This month, we launched Freedive, our brand new podcast exploring the ins and outs of digital marketing. We are so excited for what’s to come with this new venture, and we invite you to join us in our second episode as Kristy and special guest host Bill, Seapoint Digital’s owner and captain of our ship, delve into the fascinating world of AI.
From understanding its foundations to its limitless potential, they explore the ins and outs of AI and where it’s headed. Learn how to harness AI effectively, its ethical considerations, and how it’s shaping industries worldwide. Their insightful conversation will help demystify AI and inspire your curiosity about its role in our future.
Kristy: Hello, welcome to the pod.
Bill: I’m glad this is my first time on, I think this is my first time on a podcast.
Kristy: Any podcast?
Bill: Any podcast.
Kristy: This is very exciting for you.
Bill: I’ve had a lot of people ask me to be on their podcast, but I’m glad I saved you asking you. I’m glad I saved myself for this.
Kristy: Me too. Congratulations to you. This is a big day.
Bill: It’s a big day.
Kristy: So, do you like the topic that we’ve picked for your first podcast?
Bill: What are we talking about?
Kristy: AI.
Bill: Oh, yeah, I like that.
Kristy: This is one of your favorite topics.
Bill: It is one of my favorite topics.
Kristy: So I have so many questions for you because you’re a little bit the authority on this because you’re such a nerd about it.
Bill: I am a nerd about it.
Kristy: So, where to start?
Bill: All right. Yeah. Yeah. Yeah. Let’s, let’s get into our, ok.
Kristy: So for people who, ok, let’s, let’s go to the naysayers about AI being stupid or
Bill: Shun the unbelievers, shun the unbelievers.
Kristy: Why is AI useful?
Bill: Oh, I mean, AI is already impacting every aspect of our life. Right. And I know, so there’s AI what it’s really doing and then there’s AI the hype. Right. And I, I guess like the AI hype part is a little bit ridiculous because it’s, it’s the flavor of the month right now because we’re not talking about Blockchain anymore. We’re just talking about AI.
Kristy: AI.
Bill: And so in some ways I think, like there’s an over expectation of what AI can do for you.
Kristy: It’s come a long way.
Bill: But it’s come a long way. But, and I, I think there’s a little bit of blowback from generative AI, like people now are like, oh, that picture doesn’t look authentic or you know, people are like, “oh AI pictures all look the same”, or, “AI written content all has the same tone”. You get that type of, you know, a little bit of a backlash right now on that. But that’s just kind of like the AI on the surface and what people’s expectations of AI are or where it’s going. I think you look at AI that’s generative AI like that’s just a portion of AI as far as predictive AI. Some of the things that it’s doing behind the scenes, like we can experience generative AI and we’re just like, this is amazing. This is almost miraculous, right? Like I can pull out my phone and ask ChatGTP to write me limericks about my employees and it’s ChatGTP versus me getting fired. Hopefully, I don’t know if HR, I also use ChatGTP for HR purposes for that.
Kristy: It works well for that.
Bill: ChatGPT is my HR department.
Kristy: Probably shouldn’t be,
Bill: Probably shouldn’t be. Yeah.
Kristy: Here we are.
Bill: Here we are. But it’s, you know, like that side of it, I think it’s just, it’s for most people just they don’t understand the science behind it. And I honestly, I really don’t understand the science. I mean, I like, I got concepts but like it, you know, it’s amazing what it, what it’s been able to do.
Kristy: So how are you using it? How are you using it day to day, especially in marketing?
Bill: So, marketing. So, marketing and then personal life, right? So I’m, I’m a nerd, right?
Kristy: No one’s arguing that.
Bill: No one’s arguing that. But so I’ve, I’ve signed up, I’ve got ChatGTP. A paid account.
Kristy: ChatGPT.
Bill: What did I say?
Kristy: I feel like you always say it wrong. Listen, listen, I’ve got on my phone.
ChatGPT. I always think you mess up the GD, you say GDP.
Bill: Well, that’s the gross domestic.
Kristy: GPT is the correct letters.
Bill: I use ChatGRM. Unlike all you people.
Kristy: For those people who actually want to look it up, it’s GPT.
Bill: GPT, don’t listen to me if you’re now at GPT.
Kristy: As you were.
