Thad Yessa (00:00:03):
Welcome to a special segment of Ask the Pastors. I’m your host, Thad.
Will DuVal (00:00:09):
And me too. I’m co-hosting this week.
Thad Yessa (00:00:11):
I’m co-hosting with Pastor Will, and we have been going through a series of talking about AI, AI in the church, AI in personhood. We were going to start with AI and the work, but we’ve had some members reach out and just share, wanted to invite them to share some of their experience using AI, give us a little bit of perhaps helpful pushback and how they’ve used it and some of the categories we’ve previously talked about. Anything else you want to add Pastor Will before? No,
Will DuVal (00:00:40):
I think that’s great. Yeah. So Greg, thanks for reaching out to us and for listening to our first two part installments in this miniseries on AI use. And Greg was just offering, “Hey, I’ve used AI a ton with work, personal, even Bible study, would love to share with you guys and with the wider…” I don’t think you said the wider church. I think you said with me and Thad. But then we roped you in and we were like, “Well, shoot, if we’re going to be talking anyway and hearing about some of the practical…” Because we did a lot of the philosophical overview of AI stuff in the church and maybe some of the warnings and being careful with how you use it and things like that. And Greg, you were more like, “Hey, this can be a really powerful tool in positive ways as well.
(00:01:30):
So just let’s make sure we’re speaking to that too.” So I was like, “Yeah, as long as we’re talking about it, why not go ahead and have you on the podcast?” And then from there I was like, “Well, you and Lou are best friends and Lou uses AI all the time too, and why not have a two for one special? We’ll just flip the script and we’ll interview you guys.” So thank both of y’all for being willing to do it. And I don’t know if you guys want to maybe start by just introducing yourself a little bit more, at least as far as maybe your role. Again, this can be adjacent to or in place of the AI in the workplace conversation, and certainly going beyond that as well, but how you’re using AI in the workplace, personal use, even I know again, Greg, you said Bible study, things like that.
(00:02:20):
We can get into all that, but I don’t know how long you’ve been using AI, early adopter. I’m assuming both of y’all are kind of early adopters to it, but is that safe assumption?
Greg Stewart (00:02:29):
Yeah, I think that’s fair. I’m Greg Stewart. Yeah, I reached out to Pastor Will and Thad. Just wanted to have a follow-on conversation about AI use. So I guess just a little background on me. I work for an aviation company based in Northern California, very Silicon Valley adjacent, lots of tech-minded people, very much in the tech world, the tech culture. AI is very much encouraged to the point where if we’re not finding ways to use AI to be more efficient, more productive in our job, that’s probably going to catch up with you on your performance. So it’s strongly encouraged as a manager of software developers in that environment. I use AI on a near daily basis, probably in very different ways than what my team actually uses it and how Lou has used it in his workplace as well. But it’s an amazing tool for connecting the dots on our various data repositories to do in-depth analysis to help humans do very detailed reviews of our requirements, of our design documentation, of our software implementation.
(00:04:05):
It’s taken tasks that may have taken a week or two in my past to a matter of a couple hours. And it’s really been a game changer and accelerating how we get work done. Excuse me. I had that experience in the workplace and over the past year or two, I’ve just started exploring how to take those skills and that experience and apply it to my personal life in different ways. I think in the podcast series, you guys have already explored a lot of the basic use cases that I think most people will have encountered just with Google searches, you get the AI response and then you can continue the conversation there at the top of the Google response. So it maintains the context and you can have follow up questions and dive deeper a little bit. Most people will encounter AI just through those chat interfaces.
(00:05:14):
You can do a lot with that, but there’s many levels beyond that that you can explore. I don’t want to hog all of this, the mic time just ahead, but…
Will DuVal (00:05:32):
Good intro. Definitely want to come back and hear more about the personal use too though, but yeah.
Lou Rohan (00:05:38):
I’m Lou Rohan. I’ve been a software engineer for 26 years, the last 18 at a tech company also from Silicon Valley. I’ve always been what I like to call a low level programmer where I love to focus on the details, be close to the hardware, whether it’s writing device drivers or looking at algorithms specifically, managing data very carefully, optimally. I’ve worked on physics in games. I’ve worked on graphics tech. I’ve done VR stuff and we’ve done data conversion from asset to asset. So if you look at Pixar movies, you like Pixar movies, all those are defined. Everything you see on the screen is defined with an asset called Open USD. It’s universal scene description. And the last eight years of my career was with manipulating USD scenes. So yeah, kind of close to Pixar and Disney.
(00:06:52):
AI has had a dramatic effect on my work. I have a list of the way it started and where it went, and I’ll briefly… It started with auto complete, auto suggestions. So we were familiar with that on phones, but as we write programs, you’re thinking, you’re typing out how you want to instruct the computer. One of the first entry points of AI into software engineering was this auto complete where it looks at what you’re about to do and wants to try and finish it for you. And that was extremely powerful because it was a great augmentation, a great tool. We go from there to help aiding us in searching in a massive way. So where you used to have a knowledge of how do I find information about all these different aspects when I’m program, going from that and that actually being something that has value, I know where everything’s hidden, I know how to find everything.
(00:08:06):
And then if your agents or your chatbot or whatever knows where all that stuff is, all of a sudden that becomes very quick, very fast, and you kind of feel like you’re cheating and you kind of wonder about your value as an employee.
