Where Do You Draw the Line with AI? – Part Five

Hello and welcome back to the Blue Thingied podcast. My name's Luke and I'm here with Adam. How are things, Adam? It's a beautiful day.

The birds are still singing. The flowers are still blooming and I am blessed. How about you? That sounds great.

It is much the same here. On this side of the world, we are going in towards summer. And so things are warming up. The birds are coming back.

So what you're describing is starting to sound familiar to me. So yeah, technically, yeah, we're on the other side of this. So it's starting to cool down and it's lovely. So absolutely.

Good, good stuff. Hey, Luke, I've been really enjoying this series. We're about halfway through. How are you feeling about our series on Christianity in the age of AI?

Well, I feel like AI is shifting all the time. New things are happening all the time. And so I feel like every time we meet and we record together, there's something else that has happened. And so every question sort of has new meaning as we're going through it in real time.

And so I really appreciate just the journey that this has been and also just getting so many different opinions. of all of the people we spoke to, not everyone agreed. And, you know, it has been, I find that it's much more helpful when I'm trying to form my own hypothesis or my own beliefs around these things to hear the opposing arguments so that we can piece them together. And so it's been, it's been a fantastic journey. Yeah.

If we all agreed, like not all of us are necessary or something like that. I don't know exactly, but I, I truly appreciate the nuance and the agreement and disagreement. And we're really wrestling with this. And I'm excited about the question today because I think it is, yeah, the big red button question.

If you haven't, you know, go back and, you know, with no AI, that's a fun question. This question today is what I think is probably one of the most important because AI is here. There is no big red button to get rid of AI. We have to decide how to use it and give it those guardrails.

So today, what we're gonna be talking about is like, when do you just, ah, times I would never use it and when should you use it? What are those? How do you tell? And there's gonna be some nuance in this episode.

I'm excited to jump in and really get into this. Yeah, that question of where to draw the line is, it's really where we're living right now. So where do we go first, Adam? We like to start this off with kind of the hypothesis.

So I'm going to start with asking you, what is your hypothesis? How do you think people are going to draw that line? If we're drawing the line between what should not be used, where it should not be used and where it should be used, I think we're going to find that the people that are saying the areas that it should not be used, are the areas that are distinctly human and valuable to be human. And I think the areas where people say it should be used are going to be spaces that we find like repetitive tasks that no one enjoys that take large amounts of time.

I don't know exactly where we'll fall, but my hypothesis is that's going to be broadly where the line of separation is. I think you're wrong. Okay. Go with confidence, Adam.

Go with confidence. Yeah. I think that we were kind of following somewhere in similar paths. There is nuance to that that I'm excited to get into.

Yes. Excellent. Well, Let's explore some of the ways our guests said that we should not be using AI. Let's use that as our starting point.

And so let's play the first video for that, because I think this will give an interesting start to our conversation. Say I'm going to use AI to help solve this problem because sometimes it's the conversation, you know, you can't use it to have the conversation. You can't use it to address your feelings and having relationship with AI and, you know, asking AI to help you with a problem that you have and give you advice because AI will hallucinate help. AI will give you information it wants to give you based on the prompt it gets.

So I will... you know strongly recommend not to get use AI for advice if you get any advice from AI you need to check it with other sources with the professions and so forth so for example if you have an illness and you you know i have a headache and this is the symptoms and so on you put that in AI and it gives you a whole list of things these medication check your doctor still check you the professions um if you are there's a funny picture i've seen where Someone asked, can this be eaten in a mushroom or something like that? Can this be eaten? I said yes. And then the person next picture, the person is in the hospital.

But I said, but you said it can be eaten. They said, yes, but you didn't ask me if it was poison or something. You just asked if it can be eaten. So those things we just have to be mindful of.

So it can be eaten. Can and should. Yeah, can and should are two completely different questions. So I found this interesting.

