Jerlyn O’Donnell, Director, Experience Design at FCB Health New York and Franklin Williams, EVP, Director, Experience Design, AREA 23 join CI Conversations host Jennifer Sain to discuss the plus sides, the down sides, the uncertainties and the myriad possibilities of generative AI.
For more marketing insights and ideas, please subscribe to this podcast or visit intelligence. interpublic.com.
For updates on CI’s podcasts and Thought Leadership, be sure to follow us on LinkedIn and Instagram and subscribe on Apple Podcasts.
Jerlyn O’Donnell, Director, Experience Design at FCB Health New York and Franklin Williams, EVP, Director, Experience Design, AREA 23 join CI Conversations host Jennifer Sain to discuss the plus sides, the down sides, the uncertainties and the myriad possibilities of generative AI.
For more marketing insights and ideas, please subscribe to this podcast or visit intelligence. interpublic.com.
For updates on CI’s podcasts and Thought Leadership, be sure to follow us on LinkedIn and Instagram and subscribe on Apple Podcasts.
[00:00:00] Intro: Welcome to the Collective Intelligence Podcast from IPG. We deliver marketing insights that help modern brands thrive. In this episode, you'll hear about the latest perspectives featured at intelligence.interpublic.com. Listen then log on to find new opportunities for your brand to stand out.
[00:00:21] Jen Sain (Host): Hello and welcome to the Collective Intelligence Podcast. I'm your host, Jen Sain, and today we'll be talking about generative AI, a topic that is hard to escape these days, and for good reason. It has the potential to change the way we create, seek information, think about commerce and healthcare, collect data, and the list just goes on. There's quite a bit of curiosity and of course, excitement around gen AI. What is it? What can I create? Can you help me with my March Madness bracket? But there's also skepticism and even fear. Will AI replace humans in some capacities, particularly in the workplace? What does this mean for art and artists? There's so much to dive into and to do just that. I am thrilled to be joined by two thought leaders in this area, Franklin Williams and Jerilyn O'Donnell.
[00:01:10] Jen Sain (Host): Welcome, welcome. Would you both please take a moment to introduce yourselves? Jerilyn why don't you start?
[00:01:17] Jerilyn: Hi, I'm Jerilyn O'Donnell and I work with FCB Health New York, part of the IPG Health Network. I am one of the directors of experience design there, and I am super passionate and excited about generative AI.
[00:01:30] Franklin: Awesome. And I am Franklin Williams, head of the Experience Design Department across both AREA 23 and McCann Health New York, both part of the IPG Health Network. And I think that I've had a passion for just making things my entire life. So now the democratization of AI tools for helping me to generate even more than I could before, I think this has just been a natural fit.
[00:01:53] Jen Sain (Host): Great. Well, thank you both so much for being here. Um, yeah. So to kind of, to get us all on the same. I'd love to kind of establish a common definition of generative AI. You know, as much as that's possible. So to kind of just start to orient us. Right? Do you wanna kind of start with that?
[00:02:09] Franklin: Sure. I mean, I think you can go with like the textbook definition. It's training computer models to. You know, be able to use probabilistic means to represent the data that's on there, kind of like that the model has been trained on. And I think that that's fine, but I think generally for the audience, it's like generative AI teaching a computer to be creative, if you will.
We give it a lot of examples, a lot of pictures, a lot of words, a lot of sounds, and then it learns what those things have in common and then we ask it to create. So, That looks, sounds, or reads like the examples that it's seen before, right? So that's how I think about the model or think about generative AI at its core and then how it goes about doing those things well then it, I think it gets a lot more complicated, but that, that's where it'd start.
[00:03:01] Jerilyn: I love how Franklin decided to, to define it because I have this discussion with my colleague Patrick Malli on my side. He's also a very great resource for generative AI and we, we thought, like everyone has a different definition of a generative AI is, but for me it's a set of algorithms like, so you're probably familiar with them. In the, in the news right now, this tragedy, pt, that's my journey. They generate things from training data, so text, images or audio in this case. But yeah, I, I feel like everyone seems to have a different definition. Even the, the models themselves have different definitions of what generative AI is.
[00:03:39] Franklin: It's very true. And actually one of the things that as we continue to talk about AI in general, I think that that lack of definition or like core definition is one of the things that leads to misunderstanding of what it's doing, how it can do it, and so, I think at some point in life, uh, we're at the kinda like beginning, right?
Everybody's just like coming out with something and they're just doing it because they can, which is fine. Uh, but at some point, just like the internet, you'll need to start defining some rules and some standards so that people have a common language or a common way of referring to things, um, having to do with generative AI. And then I think that becomes just a much easier way to consume this as a new part of our lives.
[00:04:24] Jerilyn:Agreed.
[00:04:25] Jen Sain (Host): Well, thank you for that. And to kind of start us off with really getting into our discussion, I'd love to kind of start where we are and that's in the advertising industry. So I would love to hear what you both think about the implications of generative AI for agency work and creating assets and storyboards and, I mean, perhaps even profiling the consumer, if that's, if that's accurate. So, yeah, so why don't you tell me where, where do you see this impacting our industry?