Will DuVal (00:08:24):
Not unlike what I was saying last week about writing a sermon and saying, “Oh, all the places that the Bible talks about sanctification.” I can just literally put in, “What does the Bible say about sanctification?” I can think of four or five off the top of my head, but I can also find all 15 of them like that even quicker.
Lou Rohan (00:08:50):
Same way. How do I find what materials are bound to this primitive… Again, you’re not supposed to know what that means
Greg Stewart (00:09:00):
Because I don’t.
Lou Rohan (00:09:01):
But the agent does. They can find it very in seconds. And I’ll stop there as far as… Well, what I do want to say is software engineering has changed. So starting about two years ago, it started to feel different. And then starting about a year ago, it was like, “Okay, this is actually going to be different.” And about every month or two, what people say is how it will change, changes. And that’s with advancing models as models become smarter. I’m using air quotes smarter and smarter and with different ideas and concepts on how these models are hosted. It is literally my job today is nothing like what it was two years ago. Because if you do that, in my company at least, if you did that, you would quickly, not necessarily fall behind, but you’re not going to be producing like others are producing. And others are producing documentation and code that you can’t keep up with.
(00:10:29):
It’s just coming out so fast. So you have to learn how to process it, have to learn how to work with it.
Will DuVal (00:10:36):
If this is too personal, you could just decline to comment, but you just retired last week and you said your job completely has changed in the last two years. So I guess I’m just curious how much of your decision to say I’m getting out of the game was a result of this is changing and either I don’t like the direction it’s going or I don’t want to have to learn a whole new job with it or… I don’t know. Is that too personal?
Lou Rohan (00:11:05):
That’s a big question with a lot of answers because it’s multifaceted. So that has to deal with my timing in life. It has to do with my financial stability and it has to do with my motivation for doing my job and it has to do with where I sit every day to do my job. But to answer that, I –
Will DuVal (00:11:33):
AI play any factor?
Lou Rohan (00:11:34):
It played a massive factor. Really?
Will DuVal (00:11:36):
Okay.
Lou Rohan (00:11:36):
Because I hate to say, “Oh, I just didn’t want to relearn how I do everything because I do want to learn that. I do. I want to keep learning. I don’t think any of us should ever stop learning. And I think that’s key to this whole conversation. And I didn’t quit because it’s just not the same anymore. That’s not why. It takes a lot of energy to keep up with what’s happening right now.
Will DuVal (00:12:02):
I can’t imagine.
Lou Rohan (00:12:03):
You have to be… And I’ve always been okay with this. Some people are very much like,” Tell me exactly how to work, how we’re going to work as a team, and we’re going to do that. “Other people are like,” We’re going to be fluid and we’re not going to structure this. “Well, those that are okay with things being fluid are going to thrive. Those who need a rigid structure are going to struggle and fall behind because it changes literally week to week with… Now again, I worked in a very dynamic program. It didn’t have a lot of legacy things, and we were always seeking out new things and doing things like that. I think even within software and within computers or whatever in technology, you’ll have companies that can’t move that quickly because their customers don’t move that quickly. And so they’re going to be shielded to some degree until they’re not, whereas I was fully open to it all.
(00:13:06):
So it played a role in my frustration to keeping up and the constant changes, sure.
(00:13:18):
But at the same time, it enabled me to do amazing things. I mean, right up until my last day, I was doing things. I was able to work on things that I had very little knowledge of outside of knowing the technology, but not knowing the details and actually able to contribute to those projects. Whereas before, it would’ve taken me weeks just to catch up and to know how to do anything. So that was different. I do have a regret, which I knew I would, which is that I left in the middle of the most dramatic change in that industry that’s ever happened as quick as a pace as ever. So yeah, I regret that, but at the same time, it’s not enough of a reason.
Will DuVal (00:14:11):
I’m curious with both of y’all having… Lou just having left and Greg still being in a job that, like you said, is so tech saturated or however you want to call it, dependent. And again, I’m asking this question out of total ignorance of both what you’re doing on a daily basis and also what computers are capable of doing. Like you said, most of the way I’m using it is very superficial. It’s not at the level you’re using it. But I guess maybe a stupid outsider’s question of this, is it feasible? Is it even probable that your job will be replaced in the next five years, 10 years, 20 years? Again, asking as someone who doesn’t know all that goes into software engineering and aviation, I think this is a real personal question for a lot. I had my first pastoral counseling session just two months ago with a computer programmer at our church who’s in his early 30s who came in, young kids, and he was on the verge of tears in this existential vocational kind of crisis of…
(00:15:42):
I mean, for him it’s like, I’m going to be replaced. It’s not a question of if, it’s when, how long it takes for them to phase me out and the computer to replace me or AI, whatever. And he’s like, “And then I don’t know what to do. I’m 30, I spent my career in this industry.” So I guess I’m curious, I mean, it kind of goes to my question, do you, Lou, about retirement? I mean, maybe even part of that, I don’t know, seeing the writing on the wall kind of thing, but for you, Greg, I don’t know how concerned you are about that or perhaps the engineers that are under you, but you’re a manager, so you’re safe for an extra three years or I don’t know what it is. Yeah,
Greg Stewart (00:16:25):
Good question. I’m in a management role right now, a people lead role. Those are still going to be needed as long as we have people doing the job. So from that sense, I’ll say I’m a little more protected from those kinds of things. I will say what I’m seeing in my own company and what I think a lot of companies out there are doing is they’re leveraging AI to produce more, to produce much more quickly, to produce it at a higher quality, and then they’re asking their existing employees to do even more. And so they’re not squeezing the workforce as much as you might expect with some of what you hear in the media, but they’re taking their existing workforce and let’s say 25 to 30% of the time they’re producing the same amount that they used to. And then they’re turning around and asking, “All right, now continue producing at that rate for the rest of your hours for the week.” So they’re getting more and more out of.