And I think there's multiple ways that we could come at this. But I'm curious from your perspective, Adam, how do you think about this? Would you agree that we shouldn't be asking AI for advice? Whenever I'm looking at pretty much any window in which I see an AI, at the bottom of it, it gives a very good disclaimer.

I may be wrong, essentially, like verify. And now I think I would do that with any person. I mean, in most scenarios, if it's something important, I'm typically going to seek advice from more than one source. I'm going to use my own logic to help me frame anything.

And if one person told me that it was okay to eat that mushroom, I'd probably go ask maybe more than one person. Don't depend on AI for everything. Don't remove your thinking brain. It's not a replacement for your own thinking.

And it may have access to, it does have access to information that I probably don't. So I think that is valuable. But yeah, I guess be careful is probably my best advice. What about you?

I would say that information and decision-making are two completely different things. You can have access to all of the information in the world. I'm sure the AI, you know, using the mushroom example, right? The AI has a list of every mushroom species to be discovered.

Absolutely. It has that. It can describe them to you, but its ability to tell you what is right and wrong about whether you should eat it or not. It's going to be dependent on it, how accurate it is at diagnosing what that uh, what that is.

And usually that's going to be based on a description, a location, a, uh, or a photo, but none of those things are fail safe. And so I, going back to your point, it really comes down to what is the role of our own judgment. And my, my sense is that the role of AI is always, should be always, uh, outside of the making of the judgment call that that really should lie with us. And like, like you said, you're going to ask more than one person if you have an important decision to make.

And so I don't think we're, uh, I don't think we can alleviate that by asking grok Gemini and chat GPT, but I think, but I think that if we apply our own critical thinking skills, that is probably where we are going to get the best results with AI. And I actually want to be a little bit. uh, a little bit counter here and say that one of the examples that he used was healthcare, right? Talk to a real doctor. I would absolutely agree with that, but I don't think it is universally true.

There is some critical thinking to be had here as well, because you have to understand that doctors see a certain amount of data. No individual doctor knows everything there is to know. And so there are situations where, uh, an AI implementation will be able to diagnose something that a doctor would perhaps miss, but that's only because of the amount of data available. The important thing is that that, that information is verified by a doctor, right?

So I don't think there's something wrong with using AI. I think there's something wrong with trusting it as the decision maker. Cause at the end of the day, the AI is, it is the average of human intelligence. It's not the best of human intelligence, right?

It's collecting all of that information and like, saying, what's the most likely next word? What is the most likely outcome? And so, while ever it's doing that, it's going to need to be verified by actual thinking people. Yeah, we actually had an experience this weekend.

You know, Dr. Google was able to give my wife some advice about, hey, you probably should go into the hospital. And I'm really glad she did. At the same time, Dr.

Google didn't solve all of the things and after the event that took her to the hospital, she met with her primary care doctor and she was super excited and happy that she did because that doctor was warm and understanding and took time with her and did all of the things that the AI just couldn't do. but it did give her some information when she needed it at the same time. It can also be scary when you go online and ask it medical questions. So yeah, again, go back to use your best judgment. AI is useful.

Sometimes it can be extra scary and you can learn about every single case when this bad thing happened. Yeah, we're in a bit of a Wild West scenario with with AI and how it can give us advice about medicine, but I wouldn't ignore it. Yeah, great example. I want to transition to another way that we use AI because we obviously are talking about, you know, get advice from AI, but AI is also good at writing, at creating. written content.

And you could use it in, you know, a student say, write me a 500 word essay on, you know, the causes of World War Two, you know, if you had a class and you needed to suddenly it would write for you all of that. I also think of writing, you know, where if you have a journal, and you're journaling, hey, I write a journal entry for me on how I feel about my day. um you know like there's certain things or help me you'll write this summary of this thing there are different cases when it comes to writing and uh pastor puya he's gonna give us uh his thoughts on on writing let's go ahead and play that video uh i think this is common knowledge that you should not use ai to write Things that you would pass on as if it's your own writing. It's basically plagiarism I think so it's it's a no-no to To to write something especially in Bible study development. I think it's important to be personally involved in every step of the process.

