complicated is that to do?
Clint: You can do it in the ChatGPT interface, and it’s tedious because it takes a lot of copying and pasting, because, for lack of a better phrase, ChatGPT only has so much memory to work with, so you have to copy and paste it in pieces.
In terms of building a tool that can do it in larger chunks, if you will, and do it in one batch, if you understand Python, it’s easy. It’s a very easy thing to accomplish. There are people who are developing tools to do this. I’m developing tools to do this. If the tool is already written, it’s extremely easy. It’s a drag and drop in a drag drop chat.
Russel: Interesting. [I’m going to have to have a conversation with Scott Williams as soon as we get off of this recording.
For the listeners that don’t know, Scott Williams is our lead developer at our company. Scott and I have been working together for longer than either of us care to mention. Long time.
Clint: Scott’s a .NET developer, right? He’s a C# .NET developer?
Russel: Yeah, he’s a .NET developer.
Clint: There are libraries. There are lots of libraries. As you know, I’m a Unity developer, which uses C# and .NET, and we’re putting this into the red versus blue. I can absolutely point him in the right direction to where all of the C# and .NET libraries for ChatGPT are.
Clint: I just blew your mind, didn’t I?
Russel: You didn’t blow my mind up. What you did is you just lit a rocket, and I’m going down a path in my mind about how that could add some real value for our tools and for our customers.
Clint: I was going to say that. I was going to say that you already have a fantastic tool in the world of compliance, compliance management, and regulations, and things like that. If you added the generative AI large language model, and again, we’ll just say ChatGPT, it would be deeper.
If you add ChatGPT capabilities into what you’re doing, number one, not only would you be one of the first companies, if not the first company doing it in your domain, you would add so much value to your customers, because you would add the functionality to automate a lot of analysis.
It’s a value add, having your customers chat about their policies..
Russel: This is such a big deal in cybersecurity, because when a new threat breaks, I have to analyze, “What do I need to do to mitigate the threat?” Part of that analysis is I got to look at my program and see if there’s anything in my program, my policies and procedures I need to modify in order to address the reality of the new threat.
As there’s more and more threats and more and more policy and procedure around all that, that gets more and more complicated. If you take it in my space and you talk about everybody right now running around and they’re trying to figure out what to do about the new valve and rupture rule. If I could use a generative IA…Generative…
Clint: AI. Generative AI.
Russel: Thank you! The brain is going faster than the mouth at this moment. Sorry…
Clint: Just say ChatGPT for generic purposes.
Russel: If you can use ChatGPT to accelerate the front end analysis, it makes the actual drafting work a lot easier.
Clint: That’s the real power of using tools like ChatGPT is because it can analyze things faster than a human, it can spit out results faster than a human. It doubles, triples, quadruples, shoot, multiplies to the power of 10 your ability to analyze and produce information.
In cybersecurity, let’s get this straight, in a cybersecurity application, we’re not using it to produce content. We are producing something if I’m using it to help generate code, but we’re using it as a tool to make us operate more efficiently and a lot faster, and that’s odd.
To say you can do something faster and more efficiently, a lot of times is a paradox, because when you speed up, you do things less efficiently or you make mistakes. Because you were using…By the way, you said AI is a misnomer, and it is. It’s machine learning, not AI. AI is a broad term.
This is machine learning. It’s an application of machine learning. Machine learning is done in the training. Anyway, I digress. The real power of this is not generating content, generating creative content like reports or articles.
The real power is using it to help you with the more difficult parts of your job and the parts that you’re prone to human mistake, like doing analysis, and looking at things very quickly, and helping you. Speed, as we know, in the industrial world, speed and accuracy is money.
Russel: That’s very true, very true. In a safety environment, speed and accuracy is improved safety performance.
Russel: Interesting. I know enough about ChatGPT to see how that could work. Basically, what I’m doing is combining … Let me see, I got to try to think about how to ask this question.
Off the shelf, ChatGPT has its library of documents and information. Can I feed information in and make it part of the base data set that ChatGPT is using?
Clint: Not of the base data set, but there’s two ways to do this. There’s one thing that ‘s called fine tuning or training the data. Basically, you have this data that exists outside of its original context, and you would add to it, and you would train it, and fine tune it.