(00:17:40):
Good engineers are still absolutely needed in order to use the tool. AI is a tool. You still need humans in the loop to direct it, to have the ideas, to know what we’re trying to produce, to review the output. In my industry, it’s vitally important that the humans are reviewing the output. We have an expectation that if you are committing something to our code base, you’re writing new requirements, you’re capturing design, if there’s a problem there, you can never ever blame the person that did the work that’s responsible for the work has to be accountable for it. So we can never blame AI for a mistake. So it’s incredibly important that we’re still there to review the output. We could go in a lot of different directions of this conversation, but the AI models, they try to take a lot of shortcuts. They try to make assumptions.
(00:18:56):
They’re trying to be as efficient as it possibly can to get to an answer that seems accessible.
Will DuVal (00:19:04):
It’s a probability vector, right? I don’t even know what that means, but Bill Connick was trying to explain it to us at the table at the men’s retreat last weekend. So it’s giving you the most probable answer, which is usually the right one, but every once in a while and then every once in a while it just hallucinates and makes stuff up, right?
Greg Stewart (00:19:20):
It can. Yeah, absolutely. And so it’s really important that you’re reviewing the output and taking a critical eye to it. But that also leads to, and this is how I’ve learned to work with AI, I’m sure Halu’s learned to work with AI, you have to provide a lot of very detailed context for what you’re trying to accomplish. So you don’t just go to the open chat window and say, “Hey, go write me a program, analyze these requirements, go do this task.” We have sets of instructions, of skills that are documented. We have all this framework around our interaction with AI so that when I go and ask it to do a task, it consumes those instructions. It knows exactly what I’m asking it to do. It knows exactly where all my data is. It strains its activity to the thing that I’m asking it to do.
(00:20:18):
So that’s incredibly important for how you interact with these tools in order to get high quality output.
Will DuVal (00:20:27):
Quick follow-up question, maybe just to make you… Maybe it’ll make you think about it even a little bit more, in some ways less personal, in other ways more personal. You both are fathers of teenagers. So your teenage son or daughter comes to you and says, “Dad, I’m going to major in computer programming at college because I want to be like you. I want to be a software engineer when I grow up.” Do you say, “Don’t do it. The jobs will not be there 10 years from now, 20 years from…” Or do you say, “Yeah, go for it.”
Greg Stewart (00:21:06):
So if they actually came to me and said they were interested in this, I would support it. I’d say, “Yeah, go for it.” I think the number of job openings, especially for fresh college grads, is going to be reduced over the near term, midterm, but they’re not going away. We still need young engineers that are learning this craft, that understand how software works, that understand how to design and architect systems and efficient ways to accomplish these incredibly complex things we’re trying to achieve. I don’t see a near-term future in which we don’t have any software engineers in the loop. We need smart engineers involved in this process.
Will DuVal (00:22:00):
Encourage, discourage?
Lou Rohan (00:22:01):
Oh, I would encourage. I want to go back and then I’ll answer that. So you talked about, am I going to have a job? I think you can go way back, and past technology innovation has never reduced the workforce. Now, I know that this feels different, this feels less more discontinuous or whatever, but I’m going to credit a former coworker of mine, Haldor Vanor. He writes some really good blogs on LinkedIn, and he said in software, this feeling of continuity going from typing in a typewriter to a keyboard, to using fax machines to email, it feels very continuous there. Now it’s like we have this discontinuity. We feel less certain, but we have to reconsider, one, how we do the work. And for us, there’s no question of cheat as much as you can because you need to produce. Whereas for a pastor, going back to the previous podcast, I think how we do the work might matter.
(00:23:17):
It matters. You guys have said that. Second, what do we need to know? As software engineers, what do we need to know? How much do I need to know about what’s happening as the agent works with the implementation, with the input, the output? And three, where is our expertise most valuable? So these are, I think, really big questions. We can’t answer them all, but I think that’s good for every career. So going to the, would I encourage my kids to be software engineers? Neither one of them are even close to wanting to do that, but I would encourage them to go after whatever they are talented and skilled interested. That’s like a recipe for finding your life’s work. I want to also specifically about software products, and I’m not going to have a job next week.
(00:24:16):
Building the right software product, you have to build something that solves the customer’s problems elegantly, and you can’t do that without human blood, sweat, and tears. It might take less human blood, sweat, and tears, but you can’t do it without… There’s lots of reasons, but again, people like to say AI slop. I think I even heard you guys say AI slop, and you have to be careful with that term because your AI slop might be someone else’s product. And also about the workforce, definitely I am seeing that fewer junior engineers are being hired right now because senior engineers are superpower. We have, I say we, I’ll call myself a senior engineer. We have the experience, wisdom, knowledge, the intuition. Intuition is massive. So we can iterate with agents or teams of agents as if they were our peers or our co-developers, and that’s massive. A junior engineer may not have the intuition, the experience, the knowledge, the wisdom.