I think it's a question that we'll have to continue to wrestle of where do you draw the line? You know, I think it's time will tell. And I know that some are more open and some are more cautious. And I would say I'm somewhere in the middle of sort.

I don't know if that exists. Yeah, I've got a lot to say on this topic. But I'm curious, before I share, what are your thoughts, Luke? How do you think about passing off writing that was not yours?

Well, I felt quite jarred by the word plagiarism. I thought that was a very interesting statement. And I know there have been scenarios where Uh, AI has pulled other people's work and reproduced it as its own. Okay.

That if you want to define that as plagiarism, I think that is accurate. But when, if we are talking about, and I guess the question then becomes, what are we talking about here? Because from my perspective, when I use AI to write, if I was just saying to AI, write me a 800 word posts detailing, uh, you know, how technology should be used in education. Like I could ask it to do that and it could be written and it definitely would not be me.

But if I was using it to write, I would not use it like that. I would use it as a thinking partner. Okay. What should I take into consideration?

Here is my version of the things that I think should be involved. What blind spots do I have? And we sort of build it together. And then I would say, okay, can you collate that into a draft for me?

And usually it will put it into a draft that sounds nothing like me. And maybe it's getting better at that. And I'm trying to make it better at that, but then I will go through and I will rewrite it in it. So it's in my voice, but the, the amount of, um, Time that it has taken me to collect the information is much lower and I would say that is me I would not say that that is plagiarism So anyway, so I think it comes down to how it's being used.

What about you? Yeah And I think it's also there's an integrity part to this that is always going to be true taking a school example a kid who had plenty of money has always been able to pay another kid, hey, write my essay for me. Like that's not a new thing. And basically it makes cheating democratized.

Cheating can happen for everybody, whether you're rich or poor. So it's not a new thing, but it does have to do with your integrity. And just because something is easier now than it was in the past, does not give us the right to lose our integrity. What I will say is that when someone is giving content to me, I don't care how you got it.

I don't care if you wrote it yourself and you took weeks and weeks to write it. I don't care if you used AI to write it, as long as you own every word of the output. Now there is some interesting interactions where, hey, Claude and I wrote this together. And there is a an element of just being like, hey, being really clear, I had a back and forth.

And sometimes I will literally spend hours back and forth on content with AI, getting it right. It's not that I took a huge shortcut, but I needed to use AI in order to get because my my words just they weren't going to get there. But at the end of it, That is what I wanted to say and still having integrity on the other side of it. Yeah.

Yeah. I think that is a concept that is, we don't talk about enough and that is owning the outcomes. Like, you know, I know for, for me, when I'm, if I'm writing a proposal for a client, it will start with me. Getting the, the information that I have access to.

And yes, I will have a conversation with Claude, but at the end of the day, it's up to me to say, are these promises that are being made promises my promises? Are they promises that we are going to live up to as an organization? The problem is when AI is like set free and it's not checked. And like in those moments, like when I, when I read content that someone else has clearly, uh, generated with AI.

I become very skeptical very quickly, unless I know the source, right? If I know it came from you, I trust you. And so I trust that you have fully vetted what you've sent me. And so even if it sounds like AI, I'm going to trust the content because I trust who it came from.

But if just somebody on the internet has written something that sounds like it came out of tech GPT. The amount of, uh, weight that I will give that writing is almost nil. So, yeah, yeah, absolutely. Yeah.

Very good. I agree. Yeah. Okay.

So, uh, as far as the answer, I'm, yeah, I think that answers the question that he was stating there about being personally involved at every step of the process. I think that That's the real core of it. It's the ownership piece. But having looked at some of these ways around, or the places in which we should refuse AI, I think it's also helpful to flip the coin and look at the other side, because there are some things that I believe that if AI is capable of, humans should never do again.