Right now, that’s the traditional way and it’s the clunky way because that does require you to have enough GPU resources and it takes time to get that embedded into your version of the model.
Even though, now, they’re improving that and then there are techniques that they have that you can do it in 24 hours, but there’s a second technique. It’s just called embeddings, to where…I’m not going to get into the geeky technical details on how this works, but let’s just say it’s very similar to how Facebook and Google chat similarity comparisons and stuff work.
Basically, you’re taking that document and you’re creating a local knowledge base of that information that is a supplement that ChatGPT can reference when looking through conversations and doing what you’re asking it to do in the prompt. That’s pretty easy to do.
Russel: Given the stuff we already have in our compliance toolset, oh man, this is getting…Clint, now you are blowing my mind up.
Clint: I’m experimenting with this exact thing right now. I’m taking all of these cybersecurity regulations, the most popular ones and the ones that I’m allowed to do without licensing, and I’m putting it into a knowledge base. I’m experimenting with this right now. It’s actually something I’m about to release publicly for people to experiment with.
It’s all these cybersecurity regulations and policies. People can ask questions about these policies. How do I comply with this? How do I do this? Literally take your cybersecurity policy, upload it, and compare and say, “Where are my gaps?”
The only way that you could do that.
Russel: You could also use it to create a boilerplate, to manage a boilerplate set of policy and procedure, a generic boilerplate set.
Clint: Yes. Here’s the thing. You can only create that knowledge base by doing these embeddings or having a pre-trained data set. It doesn’t know about the…Well, I take that back. If you go into ChatGPT right now and you ask it to give you some…”Hey, what are the main sections of API 1164,” or something like that.
It knows this stuff because it’s been in there. It’s there already, but your policy, it doesn’t know about.
Now, here’s a caveat. The caveat is when it comes to cybersecurity, there’s a lot of conversation about well, the things that I asked ChatGPT and the things we use ChatGPT for, they save that data. It’s not private. I’m exposing my data. I don’t want my cybersecurity policy exposed.
There is a way to do this all locally. Basically, there’s these models that are not OpenAI models. They’re not ChatGPT models. They’re large language models that are almost exact subset.
Russel: Said another way, basically what’s going on is there’s applications out there that unlike ChatGPT, which is largely using information off the Internet, they’re not Internet facing and can work against a library of stuff that you manage.
Clint: Yeah, exactly. You’re taking the same data set that ChatGPT is based on and using it locally, and you’re making you have a knowledge base locally, so it never reaches out and touches the Internet. It’s completely private.
Right now, the general consensus is that this method is 99 percent as effective as ChatGPT, which ChatGPT is known as the crème de la crème. It’s the one that is leading the market in accuracy.
Russel: It’s interesting. It’s very interesting. Look, I want to pivot because I want to talk as deeply about what you’re doing on the coding side. Talk to us about how you’re using ChatGPT to develop code and what code are you creating? In other words, actually what’s it doing, and what’s the purpose of using ChatGPT to create that code?
Clint: I’m using it on a multitude of applications in terms of, for example, there’s a problem where one of our simulations within red versus blue. Actually, it’s in a different application, like red versus blue. We were trying to figure out a problem, and I’m not going to get too deep into it, but I couldn’t figure it out. My developers couldn’t figure it out.
It’s a niche problem, and it was one of these things that was hard to root out and figure out what to do with it. I asked ChatGPT, and this was in the early days of ChatGPT. It was ChatGPT 3.5, not even the new version of 4. I asked it, I told it the problem of having what I’m trying to solve. In 30 seconds, it solved a coding problem that we were unable to solve in three years.
Russel: Holy smokes.
Clint: It worked. It fixed it. It fixed the problem. I’m using it to help me quickly solve problems I don’t know the answer to. Here’s the thing. If you’re a coder and you know what I’m talking about, so you’re going to stack trace, or what it was called?
Russel: Let me make you pause, because this is near and dear to my heart. The thing about coding, it’s like trying to fix something in Windows these days. If I know where to go and what button to push, it takes me two seconds to fix it.