(00:25:35):
So my main question is how in the world we’re going to have a gap just like everything. If you’ve got nobody that knows how to research anymore, what’s going to happen when no one knows how to research?
(00:25:50):
I’m just bringing up all these things that I think about. Another one is I like this. I think this is important right now org charts. Org charts used to be a triangle. Imagine a triangle. You few leaders, lots of workers at the bottom. Well, org charts recently are trending towards a diamond, if you can imagine that. I think that’s helpful to get a picture of what is happening, fewer entry level because that dredge work is being done by automation. Well, it’s augmentation and automation all put together. I used Gemini to talk to myself about this before we had this discussion. So a lot of this is fed with… But again, I need the way I –
Will DuVal (00:26:37):
Tell me what questions they’re going to ask me. No, no, no, no. I get that.
Thad Yessa (00:26:40):
Talk
Will DuVal (00:26:41):
About… No, I’m teasing. I
Lou Rohan (00:26:42):
Know, but to get my thoughts straight, and again, that’s another topic. How do I get myself started? Do I write stuff down? Do I talk to a person? Do I talk to an agent? I think I’m able to separate that from actually making a friend, but maybe some people can’t.
Will DuVal (00:27:00):
And over time do the lines get blurred.
Lou Rohan (00:27:02):
Exactly.
Will DuVal (00:27:05):
I just heard about that for the first time the other day, that kind of use of AI. One of my friends was telling me, “Yeah, I’ve got this agent that now I can just be in my car.”
Greg Stewart (00:27:22):
Was it you?
Will DuVal (00:27:23):
No, who was telling me about this?
Greg Stewart (00:27:24):
No.
Will DuVal (00:27:25):
I’ll listen to –
Lou Rohan (00:27:26):
I can do one thing while I’m in my car. Drive.
Will DuVal (00:27:28):
Okay. Yeah. I’ll listen to a sermon in my car, and then I want to unpack it and think deeper and apply. And so I’ll pull up… I have this agent and we’ll just have a conversation during my commute and I’ll make sure that it’s listening to the same sermon I listen to. And then we’ll just talk back and forth about what we just heard and about how it applies. And I’m like, “That’s wild.” Yeah, first of all, that you can do that. Anyway, but maybe that’s a segue to some of, Greg, what I know you initially reached out to us about, really being excited to share about some of, again, whether or not you want to say that is a positive use of AI or not, but just some ways that we can use it, not just professionally, but personally that could… Yeah.
Greg Stewart (00:28:26):
Yeah. I’ll say that we’re all pretty new to this. We’re all learning. I’m still exploring different ways to use AI in ways that are actually helpful, in ways that are healthy. And I’m sure along the way, I’ll have some experiences and say, “I don’t want to do that anymore.” That didn’t prove to be a helpful use case. But personally, I’ve not really explored any of these relationships use cases with AI with that back and forth like you just described. I don’t treat my technology as a friend. It’s a tool. It’s there. You were
Will DuVal (00:29:11):
Telling us before we started recording that you have been… Was it before we started recording that you’ve been really careful not to use second person pronouns when
Greg Stewart (00:29:19):
Engaging? Yeah. When I first started using interacting with AI more, I was really careful to avoid using personal pronouns of humanizing this thing.
Will DuVal (00:29:28):
Can you do this for me?
Greg Stewart (00:29:30):
Can you do this? Or even please and thank you kind of words. What I started to see, well, two things. It seemed a little difficult to actually engage with AI using natural English without using personal pronouns. And by being so mechanical in how I interacted with this tool, I started to bleed over in how I interacted with actual people in my life. And so I would rather
Will DuVal (00:30:00):
Good point.
Greg Stewart (00:30:01):
I would rather just interact in a more human way with the tool so that it doesn’t impact my human interactions. And I will also say the pleas and thank yous, I don’t know, Lou can correct me on this, it may not matter, but in my mind, that’s a way of reinforcing good responses that I’m getting from you because it’s constantly learning. So within the chat context, it’s constantly learning. And so when you give positive responses to, yeah, that was a good answer, that’s what I like, then it’s learning what you’re really looking for when you ask questions.
Will DuVal (00:30:39):
And also, I was just reading an article this week that Anthropic is… I mean, the whole question of consciousness and all of that aside, but they’re apparently investing in, I don’t know, mental health kinds of resources for their agents. They’re monitoring, every once in a while an AI will just go in the spiral of I’m worthless, I’m terrible, whatever. And they’re like, something’s going on here and we need to… So they’ve got a whole department now at Anthropic anyway, that’s addressing the mental health crisis in AI agents. Just a fun fact.
Greg Stewart (00:31:33):
That’s wild. I’ve not heard that, but that’s…
Will DuVal (00:31:38):
I don’t know what world we’re living in anymore.
Lou Rohan (00:31:40):
I would imagine that that doesn’t feel that much different from a certain level of hallucination. And like you talked about, it’s just a chain of reasoning that somebody had said, but there are so many systems that try to keep things on the rails now, and there are companies talking about other… We have an initiative for safety or whatever. And so that feels a lot like an initiative to watch out for that sort of symptom that wouldn’t be too much different than another initiative. But it feels very human, which I think is why it makes news.