And that's an enthusiastic response from me. There's definitely some good insight to be found in that. Pastor Wes, I really appreciated his perspective on it. So let's play back that video and just hear how he's using it and how he believes we should be using it.

So that's for me where I have stopped. I no longer use it to even take content from me and create something else, something new. Where I do use it is as a thinking partner. It's helpful for me to talk things out.

It's helpful for me to kind of sound things out. And so I have a recorder. I have an app that I use for talking to my phone or my computer. And then I will put that into Claude. and then use that, not for it to take that and give me something new, but I will begin to ask it to kind of like process through whatever it is with me.

So also asking it specific questions. You know, I did this, I tried this for sermon prep, my last sermon, and it was actually really helpful. It's a challenge to think about how to preach in the context of this, but used to I would write my sermons. I would manuscript them.

And the reason I would do that was because I wanted to think through what I was gonna say. I wanted to, it slowed me down because I'm ADHD, my brain's just firing super fast all the time. And it made me slow down to kind of think through the connections that I was making and the argument that I was making in my sermons. And then I finally, I got to a place where I could do that more with an outline and got a little more freeform. but my my biggest concern when i'm preaching is is over repeating myself um like getting caught in like an interesting idea and circling rather than moving down the road and so um my last sermon like after i was when i was ready to write instead of writing i just got my little recorder and i preached the sermon just walking around my house like i just went through my entire line of thought and then i sat down with claude and i was like hey You know, let's look at this.

Is there, you know, any gaps in logic? Is there any places where I was unnecessarily repetitive? Is there any place where, you know, I got caught off on a rabbit trail and was able to use it kind of as a mirror to look back at that before I get into the sermon, right? Because I was listening back and I see that stuff after the fact.

So I've been using it like that a lot more just talking to it and having it help me think through things rather than asking it to create things for me. So with some of those thinking there, there were a number of things that he said in his preaching experience that are very similar to mine. So a lot of that resonated with me, Adam, but I'm curious, have you used AI in your sermon prep and did it look like that? Yeah, so while I have read the Bible, I don't have it memorized or cannot. think of every single thing that's in the Bible.

And I recently did a series on the fruit of the Spirit, and you know, love, joy, peace, patience, kindness, goodness, faithfulness, gentleness, and self-control, which I'm glad I could memorize all of that. But when I'm thinking of any one of those as a topic, because each of those was its own sermon, hey, Claude, you know, help me think of like, what are some examples in the Bible? of someone who has exhibited self-control. Is there a Bible story that really, and it'll give me three or four different ideas. And there are some that I had already thought of and some that I hadn't.

But having that, hey, can you tell me what was historically happening at the time of Paul or something? What other things, maybe what were the things that are going into why he might be saying it that way? So yeah, that type of feedback. having a back and forth, basically having a biblical scholar with perfect recall at my disposal is really helpful. Yeah, I like that.

I can resonate with that. One of the things that he said about his own preaching in that his biggest problems is repeating himself getting caught in an interesting idea and circling around it. These are things that I definitely relate to very strongly. And so the way I have mitigated that in the past is by doing precisely what he did of like just preaching the sermon at home generally when when i preach in the week leading up to it i try to go through it in detail like you like just as i would preach it three four times in the week leading up to the seven.

When i when i do that it helps me to identify those things what i haven't done is that employee to coach me. on that. And so that's something that I would definitely try. I hadn't thought of that before, but, uh, what I have done in the past is record myself and re-listen to it and just figure out what irritates me, you know? So, um, that whole process just gets simpler if you have a tool to do it.

And AI is very good at identifying those tools. Just before we were hopping on to record today in this, in this session, uh, there was, I was writing a script for a short video that, that I need to make. And as I was writing it, I was really happy with what I'd written, but I then ran it through Claude and just asked for feedback on it. And Claude identified three places where one thing that I'd understated, one thing where I'd spent too much time.