[00:04:51] Jerilyn: So I think it has positive effects. Some of my colleagues will disagree, but I think, um, AI can help advertising agencies create more effective and efficient campaigns and deliver the best results for clients. And there's several ways that it can help.
Some of it is automated work, some of it is personalization. Can improve targeting and also like help us be a little bit more creative, optimizing our creative process. Nothing like it has to be creative. It just helps us with the things that we probably don't wanna waste time on. Um, consuming large amount of data, for example, and we can iterate faster.
So testing things, just understand the effectiveness of the ads that we produce. And also I think, Targeting the right audience or the right messages they can help with with that as well to speed up our process. And so that is less wasteful ad spend. We're always thinking about ways to do that. So I think it's a positive effect on, on in our industry.
[00:05:54] Franklin: Yeah, and, and probably to build on that, I mean, we're already using it, have been using it for quite some time. I would say that there's a lot of talk about omnichannel, omnichannel ecosystems and journeys and all of that stuff within the advertising agency. Not just healthcare, but just at large. And to be able to do that effectively, the use of AI is kind of critical to that. Making sure that we have the right audience targets, personalization, optimization along the journey that is kind of like the bedrock of those ecosystems. So we can easily see how that is beneficial to us because we can't go through every single action that a user takes to be able to. What the next implication should be along the journey based on where they go and where they've come from.
That's just a lot of data processing to do, and if we tried to do it manually, well, we wouldn't get anything. So I think that for those use cases, we're like, yeah, that makes a lot of sense and we can and should use that. Um, I think it's kind of the creative implications that have probably been the talk of the town most recently because you know, you have art directors and creative directors who are like, wait, now the novice copywriter or strategists can create beautiful art that you know is useful. And then on the copy side with large language models like chat, GPT, it’s like wait, so now you can create scripts where in the past you, you couldn't. And I think that, yes, in those ways, you start seeing it creeping in on the kind of like jobs that other people are doing, but in the exact reverse, the people who haven't been able to do this their entire life now have access to be able to do something and express themself in a way that is new and novel to them.
I think that there's, you know, always gonna be that benefit and that, uh, dichotomy of, you know, is it good or bad? It really just matters based on like perspective. But that does help in the advertising agency get more people into kind of like that ideation process and being able to bring their ideas to life in a way that they can communicate it to the creatives that, you know, have and start to generate and, and, and refine those ideas over time for our clients. And then lastly, I would say probably just. We talk about costs very often, and Jerilyn to your point about repetitive tasks, like cost is a big part of that. I'll give you, I'll give you an example, and this is just something that I was doing very recently because I was like planning vacation and I was looking at two different websites for two different hotels and I was curious.
I was like, okay, so I want to know what all of the kinda like fitness amenities of one hotel is over the. So I went to the website and I looked at their fitness page and they have, you know, big, long pros of things and I was like, okay, so I can read through all of this. I can try and find each little item that they have and then put that into a spreadsheet and then figure out that, and then do it again on the other website. Or instead, I copied all the text on that page, put in Chat GPT, copied all the text on the other page, put into Chat GPT, and I said, you know, tell me all of the fitness options. As a list from each one of these hotels. And then it just gave me a list, bulleted list of those fitness options one and then the other, it took me 40 seconds and that would've taken me, you know, five, 10 minutes. So if you just think about efficiency and cost, I can now move on to the next thing. And I'm just, you know, grinding through work. All of those are very beneficial in what we do. We just have to figure out how to operationalize it inside the confines of that ad agency.
[00:09:27] Jerilyn: Yeah, AI, it's just Franklin's point, like the large amount of data that you can't like skim through it identifies patterns and and things like that. So for creative, I can see where the writing could be useful because you can see like what message you just put out in the world. Is it effective? So you can target your audience better, like what colors you're using, the fonts you're using. That's easier detected by AI to see those trends that are gonna go viral, for example, in an ad.
So it's just. A lot faster, like gives us the results right away cuz we reiterate a lot in our work. So just wanting to know like, hey, our ad dollars are being spent at this point, can, can we shift to something else? Like if it's not working and quickly, like AI could help us with that.
[00:10:08] Franklin: Yeah, for sure.
[00:10:09] Jen Sain (Host): So we've talked
a bit about our industry and Franklin, you know, in your example you were looking at a travel. I'm wondering if there are any other consumer categories that you feel might lend themselves more to embracing generative AI or not. And you know, that could be CPG or or healthcare or, or really anything. I'm curious what you think about that.
[00:10:33] Franklin: I mean, I don't think that there's an industry that can't benefit in some way from generative ai that might be a little. In general as a response, but I feel like any industry that I've thought about so far over the course of the last few months, I can see where the benefit would lie. Um, I remember, so we just went out to CES this year and we were looking at some of the new, like just technology, AI generated or AI-based technologies that are out there, you know, for, you know, at-home stethoscopes that will make sure that you're not having arrhythmia or whatever, like great stuff.