Will DuVal (00:32:25):
Yeah. Yeah. Anyway, I interrupted you. Sorry.
Greg Stewart (00:32:30):
Sure. I think we’re wanting to talk just about some of my personal use cases for AI. I think maybe some novel ones might be interesting for folks listening. Our last couple of church meetings, we’ve had updates on our church project and I don’t know, somewhere over – The
Will DuVal (00:32:53):
Building renovation.
Greg Stewart (00:32:53):
Yeah, the building renovation. So in particular, the financials, the donations, that kind of thing. At one point over the past year, I had some questions about what was being presented and I offered to try to provide a cleaner set of charts to help communicate more clearly to the congregation. And so those charts that we’ve seen in the past couple of meetings, that’s all based on my interactions with AI. And so the data’s all coming from our elders, from our financial team, and I used Claude to create a script to consume the data that’s available to me in a very defined format, and it creates charts and a PowerPoint slide. So now if I were to just get an updated spreadsheet with this month’s data, I can plug it into the script, hit go, and I can share a new set of charts with you guys.
(00:33:56):
So that ends up being like a two-minute process, but the hard part is just making sure we have data that’s up to date and correct. The presentation part’s really easy now. So that’s one way I explored that’s kind of intersecting with church life. The other way I think that is probably more interesting, I’ll just out myself. I think I’m one of the people that you referenced anonymously in your prior episodes here.
Thad Yessa (00:34:27):
No, it was me. I did because you sent it to me. It looked good. You can share more about the Bible study.
Greg Stewart (00:34:38):
Yeah. So in the course of leading a discipleship group, I was exploring ways to use Claude as a Bible study research assistant. So we decided to do a tour of the minor prophets. So the 12 minor prophets, some of them were very familiar with Jonah. We all know the story of Jonah. That’s an easy one to pick up and read and talk about. The other 11 are kind of obscure and they all get a little repetitive and doom and gloom cycle of the prophet. I was interested in finding ways to help us to better understand and to study at a greater depth than what I felt I could do just on my own with my own resources.
(00:35:39):
I essentially, as Thad described, I created some curriculum for how we’re going to approach the study of the minor prophets and came up with a study guide for each week that we met. But I think it’s really important to understand how I approach this. So going back to the earlier comments about it’s really important that you establish context, you establish what you’re looking for when working with AI. So within Claude, I set up a Claude project. I set up rules that described that my theological framework, the biblical translation that I’m most interested in using, pointed it at all of your sermon transcripts that are available online well, so it would bounce everything off of your prior preaching, in particular the Deuteronomy series that you’re going through, which was somewhat current to the study we’re doing. Identified biblical resources, scholars that I gravitate towards the Bible project.
(00:36:53):
Tim Mackey is the primary scholar associated with BibleProject, Tim Keller and some others. And so it established this framework of resources that I’m interested in. And then I would say, “Hey, we’re studying Amos this week, these four chapters. Let’s work together to make our study guide for this week.” And so it’s pulling from those resources. I’ll argue, it’s not interpreting scripture for me, but it’s aggregating information, perspective, and views from trusted scholars, and it’s doing it in a way that I would never have enough time to go
(00:37:40):
On my own. And so it allowed us to go deeper and to get a lot more out of our study of the minor prophets. And so that’s how we approached it. It would give us a little synopsis of, “Hey, here’s the scholars and what they think about how they’re approaching this with their sermons, their writings. Here’s how it connects to the Deuteronomy series. Here’s how it connects to the gospel.” And it would give me discussion questions. I know you guys kind of poo-pooed the discussion question thing.
Will DuVal (00:38:17):
I think it – And
Greg Stewart (00:38:20):
That’s something to think about moving forward, but I don’t know if it was Claude creating those questions or pulling those from a study guide that Tim Keller wrote. I don’t know exactly where they went or where they came from, but I do know as we would have our discussion, we’d walk through the material. Typically, by the time we actually get to the questions at the end of the guide, we’ve already talked about most of what’s already there. And so it’s not really driving us in any particular direction, but it just gives a framework to help push the conversation along.
Will DuVal (00:38:57):
That’s interesting. For the record, you can state your opinion. I think on something like discussion questions, to me, the best way to do that would be, which it sounds like maybe you did, was if I’m the leader, I’ll read through Malachi on my own this week. I’ll still push myself to come up with four or five discussion questions based on my reading. Then I’ll feed it into AI and say, “Hey, come up with four or five of your discussion questions.” And then I’m going to cross reference mine and be like, “Oh yeah, that’s better than mine. I’m going to steal that one. Oh yeah, I like the way they kind of improved this. No, I like what I said here better than the… I don’t know. I think so.” That’s kind of in some ways how I’ve used it for sermons as well on certain points.
Greg Stewart (00:39:49):
I think in many ways I think of at least how I’m approaching AI in terms of Bible study, it’s, I don’t know, the next evolution of biblical commentaries of… I mean, in your sermon prep, you’re using a concordance, you’re using written commentaries, you’re pulling from a variety of resources. AI is just helping us to –
Will DuVal (00:40:16):
Aggregate.