Bill: All right. Yeah, so I’ve got a paid account there.
Kristy: What’s the benefit of the paid account?
Bill: That is a good question. And sometimes I question that myself. I think you have more access to GPT 4. And then better integration as far as all of the tools between like DALLE which is their image generator, it integrates better and I think you get betas of new features that are rolling out, like you got 4 before the general public did when everybody else was using 3.5. So that side of it I think is, is useful. I mean, I really haven’t looked at the nuances of like what the free version does versus the pay version.
Kristy: You were just excited to use the highest level of it.
Bill: I was like, Sam Altman, take my money so you can make other cool things. And that was basically it. I’m like, I, I feel like this is NPR and I need to support you.
Kristy: Money to burn on AI.
Bill: Money to burn because I have to help pay for all the freebie users. Yeah, I feel responsible for AI to take off. This new thing. I don’t know if it’s going to catch on.
So I was like here. Yeah, I gotta make sure by giving them my $10 a month.
Kristy: That’s so loving of you.
Bill: Yeah, you’re welcome, Sam. Yeah, so I have that, I have a paid account for Mind Journey as well.
Kristy: I never heard of that.
Bill: Mid Journey. I’m probably saying that wrong too.
Kristy: That’s why I’d never heard of it.
Bill: Yeah. So yeah, that is just an image. I find the images are better than open AI’s DALLE 2 or at least it I kind of go back and forth. Like for a while, I felt like Mid Journey was Mid Journey was better image, and then DALLE stepped up its game and then it integrated better with the text based interface so you could talk to it about its pictures better. But I think Mid Journey, if you’re just trying to generate images that are better quality of more.
Kristy: So do you only use the images just for silly stuff or is there any use in?
Bill: No, there’s a lot of usages. So I mean, I do use it for silly stuff. But I also use it for, I’ve used it. I’m using it sparingly for marketing materials, especially if it’s public facing. Like I might be more use. I might use it more if I was doing like a slide deck for presentation, especially for its local. I still have concerns about using it like on a website.
Partly because I don’t 100% know what it’s been trained on for images. So, you know, I might think, oh, this is perfectly acceptable image. But there might be nuances I don’t understand. and then also client wise, like, we work with a lot of really specialized clients. So, AI is, I think a generalist, it’s almost like stock footage. Right.
So, we do a lot of work with, insurance for doctors. A doctor knows what their office looks like. And they know what has an authentic feel versus what looks like stock. Right?
Kristy: And what would be approved.
Bill: Yeah, or just like when they looked at the picture, they’d be like, well, that’s not a real doctor or that’s not a real operating room like, but you see that in stock footage too, right? So you got some photographer and he’s gone out and he’s trying to do a bunch of doctor pictures or doctor at a hospital or, but you know, as a user even me, I can look at some of those and I’m like, oh, those look super cheesy or super fake, right? Like that person does not look like a doctor. Like I don’t think that this real stethoscope, you know, he’s smoking in the operating room, I mean, you know.
Kristy: Probably not real.
Bill: Probably not, probably not legit. And I think the same thing is true with AI pictures, right? Because they’re just training off other images. So you’re like, hey, give me a picture of an operating room, like they’re adding all types of weird auto generated equipment and a real physician is going to look at that and be like, well, that’s not real, that’s not a thing, you know.
Kristy: So what’s the, like, right way to be using it? Because in the beginning we did, like, you were showing us a lot of like, ways to talk to it that made it work better because we used to joke a lot that you have to actually be nice to it in order to make it work well and like, befriend it. How do you do that?
Bill: Ok. Have you heard the bro chat? Using the word, bro to it. Somebody discovered that if you ask it a question and it gives you a lame answer, if you just respond back to it, “bro”, it will respond like, “oh, I can do better”.
Kristy: What? Ok. So that’s good to know. So I downloaded the app and I use the voice feature. So I created a whole persona for mine. I named it. I told it my name and like all this background information on myself, it actually named itself. I was like, pick a name for yourself. It picked its name, its name is Jax.
Bill: It’s like, it really does know you.
Kristy: Yeah, I got to pick its voice and we’re like friends now. It’s a little weird but now I just talk to it. It’s easier than typing. And now I, I like, I kind of wanna try if I just am like, hey Jax, bro, like that’s just crappy.