And as I was reading through, I'm like, yeah, Claude's right. Let me fix that, right? And so it's just that companion that works really in a very strong way. when it comes to good communication, it's difficult. It's difficult to know which order of things to put first and having that AI right there to be able to like, hey, try reorganizing this.

Yeah, that thinking assistant is pretty special. Yeah, it is. I also just want to just jump in and just... there's a, an idea that I was, I was reading somebody that was reflecting on what AI written content sounds like. And one of the reasons that it sounds that way is because it is too perfect.

It never, there's no aberration from the rules. When a real human rights, even a really good author, they have these little rabbit warrens and sometimes like the, the, the order of things will be swapped. And so I don't think we want to be I don't think we want total perfection, but we want to highlight when something is getting between the message and the audience. And if we can figure that out, we can fix that.

But I think we should, to a degree, embrace the imperfection of how humans communicate. So what you're saying, Luke, is that after you write something, before you... pass it back to me, you go and make some like spelling mistakes and a few things to. It wasn't, it wasn't AI. Actually, it's funny you mentioned that.

I have a habit and, you know, I've been working with Dave, you know, for 10 years in various capacities. He's been making fun of me for this for 10 years, but I have a tendency when I'm typing of holding down the shift key fractionally too long. So I double capitalize the beginning of a word. So when there's a capital letter, there's typically two capital letters.

He said, every time I see that, I know you haven't written with AI because it's just like, that's your trademark that you leave through your, through your content. So that's funny. That's yeah, your, your own watermark. Cause now Claude is watermarking all of our our written content.

You've got your own. So yeah, I know. I feel like in the past, that's something I've needed to edit out. But now I'm wondering, does it actually speak on my behalf if I leave it in like, I don't know.

There. That's Luke. So yeah, we've been thinking of it as a thinking assistant. But also kind of as I was talking about earlier, it's it can be an incredible research. engine, a research assistant, when you have access to all of these these great works that would have in the past taken you forever to just sort through and filter.

And I want to get to this next video clip with Patrick Jean from Amazing Facts. And he's going to share some of his process and how he's using AI. Let's let's jump to that now. Well, you know, I think AI is a resource.

It is a tool. It is an engine. a very advanced search engine in organizing and putting ideas together. Now, just like a person doing research, people come and research with their own bias. And I think that's true of AI, depending upon what the program is.

And even today, it's becoming evident that certain search engines or AI engines have a particular bias depending upon what it's been created for. Some are leaning more to the idea of engineering or building software or doing more heavy mathematical type things. Others are geared more for general information. It just searches the web and finds what's the most popular and it puts it out there.

So the general tone of the internet to a large degree creates the slant or the bias of the AI. And so you need to be aware of that. by and large the Internet has a lot of information and the majority of the information is not necessarily in harmony with the Bible. For example, on our ages or creation or the marriage relationship or family relations, those type of things. The Bible has a lot to say.

It's quite different than what the majority of information on the Internet says. However, there are others who are recognizing the potential of AI as this very Deep search engine and its capabilities and so they are building AI with a certain parameter where they will provide the information and allow AI to do the research but within a framework of what we understand to be truth meaning the word of God or Commentaries connected to that that we trust that are reliable or other writings that are in harmony with the principles of scripture and these AI engines focus on searching this particular Library of information and I think that's much more reliable than just using a general AI Although you know again, it just searches the internet and it finds whatever you want AI of course is also Coming to the point where it knows who you are based upon what questions you ask and it will create material to try and support your idea or your bias For example, AI that I use, it knows that I'm an Adventist pastor. And when I do research or ask questions, it will often give me an Adventist answer. Now, it might be quite different for someone who's coming from a Baptist perspective.

If it knows that, it'll build content to kind of support that. So there is, in AI, there is biases. We need to be aware of that. And I think there are tools now, and I think there'll be more tools available that will narrow the search. so that when AI works, you know, you're getting the information that you trust versus just everything out there on the Internet.

I know amazing facts in the process of doing that. We're developing AI so that it could search our materials. We're using it for Bible school, providing Bible truth. But again, we're creating the parameters.