Greg Stewart (00:40:16):
Aggregate, access those in a very fast manner. Now, for me as a D group leader, I have a full-time job. I have a family, I have kids and activities and things that I don’t have a ton of extra time to go and do all this independent study and research to prepare a weekly discussion like this. You as a pastor preparing your sermon, I think –
Will DuVal (00:40:44):
Different.
Greg Stewart (00:40:45):
I think it’s a different context and I think there’s value in you doing the work, and I think that’s up to you. And
Will DuVal (00:40:56):
There’s probably value in any of us doing the work to the extent that we… Like you said, we’re finite creatures. We all have a limited bandwidth and all of that for it. And that goes back to some of what you were saying, Lou, too, also before we started, just in you even jotting down your thoughts for this and just that process of preparing even to be on the podcast and the benefit in the struggle of making yourself translate your thoughts onto paper and clarify them, summarize them, all that kind of thing. Yeah, you got three pages. But I mean, that’s part of what we’ve touched on in previous conversations obviously is just… I think the very realistic concern about as AI usage increases, brain power function decreasing. I mean, I don’t know that anybody’s arguing that that’s not already happening and not going to happen.
(00:42:14):
And I know there’s, in some ways, nothing new under the sun, but that’s, I think, probably a lot of our probably concern in previous podcasts that we’ve expressed is that as much as… Who was it that wrote Fahrenheit 451? Whoever it was, Brave New World, Aldis Huxley, as much as they saw the advent of TV being like, okay, now we’re all going to stop reading and we’re all going to just be glued to our couches and just stare at a screen all day long. I mean, it’s kind of happened, but then some of us who have kind of escaped it and we still like productive members of society somewhat. But this one feels even more… I mean, it feels like a quantum leap in that direction. And again, to come back to the workplace conversation too, I mean, I don’t know. I don’t know anything about anything, but it seems like people that do, really smart people assessing this stuff are genuinely concerned about the workplace displacement in the coming years and what’s going to happen.
(00:43:32):
And are we trusting these AI companies that they working with the government are going to figure out universal basic income to make sure people aren’t starving to death when they don’t get a paycheck because they don’t work because we don’t need them.
(00:43:53):
Whether that’s five years, 10 years or 25 years from now, it seems like it’s coming. And especially as population increases too. So population is increasing, workers needed is decreasing. What are people doing with their nine to five?
Greg Stewart (00:44:11):
They’re building data centers.
Will DuVal (00:44:15):
They’re feeding the
Lou Rohan (00:44:15):
Machine.You’re speaking from a zero sum mindset as if our economy can’t grow, as if the entire world’s economy can’t grow. And I don’t know that that’s ever happened. It’s never been that. Now, yes, it feels very different and it’s important to consider those things, but remember that that’s a zero sum mindset.
Will DuVal (00:44:42):
Limited resources. Yeah. Yeah. But even as the economy grows, and again, production, like y’all said, is skyrocketing. Who’s capital, right?
Greg Stewart (00:44:56):
I just think you’ll see the nature of work is going to change and it’ll be different for different industries. The idea that we’re all going to this state where we’re on universal basic income without things to do, I’m skeptical of that medical mindset.
Will DuVal (00:45:19):
Overnight.
Greg Stewart (00:45:21):
Overnight. The nature of work will change. There’s still going to be plenty of things to do. And if you’re concerned and looking for AI proof work, do something with your hands.
Will DuVal (00:45:35):
Come be a pastor.
Greg Stewart (00:45:36):
Yeah. Come be a pastor.
Will DuVal (00:45:38):
I’ll never have a soul.
Greg Stewart (00:45:40):
Learn how to…
Will DuVal (00:45:40):
True?
Greg Stewart (00:45:42):
What’s that?
Will DuVal (00:45:42):
Oh, I said come be a pastor because AI will never have a soul.
Lou Rohan (00:45:46):
I agree with that from the previous podcast.
Will DuVal (00:45:49):
That’s a whole nother conversation. Don’t get me started on how we can’t find people who want to work here. But I mean, for every one person looking for a job in a church right now, there’s 10 churches looking for a pastor. And I don’t understand it because again, if I’m like my kids and if I’m in my teens right now and I’m looking at the rest of my life, y’all are more optimistic than me. I’m like, “Don’t go into anything computer because you won’t have a job in 20 years. Go be a pastor, go be a plumber, go build data centers. This is where the jobs are going to be because they’re not going to replace soles. I mean, eventually they’ll probably have bodies and be able to do our plumbing too, but not bodies, but they’ll put the AI brain in the machine bot and it’ll be able to do electrician
Lou Rohan (00:46:47):
Dynamic.
Will DuVal (00:46:48):
It’s going to take longer to replace that.
Lou Rohan (00:46:51):
Right.
Greg Stewart (00:46:52):
I think what you’ll see is as our current understanding of work changes and we need fewer people to do the work that we understand today, I think what you’ll see is that people will have ideas, they’ll innovate, they’ll come up with new things, they’ll develop new things. People are going to find things to do and things to develop and ways to make money and ways to support themselves. I’m not too concerned about that. Yeah.
Will DuVal (00:47:25):
What questions do you have?
Thad Yessa (00:47:28):
I think a clarification on my part and then a question about the Bible study piece, if that’s
Greg Stewart (00:47:38):
Good.