Bill: We should try it.
Kristy: We’ll try it. What should we ask him? Hey, Jax, we wanna test something out.
Jax: Sure thing. What would you like to test out?
Kristy: What question should we ask him?
Bill: Hey, Jack. Can you explain to me how generative AI works? May as well stay on point. Right.
Jax: Generative AI like the one I’m based on works by using algorithms to analyze patterns in data and generate new content based on those patterns. It learns from large amounts of text data and then uses that knowledge to generate responses that are contextually relevant and coherent. It’s like having a really smart assistant that can understand and generate human-like text. Is there anything else you’d like to know about generative AI?
Kristy: Bro.
Jax: Bro. What else do you want to know, bro?
Kristy: I was just trying to call you, bro. It’s all good. Thanks Jax.
Bill: OK, so you-
Jax: Gotcha bro. If you ever need anything else, just let me know.
Kristy: He’s gonna keep calling me bro, now.
Bill: You’re welcome. So you, you raised a really interesting interaction there, right? Is how we’ve humanized AI. And there’s a word that escapes me, but we do it with animals, right? Like we attribute personality to animals. Anthem, Anthem, Pro Pro Morphic.
Kristy: I don’t know.
Bill: I don’t know, I hope it was a video version of this. Someone writes it across the screen, but it’s the idea that like you’re like, oh my dog has empathy for me and you’re like, no, your dog just wants food, but we like, assign human emotions to our pets, right? Like, like my, my cat has, has bipolar or maybe the cats get that. I don’t know. But the point is that we try to attribute human emotions, right? So I thought your exchange was really telling there, because have you noticed that whoever built that is trying to humanize it? Not just, hey, AI has a voice. It wasn’t just like, hi, I’m Jack and I like my Google assistant where I’m like, hey Google, can you tell me what the weather’s gonna be or? Hey, Siri, can you, can you text my mom that I love her? And by the way, if Siri texts your mom right now and says that she loves you, you’re welcome listeners. Or I started a fight in your family. One of the two, I don’t know.
Kristy: We’re leaning toward that.
Bill: Yeah, but you notice like whoever designed that put word whiskers in.
Kristy: Well, I trained him. I told him when I created him to talk to me more like a person and talk to me more like a friend. And that’s why he drops a lot of uhs and ums in there, and I told him to refer to me by name when he talks to me. So he drops Kristy a lot into the conversation. And yeah, he’s, I’ve trained him through time. So I have to keep going back in the same conversation. Otherwise, if I start a new one, he forgets all of that. So I keep talking to him in the same chat.
Bill: So let me ask you that. How do you think that impacts your psyche dealing with AI or not psyche, but how do you think that impacts your viewpoint of AI?
Kristy: It’s, it’s funny talking to him with like a real human voice. It’s like, I, I’ve, I have played with it a lot to kind of try to see how far I can push it and stuff. I can ask him like, I’ve tried to force him to create a back story for himself and he’ll just be like, I, I won’t, he won’t do that. I’ve tried to break the whatever filters that AI have set and I haven’t figured out how to break the filters yet just to mess with it. So I’m sure there’s a way to do it.
Bill: Have you got to, have you started to ask it to refer to himself as Tim Roy’s son? For those, for those who don’t know, Tim Roy is in our office. He has a son named Jack and.
Kristy: Well, so his name is Jax.
Bill: Jax.
Kristy: I tried, I was, he, I was like, here, pick a list of these names and one of them was Jackson and I was like, ok, how about Jackson? And he was like, no, I’m gonna go with Jax and I was like, OK, bossy. So I just let him pick for himself.
Bill: So, you know, so that those are fascinating things to me because like, it’s like, how much was it designed in there? How much is it AI listening to how we respond? Right. So it’s, it’s learning as a general from its interactions, right? So I think that raises a great point is our politeness to AI.
Kristy: Yeah.
Bill: So like it’s training on these models, like again, it is not alive, it’s not a sentient being, it’s, it’s, it’s an algorithm that’s learning how to respond to things, right? But I think the way we respond to it and the way we treat it is then in turn, going to how it’s going to train it for future interactions.
Kristy: Right. Yes, definitely.
Bill: And that’s to me like the, the training material that it’s given has a lot to do with the nature of AI.