So the search is guided. It's not just open to the Internet. Yeah. brings to mind when I hear Pastor John talking about the work that Amazing Facts is doing with Blue Vineyard actively is a very just tried and true saying that when it comes to technology and computers specifically garbage in, garbage out. And that's very true with AI.

If we give it information that is faulty, it's going to give us faulty information back. And would you agree, Luke, that there is information on the internet that contradicts itself? Wow, really? Surprise, surprise, Adam, just a little bit.

Like you could justify anything if you look in enough places. Yeah. Right. Yeah.

And so I think that guardrails It's not just, hey, AI, let's just use AI and not give it those guardrails. And what amazing facts and Blue Vineyard are doing are like super duper guardrails because we're going to make sure that we're giving it good information. Yeah, I think it really... I liked how Pastor John started, he's talking about bias.

And I think we forget this. I remember when... when AI was kind of becoming more popular and it was starting to personalize well, I remember people were posting, Oh, look, you know, even if you ask AI, it agrees with me, you know, like if you ask AI, you know, what, what is the, what faith community should I be a part of? It says my faith community. And I saw people post that who were Baptists, Catholics, Adventists, uh, you know, Muslims, you know, Buddhists.

Oh, really? So AI is just agreeing with you. And that doesn't, that seem a little convenient, you know, like it just, once you realize the biases that are in place, you become skeptical of that. It becomes, it becomes more useful to you.

But yes, I, I agree that the data set of the internet is way too big for us to trust the outputs because there's a lot of content out there that I certainly would not want to be influencing the decisions. And so what you, what we need to then be able to do is that that degree of curation. And I think that curation aspect of it, that is really a hallmark of this era of technology, because we went from a time where information was expensive and scarce. We came to a time where information is now abundant, but we haven't really figured out how to vet it well or how to uh, condense it well.

Like that's, that, that curation element is now becoming even more important than it ever has been before. So yeah, I agree that guardrails are absolutely necessary. And I think what we will see is the ongoing hyper specialization of AI where the data sets are limited to good data, because that's the only way you're going to be able to get increasingly better outcomes. And yeah, I think that's going to happen more and more, but I think it's especially happening. right now in the context of like, you know, faith technology.

Right. And one of the things that you were talking about, when can you trust AI for advice and things? Yes. An encouragement that I would give people is when you ask AI something, ask it to cite its sources, like make sure that you are able to list out where it found its information and take some time to go through that so that you're not just taking... whatever the internet says about mushrooms or about the Sabbath or anything in between that you are being intentional with how you're using AI, give it it, do your best to give your AI guardrails.

But I will say that is one things I'm really excited about with Ezra, this project that we're working on with Amazing Facts, because it is giving it that quality database to work from. Yes. And I think it'll, it'll be true for so many organizations that are dealing with this technology that content is becoming the moat. If you have good quality content, if you have good quality, you know, uh, libraries of content that you can pull from, then you can really leverage AI.

AI as a producer is, is sub par at best, but when you're using it to leverage good information, then it becomes extremely powerful. So I am really excited, Adam, I'm really excited about where this conversation goes, because this is the series has we've been building on it. And I think this particular episode has been one of the most practical episodes that we have that we've done so far. And I'm excited about where we go from here.

So Just as we as we wrap this one up, I just want to encourage anybody that's listening. As I said earlier, if you haven't heard our earlier episodes, I encourage you to do that. But if you are not subscribed, please subscribe to the podcast because that helps all of the platforms know that this is valuable to listen to and that more people should hear it. And also make sure that you don't miss out on the episodes that are coming up.

If you have an organization and you're interested in how technology or AI in particular can serve you, BlueVignette is our organization and we're very happy to have a conversation about ways that we might be able to serve. You can find us on the website, that's bluevignette.com, but also we're around on all the socials. But with that, we'll leave it there and we look forward to seeing you guys for the next episode.