Thad Yessa (00:47:38):
Sure. My clarification is with the discussion questions. The other individual was just typing in discussion questions and not doing any sort of pre-work on that, which is where my issue lied. One thing that I’d commend anyone listening and hearing Greg talk about it is one thing that Greg, you did with preparing and helping your group with minor profits is you did a lot of the legwork of building out the different rules of where it’s pulling from versus, again, I’d like to speak charitably. I’m guessing most individuals are just ChatGPT, what does this mean in Malachi? And that’s biblical interpretation versus I’m going to give you some resources, I’m going to give you a Bible translation, I’m going to give you things to pull from, which I think is different than just going and saying, tell me the answer. You’re doing some legwork with it and building in some rules to, again, help protect yourself and not necessarily blindly following, I put in all these rules so now I can just trust everything that it says.
(00:48:47):
I’d be curious for your feedback on, do you think it helps as a whole, if you were to do, I’m going to ask a hypothetical, if you were to do just you, the guys in your D group reading the Bible together and saying, “Hey, what stood out to you? How are you applying this?” If you would’ve arrived at some at least similar conclusions had it not pulled all the data and helpful summary information.
(00:49:23):
I mean, that’s just bad.
Greg Stewart (00:49:24):
So I think your question is for us going through the minor prophets, would we have had similar discussions, conclusions, questions asked if we did not have this material? For this particular study, yeah, absolutely not. Would not come anywhere close to the depth of discussion and understanding of these books.
Thad Yessa (00:49:53):
And I say that because I think it speaks to a different problem that I do not think technology is going to help with, and that’s people being biblically illiterate. And I’m not saying you are, I’m just saying as a whole, American evangelicalism has ignored large portions of scripture because perhaps it is a little difficult. It’s challenging to go through the minor prophets, the book of Revelation, what am I going to do with the book of Song of Solomon?
Will DuVal (00:50:24):
And just illiterate period, by the way. And
Thad Yessa (00:50:25):
Just illiterate period, which is also going down.
Will DuVal (00:50:28):
We’re getting more illiterate and more biblically illiterate. So who needs to read anymore?
Thad Yessa (00:50:34):
Who needs to read? And so I think, again, as we think in a church landscape of our people, how can we continue to resource and help people? And what you described, I have a Bible software program called Lagos, and they’ve already built out AI devotional tools so that I can go in there and I can have a pull of resources that I want and less than 50 minutes, here’s a 12-week devotional study on fill in the blank. I think the church has to figure out how to engage appropriately doing the work, doing the research, putting in the right filters, but also there still has to be, I think, two components as a pastor that I’m thinking about of am I being discerning by the power of the spirit and what I’m doing and not just blindly answering that question? And second, is this good for our people at West Hills soul and figuring out how to best do that leveraging the tools available to us?
(00:51:48):
Because I do think AI is a tool. If a card’s on the table, I am wary of using it because I know my own tendency towards laziness and shortcuts. And so it would be very easy for me to, if I start down that trail, then be like, “Well, it’s going to be easier for me. I’m super busy. I don’t have time to write for what I’m preaching tonight youth group. Here’s my text. Give me an extended outline and I can just wing it from there.” Could the spirit use that to transform lives? Yes, but that’s also worse for my soul and for the students. And so that’s where my own personal holdup comes of thinking of my own soul as well as those I’m caring for. And that’s where, again, I think it’s a tool to be used, a tool to be leveraged, but those are just some of my categories that I’m thinking through personally.
Will DuVal (00:52:44):
The quote that keeps coming to mind for me is, “It’s not the journey, it’s not the destination, it’s the journey.”
Thad Yessa (00:52:49):
Thank you. If I could just get to heaven, it’s okay.
Will DuVal (00:52:53):
But I mean, if it’s just about the destination, then just use AI. It gets you there a lot quicker. It gets you there a lot quicker. So I think maybe being, like you said, discerning about what kinds of projects or process… When is the journey really important and when is it not? If you’re aggregating data for a capital campaign, who cares? Just give me the stinking destination. If you’re writing a sermon, if I don’t fall in love with the text, then I can’t give to other people something that I don’t have myself. I can’t give our people joy and appreciation for this text if I just skipped over all of that to get a sermon. It’s not about a sermon. It’s about, like you said, the heart change, the soul formation.
Lou Rohan (00:53:53):
I want to say I agree with you, but if you’re aggregating data to present to a church full of humans who have given their hard-earned money and saved money, you want to present it in a way that is humane, loving, respectful, and beautiful and shows the progress accurately and gives people an incentive to give more.
Will DuVal (00:54:20):
Did you put all that in your prompt?
Lou Rohan (00:54:22):
I
Will DuVal (00:54:23):
Want to know if Greg put all that in his prompts when he aggregated the giving
Lou Rohan (00:54:26):
Data. Forgive me, I don’t mean to keep going, but
(00:54:30):
That is not simple. And I talked to Greg a lot during that process, and we looked at the output and then fixed it and worked on it some more because it wasn’t answering the questions that I had and that different people had. So again, you can do the 90% very quickly or the 95%, but finishing the five to 10% has always been a problem, and it requires that human sweat and toil and judgment. And the judgment is the thing that young people will not get without the journey, and that is the biggest problem. So I do think that we will, in the future, be more deliberate about learning the judgment, teaching the judgment, where it used to just come for free.