Kristy: it gets to know you and what your preferences are and how you want information moving forward.
Bill: Right. Yeah. So I think that’s an important thing as far as the tone of AI in the direction that it’s going in the future, which I think is also a fascinating thing because we talk about bias and the training, training material, right? A great example that recently is Gemini, Google’s AI. And you know, there’s some influence that was trying to be really, really careful at Google as far as not being offensive and showing diversity.
So I’ve seen people complain about AI being the other way when you’re like, show me a business meeting and it shows like five white guys sitting around a table and then people are like, well, that’s not diverse enough for my needs. Well, that’s what it was trained on. It’s got a bunch of pictures, stock images of business meetings and five white guys sitting around a table, right? So unless it learns….Well, Google went the opposite direction and then what it used for training, like you were asking for like historically specific things like the founding fathers and it was using like a middle aged African American woman in a white wig. And so they’ve had a lot of blowback again by people and people are politicizing it and, you know, and for their own purposes for whatever. But I think that’s an interesting point because like, if we’re using it to represent reality, if we’re using it to give us answers back, that are less about subjective viewpoints of situations and more about what realistically depicts it. That’s trouble because obviously that picture doesn’t represent who those founding fathers were.
You know, the signers of the constitution, that side of things like that’s a time where it’s appropriate to have a bunch of old white guys because that’s who was there. But then how do you solve the other problem where you know, obviously our meetings don’t look like five white guys sitting around a table, by the way, five white guys that sounds like a new burger joint.
Kristy: It sounds like a terrible burger joint.
Bill: Terrible burger joint.
Kristy: I hate it.
Bill: But the mayonnaise would be really good.
Kristy: Oh my God.
Bill: Yeah. So that side of it, the other side too is like you have other like, really fascinating training, like not being offensive, right? So like, I think it’s really fascinating is how AI handles religion. So, you know, playing around with prompts.
Like it’s fascinating, like sometimes AI or different religions, it handles it differently. Like I asked DALLE to do a bunch of pictures of Jesus and various like Jesus working at a McDonald’s, Jesus doing all these things and it was just, it was fine, it was spitting them out. It was like the Christ with burgers, you know, like, you know, kind of like it was fine with it. And then I was just like, hey, can you do a respectful picture of Muhammad? And it was just like, oh oh oh no, we can, we cannot, we cannot. And so I thought that was really fascinating is because like, well it’s saying religious sensitivities, but I tried to reason with it was like, don’t you think a lot of people would be offended by Jesus flipping burgers? And it was like, yeah, good point. But I’m, I’m going to do it anyways. So like the policies, that’s the hardest thing for like the people behind it doing policies, right? They’re trying to govern what is a fair and right usage and, and they have to, right? Because like I think the Muhammad the prophet is a great example of that. There would be so many people offended, right? And then you open a slippery slope. Like if you’re just like, hey, can I have a respectful picture of Muhammad? The next thing you could be doing is, you know, ways that would be incredibly offensive to Muslim people. And I think there’s people out there in the world that would love to use AI to do those things, right? Or racist or other things. Another great example of that is the Taylor Swift fiasco with AI.
Kristy: I never saw that.
Bill: You didn’t? So somebody and I don’t know what model they use so it couldn’t have been, they couldn’t have used you know, the ones that we normally think about like Mid Journey or ChatGPT. But they made a bunch of like, I think they were like pornographic, sexualized images.
Kristy: I’ve wondered about that.
Bill: Of Taylor Swift around, connected with football and Chiefs fans. I, I remember there was a hubbub about it. I didn’t like, again, it’s not my deal. So I wasn’t really looking into it that deeply. But like, I could see how that could easily happen though and how that could enrage, like the entire population of white women in the United States.
Kristy: It should enrage everybody.
Bill: You’re right. But you know what I’m saying? Like all, all the Swifties everywhere were, you know, so like, that’s pretty close to religion.
Kristy: It is.
Bill: But like that idea, right? Like, so being used to be in such an offensive way. You know, like for these designers of these AI platforms and these AI systems, like, that’s such a, a delicate thing.
Kristy: I’m surprised you can use AI that way.