Trying to figure out where to go and what checkbox to mark can take me days. In effect, you’re using ChatGPT to do the day’s worth of work, to figure out where to go, find the checkbox, and execute the check.
Clint: Yeah, and I meant to say Stack Overflow, sorry. For example, the de facto place to go when you are looking for coding answers, there’s a lot of places. People go to Stack Overflow, and that’s your help. You’re searching the Internet, looking for these things, you’re going through W3C, you’re looking for places to find information.
ChatGPT can give it to you instantly, and give you working code samples instantly, and solve problems with coding instantly. I’ve written videos that I’ve done where I’ve literally said, “Hey, I want to develop a simple video game.” I told it, “Write me a Python program that can play blackjack,” and in literally three seconds it’ll write me a Python program that would literally be a blackjack simulation, text based, but still. It will write complete applications. If you want to get into it, get complex with it, and you know how to work with it, here’s an interesting concept.
You have to understand how to supervise ChatGPT. You have to be the director and know how to work with it. Once you know how to do that, it will develop an entire application if you want to. It won’t replace humans, and here’s what I mean. Everybody’s scared of ChatGPT replacing Jobs, replacing humans, context or whatever. ChatGPT is not perfect. You still have to have a human that knows enough about the subject, enough about the coding or context to be able to find mistakes, fix mistakes, guide it in the right direction.
Without guidance, it can be frustrating. If you know how to guide ChatGPT and you know at least enough about what you’re doing, it is a beautiful dance.
Russel: This is interesting, Clint, because I’ve not done a lot of programming in my career. What programming I did, I did many years ago in the ’80s. Late ’70s and ’80s, actually.
Even then, when you got to a certain place with a certain algorithm, what you would learn is, “This algorithm can do well within this space, and here are the things I have to do to teach the algorithm in order to get me the result I want.” This is the same kind of thing.
That’s one of the things that people don’t understand about artificial intelligence is artificial intelligence is not really intelligence. It’s just taking what human beings do and speeding it way, way up.
Clint: It’s analytics, not intelligence.
Russel: Yeah, it’s advanced analytics is what it is.
Clint: I’m going to tell you something right now that’s going to literally blow your mind. Not literally.
Russel: That’ll make it three now, so go ahead.
Clint: I just told you you have to have a human there that knows what it’s doing to keep it on track, and to guide it, and to catch potential mistakes, etc. Here’s the other thing, and I’m experimenting with doing this right now. I have multiple instances, agents, if you will, of ChatGPT that communicate together in different roles.
I will have one agent take on the role of the expert programmer and developer. You’re an expert in whatever programming language, and you’re the supervisor. I have another one that’s the coder, and then I’ll have another one that is the tester. I’ll have one have the director agent feed the context to the coder.
The coder writes the code. It will send it to the tester, because, yes, I can have it execute code, and I can have it test it. That’s another subject about having it interacting with your operating system. It tests it, it gets the errors, sends it back to the coder, the coder will correct the errors, the director will continue to keep it in context.
I will have these agents all communicating in concert, so I literally do have them replacing humans in some aspects. It’s about a 90 percent improvement in efficiency and reduction in my efforts, because now, instead of me being the one that’s directing it and telling it what to do, I’m having other chatbots direct it and tell it what to do.
It gives me a more polished product in the end so I spend less time checking over it.
Russel: Oh man. You want to hear what my takeaway is from this conversation?
Russel: If you’re not already playing with ChatGPT and figuring out how to add value to your organization and operations, you’re behind the curve.
Clint: Yes. There’s a saying that I saw a meme and I’ve adopted it, and I believe it. “AI,” and we’ll put a quote in AI, just like ChatGPT, “AI is not going to take your job. A person or organization that knows how to leverage AI is going to take your job.”
Russel: I would say, too, that if you want high paying, good jobs, you’re going to have to understand how to apply AI. It’s getting to the point, too, that you don’t have to be super technical. 10 years from now, AI is going to be on everybody’s desktop, just like Microsoft Word.
Clint: Sooner than that. Sooner than that. Microsoft and Google are already adding AI copilots to their applications.
Russel: It’ll be there sooner, but what I mean is…
Clint: You’re talking about J.A.R.V.I.S.