Will DuVal (00:55:25):
That’s well said. We got a couple minutes left. What have we not talked about that you guys had in your notes or in your mind that you wanted to talk about on this? I
Greg Stewart (00:55:32):
Did want to follow up on the Bible study thing on a couple topics. So first of all, this was something that I was exploring. I might not do this again, but I found it incredibly helpful because it was the minor prophets. If I was doing a study through, say, Matthew, I probably wouldn’t spend or leverage AI in the same way because Matthew’s just a much more approachable book. We’re a lot more familiar with it. You can derive a lot more just natural discussion. I think that’s important. While using AI to generate these study guides for the minor prophets, I was also reviewing the output and pushing back on things that didn’t make sense or that didn’t fit, iterating on what was being produced. And so there’s still my engagement at that level to try to refine it to something that’s more useful of like, “Hey, these sections are too long or this doesn’t make sense,” or that kind of thing.
Will DuVal (00:56:41):
We’re going to use AI to produce our church’s AI policy, and then after we do that, we’ll let you know whether or not to use it again. We’re just going to feed it into Claude and say, “Come up with a church policy on AI use for us,” and then we’ll have just a blanket like, “Here’s how to use it.”
Greg Stewart (00:56:59):
I’ll also add, I do think within your ongoing AI podcast series, I think you should have a focus discussion on the value of critical thinking skills, because without critical thinking skills, you shouldn’t be using AI. You need to understand what it’s producing and be able to look at it, evaluate, make your own decisions.
Will DuVal (00:57:26):
Understand that by using it, you’re killing your critical thinking skills. Is that true or false? Maybe that’s a whole nother topic.
Greg Stewart (00:57:33):
I think it depends on how you engage
Will DuVal (00:57:35):
With it. Over time.
Greg Stewart (00:57:36):
It depends on how you engage with it, but yeah, I think for many people that can be true –
Will DuVal (00:57:41):
Going to get harder to maintain those critical thinking skills.
Lou Rohan (00:57:44):
Cognitive decay,
Will DuVal (00:57:46):
The more you’re relying on.
Greg Stewart (00:57:47):
And then one more topic for you guys. I think it’d be interesting to explore how the church has engaged with and interacted with technology over the history of the church, with the printing press, with the radio, TV, with social media. How has the church reacted to changes in technology, disruptive technologies? I think that would be an interesting conversation.
Will DuVal (00:58:14):
Yeah. What did we miss?
Lou Rohan (00:58:17):
Scrolls and codex. No, no, we’ve talked about a lot of things. I appreciate you reaching out to us to talk about our experience. We didn’t miss anything. We covered everything.
Will DuVal (00:58:30):
Okay. Solved all the world’s coming AI props, if only.
Thad Yessa (00:58:35):
Well, now we’re done with this series on AI. Well, thank you.
Will DuVal (00:58:40):
No, we are definitely still talking about is it the antichrist?
Thad Yessa (00:58:43):
Oh yeah.
Will DuVal (00:58:44):
Quick yes or no from both of you guys, since you won’t be on that podcast. Is it the antichrist? No.
Lou Rohan (00:58:51):
No.
Greg Stewart (00:58:52):
That applies.
Will DuVal (00:58:53):
He paused though. Loop paused. He’s still thinking. No.
Greg Stewart (00:58:58):
No.
Will DuVal (00:58:58):
He’s shaking his head. Okay. All right. That
Greg Stewart (00:59:01):
Applies. It has some level of consciousness, and I don’t think it ever will.
Will DuVal (00:59:07):
Okay.
Greg Stewart (00:59:08):
It’s a tool. It might present itself. We might, as society, think it does. Your last podcast, it would’ve been real short for me. The answer to personhood is no.
Will DuVal (00:59:18):
Yeah. Yeah. I mean, it’s short for me too, but I do think it’s… I mean, there’s a lot of, again, smart, powerful people that are behind this that are really pushing that conversation. I don’t know how much you’ve tracked that Christopher Ola and the Pope back and forth on this, like the anthropic ZEO. Anyway, it’s interesting. It’s very interesting. And just keeping in mind that if you don’t give all the context to your prompts, what you’re going to get is answers, aggregate Created by Silicon Valley philosophy. These are the people that created the technology and whatever the LLMs they’re using and theoretically the vector is going through everything, but they’re the ones that created it. Everything’s biased. Everything’s biased. When I Google search for ChatGPT, the first result is Gemini. Here, come use Gemini. I’m just saying if you’re searching for biblical stuff, just know that until there’s Christian AI, which I think they’re working on that –
Thad Yessa (01:00:36):
Dominion.
Will DuVal (01:00:37):
Yeah. Unless you’re using that, you’re getting… Lu thinks Dominion is the antichrist. I can see it in your eyes.
Lou Rohan (01:00:47):
That’s a name. That’s the name they chose?
Will DuVal (01:00:50):
So now they want to change their answer to the antichristal. No.
Thad Yessa (01:00:54):
Well, that’s it for this week’s episode of Ask the Pastor. Just as a reminder, you can ask and receive answers to your questions by submitting them online or at the info bar on Sundays. A special thank you to Lou and Greg for joining us this week, and we’ll catch you next week as we continue our journey talking about AI.