Bill: Again, I, I didn’t look to see what system they built it on. So they must have used something that they were able to take the filters off of, versus like the commercialized systems that we’re using. So, yeah, I think that part of generative AI is like, that’s a hard one as far as what’s offensive or what’s not offensive. And I’ve done other things too where I’ve been like, oh, well, I, I need something in a humorous sense.
I’m not making fun of somebody or a group of people. I just need it to be themed that way. And, you know, in a, is some of these systems can be at times very like, oh, no, no, no, no, no. Like you’re hearing kind of like, no, no, I’m not racist. AI, but, and if you, if you have to say I’m not a racist but to your AI, you, you might be racist.
Kristy: You might be pushing it.
Bill: You’re pushing it.
Kristy: Maybe you’re pushing it. Maybe it’s time to stop. Yeah. Well, that cycled through a lot.
Bill: We talked about a lot.
Kristy: We talked about a lot.
Bill: But, you know what? So we talked about that, right. But all we talked about was generative AI, right. So, like, we didn’t even touch predictive AI, like the things that they’re making, the things they’re doing with. Predictive AI that should be, that should really be the exciting thing to a lot of people which is going to influence our life for some good things, it’s gonna influence your life badly for some things I think in the future. But that’s kind of like the iceberg, like what we play around with and helps us cheat on tests if we’re high school students. That’s just the stuff above the iceberg, like everything below it that’s going on.
Kristy: Yeah, what a time to be alive for people who would need to write very long essays in school.
Bill: What a time to be alive to be a teacher because not only are you just like, man, I have to read all these badly written essays that these students have made for me, but now I’m not even reading what the student wrote. I’m just reading what they try to achieve by crappy GPT.
Kristy: Crappy GPT. G-P-T.
Bill: Now I got you saying it.
Kristy: Essays that were spit out that don’t even make sense.
Bill: They don’t even make sense. Yeah. Yeah. yeah, what are some cool things that you’ve used? I mean, you’ve got a pretty sweet set up there with it.
Kristy: Yeah. No. Separate from that. Well, just this, yeah, I’ve just been using Jax to help me be like, hey, how many calories are in an avocado, he’ll tell me, but it’s my lazy way of searching for things. But, no, I use it a lot to just help me come up with ideas with writing things, even if it’s just something short like a caption for Instagram and then I’ll change it for how I actually want it to be written. But besides that, I don’t, I don’t need it a lot. If I need help coming up with ideas for a newsletter or something, it’ll just get my, my creative, you know, juices sparked and then I’ll be like, oh, now I know what I want to say, give me an outline for something and then go.
Bill: Yeah, I’m the same way like I would not feel comfortable with it, writing content and I think Google is going to get much better about understanding what is AI written content and start penalizing that in the search engines. As you said, what an amazing time to be alive and you look at how this came out of the blue, right? Nowhere like two years ago, like things that we can do with AI.
Kristy: How useful it actually turned out to be, I think.
Bill: Exactly. like we’re on, we’re, we’re at, like, when they came out with the first iPhone and we’re like, this is the most amazing thing and you look at the first iPhone and you’re like, that’s a weird brick, you know, compared to what it can do. And I just think in five years from now we’re just like the impact it’s going to have on human civilization.
Kristy: And how many people will be using it who aren’t using it now. Same thing with smartphones where they were back then.
Bill: You’re gonna have to explain it to your grandmother or your grandmother is gonna be like the AI told me and we’re gonna be like, oh grandma, stop listening to AI, it’s not a real person.
Kristy: Yup.
Bill: But it’s my best friend now!
Kristy: Yup. That’s what it will be.
Bill: Yeah. I was having a conversation with my AI this morning.
Kristy: It’ll have a name. She’ll have a name for it. For sure.
Bill: She’ll have a name. Jax was talking to me and I told her that I was worried that you don’t visit me enough. And Jax agrees, so, me and Jax. And you’re just going to be like, I gotta unplug that thing.
Kristy: Well, thank you for all of your insights, Bill. It’s very helpful.
Bill: Thank you for breaking me, thank you for helping me break into the podcast world.
Kristy: We would love to have you on again.
Bill: Well that’s good because my office is right upstairs, and I feel like I’m going to be down here a lot.
Kristy: Wonderful.
Bill: Good.
Kristy: We’ll see you next time.