Russel: No, what I’m talking about is 10 years from now, everybody’s going to know how to use it. Take Excel, when Excel first came out, everybody’s like, “This is cool.” It was very clear very quickly that you could replace a lot of accounting and reporting tasks pretty quickly and easily in Excel. Then, people started really understanding all of the formulas that are in Excel, and then they started creating more and more formulas and putting them in there, and then they wrote macros. Then, people started extending Excel very substantially by writing code.
Nowadays, what you have is three kinds of users of Excel. You have the regular user that’s just doing simple accounting tasks and reporting. You have the more advanced user that’s not doing macros, but is very, very competent with all of the advanced statistical, and lookup, and database type functions that you can do in Excel.Then, you have another level that’s more, “I can use Excel to help me leverage what I’m doing in code.” You’re going to have the same kind of thing with these generative AI type applications. You’ll have a spectrum of users.
Clint: Even that is closer to what you think. There are already applications being developed, and there’s already applications in beta right now. Even Zapier plugins to where you can use a natural language processing request, just a, “Hey, Excel, I want an Excel spreadsheet, a workbook, whatever, that has this, and this, and this, can do this, and this, and this, and this.”
It will create VBA script that you can…If you combine this with Python, it will completely, 100 percent, automatically, just using your text input, a description of what you want, create a completely formatted, with all the formulas, with the macros and everything Excel spreadsheet to do exactly what you want. I have it doing that with Word files, Excel spreadsheets. You can already do all of that. It’s just right now, there’s not polished applications developed 100 percent yet that are out of beta, but we’re already getting there, just say, “Hey.” I already have it set up to where I can have it already reply to my emails. It’s easy.
Russel: Do what?
Clint: Yeah, it will sit there and watch my inbox, and for certain types of requests, I’m using Zapier plugins, because Microsoft’s Copilot isn’t available yet. It will actually respond to emails appropriately. It’s like a virtual assistant. It’ll automatically respond to emails.
I can tell it things like, “Hey, schedule a meeting with this person at this time,” or whatever. It’ll automatically craft an email, reach out to that person, play email tag on times, and then schedule it automatically for you. It’s a virtual assistant. You’ve already communicated with one of my virtual assistants.
Russel: I was going to ask that question. Do I need to start asking, “Is this Clint or Clint’s AI?”
Clint: It was the meeting request. You sent the meeting request for this. I didn’t accept it. It automatically accepted it. I simply told it, “Hey, when Robert…” Not Robert, sorry. I was literally just talking to Robert about this, too. “When Russel sets up a meeting, just accept it for me.”
Russel: This is why I like talking to you, Clint.
I’ve realized how freaking far behind I am with tech. Oh man.
Clint: Less than a percent of a percent of people…Here’s the statistic that I heard, and I don’t remember where I found this, but less than one percent of the population is aware of ChatGPT, then a fraction of a percent of that percent know how to use ChatGPT effectively, and then a fraction of that percent know how to use it as deep as what I’m talking about right now.
We’re in the beginning inkling stages of this. I am convinced that this technology is going to be as existentially important to the human race as the printing press and the Internet.
Russel: I agree with you. In that, I absolutely agree with you. We have no idea where this is headed. There’s going to be a lot of early adopters, and there’ll be a lot of things that people try to do that are not commercially viable, and there’s going to be a lot of other things that people are going to build major enterprises out of this technology.
Clint: Absolutely, and it’s already started.
Russel: It’s going to radically impact the human condition.
Clint: Absolutely. That’s an interesting concept. I know that the purpose of this podcast is not to talk about how this affects the human condition, in terms of copywriting and plagiarism, and whether it’s going to terminate us like Arnold Schwarzenegger.
The thing is that, and I’ll say these very quickly, in terms of plagiarism, and copywriting, and things like that, I have my particular stance on that, on the ethics of that, which I’m not going to get into, but it goes back to what I said earlier.
As long as you have a human supervisor that can look over this and check it real quick, you’re not at risk of that. I don’t…
Russel: There still needs to be a human being standing behind it.
Clint: Absolutely. Do I think…
Russel: By the way, Clint, do you know the difference between research and plagiarism?
Clint: You literally just touched on the thing I didn’t want to talk about, which I have my stance on. I have my long winded version. Tell me what you were going to say on that.
Russel: Research is when you take a little from a lot of places, and plagiarism is when you take a lot from a little place.
Clint: Thank you. You know what? I’ve never heard it put that way, and that’s perfect. I have a short video that I created on why ChatGPT is not plagiarism. What you just said right there was two sentences or a one sentence version of a three minute video that I produced. That’s it. ChatGPT, it takes a little bit…
Russel: From a lot of places.
Clint: …from a lot of places. It doesn’t do the latter, and that’s why it’s not plagiarism. That’s what we do as humans. Hardly anybody has an original thought anymore. Everything that you say, do, write, or whatever is a little bit taken from a lot of places. That’s what ChatGPT does.
Russel: Now, we’re going to get into a whole metaphysical conversation that we shouldn’t do on this podcast. That’s something we should do over a cocktail.
Russel: That needs to be our next step.
Let’s wrap this up. What would you tell pipeliners that they ought to be thinking about as it relates to ChatGPT and its capabilities?
Clint: I would say, number one, use this to accelerate your knowledge, and reduce your learning curve, and to make your life easier.
Everyone can start using ChatGPT, the web interface, right now. I would say start using it. Start learning how it works for you. Use the interface to start asking it questions about your job, about regulations, about issues with pipelines, detection, whatever. Anything that you may need to upgrade your knowledge on or whatever.
Chat with ChatGPT. Use it as a tutor. Use it as a form of reference. Again, be careful because sometimes there’s this thing called hallucination. If ChatGPT doesn’t know the answer, sometimes it will make stuff up that sounds real, but it doesn’t do it as often as people say.
Just be sure you do double check things, but for the most part, you are going to…You can use it to accurately learn things. That’s your first step. Start there and start seeing its capabilities. Then, dig deeper.
OK, I want to know about this. What about this? You can even test it. Ask it questions about things that you already know a lot about to test its capabilities and see how comfortable you are with it.
Then, I would say the best thing you can do is experiment with it and see how it’s useful to you. See how you can use it for you. Then, watch videos on YouTube or talk to people who have already been using it to get more ideas.
It’s going to be a rabbit hole, but you need to figure out how to guide yourself through that rabbit hole on your own. I can’t tell you how best to use it in pipeline, but maybe Russel can. Right, Russel? You can guide people on what they can possibly use it for.
Basically, take domain expertise and combine it with the capabilities and the knowledge and the experimentation with ChatGPT, and that will guide you to the right place.
Russel: This has been an awesome conversation, Clint, and I appreciate you taking the time. My takeaway is everybody needs to at least get their hands on it and play with it.
Russel: That’s my takeaway. To that end, what we ought to do in the show notes is put some links in there so that people can get to a place where they can create an initial account and start to play with it.
Clint: It’s impossible for us to cover all of the intricacies even in just a single domain of what ChatGPT can do. This was an awareness podcast. This was an awareness thing. I said video, I meant podcast. This is an awareness of, hey, this thing is out there. Here’s how I can help you. Welcome to the rabbit hole.
Russel: That’s exactly right. Awesome. Listen, man, great to talk to you. Thanks for coming on, and we need to get together soon.
Clint: Absolutely. We do. It’s going to have to be over a beer or a cocktail because it’ll get right into those conversations that we said were probably too deep for this podcast.
Russel: Yeah, the ones we don’t want to record.
Clint: Yep, absolutely.
Russel: All right, man. Take care of my friend.
Clint: Thanks. Take care.
Russel: I hope you enjoyed this month’s episode of the Pipeline Technology Podcast and our conversation with Clint. If you’d like to support this podcast, the best way to do that is to leave us a review. You can do that on Apple Podcast, Google Play, wherever you happen to listen.
If there’s a Pipeline & Gas Journal article where you’d like to hear from the author, please let me know either on the Contact Us page at PipelinePodcastNetwork.com, or reach out to me on LinkedIn. Thanks for listening. I’ll talk to you next month.
Transcription by CastingWords