But with the advent of all of these amazing new at home diagnostic tools or tools that are at least creating more and more data, the ownership of the like physician to be able to do something with all of that data. Is insane coming from all of these different platforms. Now I know what's happening with your heart and your lungs and your blood and your, like, everything.
[00:11:35] Franklin: And so now I'm supposed to make an assessment about you and God forbid I miss it, right? And then you say, well, you had the data and now you know, I, I didn't catch the cancer early enough. And so now we're talking lawsuits and all of those things. I think that there's clearly a benefit at being able to just turn through massive amounts of data that gets produced by us humans to be able to help kind of like mitigate the onslaught of data that physicians need to go through and instead start surfacing some of those insights.
[00:12:08] Franklin: Hey, with all of the scans that we've taken over the course of the last couple of months, I realize that there's a pattern that looks like this. You. Cancer might be growing, uh, and maybe it's undetectable by the human eye, but a computer might be able to get there faster and that that's already happening.
[00:12:26] Franklin: So I definitely think that where we're talking about massive amounts of data and organizing that, and kind of transforming that into other types of data that are more useful and usable to us, I think that's where you find the power of AI.
[00:12:42] Jen Sain (Host): So I'm hearing a lot, you know, about, you know, data collection and then using that data to make informed decisions or create efficiencies. You know, I had said kind of in my opening remarks that a lot of consumers who don't have the same lens as you both do, as you know, subject matter experts have a level of fear around it in terms of their job. So what we've been talking about sound makes it sound like a really good thing. Of course. And then with the caveats that you were just saying in the healthcare space, but what would you say to, you know, the average person who has these fears that AI is taking over? Are they founded, are they not founded?
[00:13:19] Jerilyn: Well, some of it is, I have to be honest, but I like to look toward the future. While AI might take some jobs, it will create new ones. And I, I really believe that, um, it's just, Efficiencies that's gonna happen between those. And it might take us a while to figure out what that looks like. And of course, like we're, this time is kind of uncertain. Like there's a lot of regulations that happening and, and everything like that, so we don't really know what's gonna happen with AI in the next few months. However, I think people should see a positive when they're using this because there's large companies that can benefit.
[00:13:56] Jerilyn: There's small companies that benefit, the individual can benefit from this. I mean, I definitely am, a lot of my work is super, super efficient now because of my access to these tools. Yes, it is kind of scary. So I, I totally get people who are kind of worried about jobs right now, but I think the positive of the new things that'll happen because we get some time back is more beneficial than, that's fair.
[00:14:18] Franklin: I think that it's very easy to focus on the things that you will lose versus the things that you'll gain, because the things that you'll lose are very clear because you're doing them today, right? You know about the jobs, you know about. You know, if self-driving cars become a real thing, then what's the point of having truck drivers to drive, you know, for 20 hours at a time across the country to deliver something that makes no sense.
[00:14:41] Franklin: And so you can see why that would be lost, and you can see how, you know, uh, that industry may be changed by it. It's harder, and it just always is the case. It's just much harder to see the benefits long term and how many of those benefits there will be. Because people innovate on those benefits all the time.
[00:14:59] Franklin: Something that you don't think about today, you realize, oh, this community no longer is doing that job. So now we have the opportunity to tap into them to do something completely different that you just may not have seen because that. Just didn't exist. That ability to tap into them that way didn't exist.
[00:15:16] Franklin: So for me, when I start thinking about, you know, the change that will come, and especially the areas that are affected, I mean, I think back to like the Ford model T, right? Okay. So now we have this car that unseats all of these horse-drawn carriages, and you're just like, well, I guess all these horse-drawn carriage drivers don't have a job, but guess what? Mechanics didn't have a job, right? In that in the same way. So now we have this whole new industry that starts getting created and then mechanics turn into race car driver mechanics, they're building cars and just like that, you know, we have new industries that emerge out of it. So it's not a, because this happens, everything is like a negative. I think it's just. We have to see where it goes and we have to do it with our eyes open and be smart about it because you know, everything has its positives and its negatives, so we just have to be clear that we're trying to push as much as possible for the positive.
[00:16:09] Jerilyn: I love that example so much. That's a perfect example of that. Can you imagine we're still using horse drawn carriages?
[00:16:17] Franklin: I absolutely. As I sit here in Manhattan, like 17 cards, zip buying like the last four seconds, I could go for a little bit more slow of a pace sometimes.
[00:16:28] Jerilyn: Yes. Tranquility cannot be overrated.
[00:16:32] Jen Sain (Host): So not to kind of focus on the quote unquote negative implications, but kind of thinking of it from me, you know, particularly, well I wouldn't even say consumer, you know, there could be professionals who have this concern as well, but that's kind of, I guess, content authority and identity. Like, you know, who is the artist or what is the artist if this can be generated by AI or you know, just kind of thinking about ownerships and the, and the definitions of what an artist or creator is. Do you have any thoughts about that?
[00:17:03] Franklin: The, the short answer is, I don't know. And I think, I don't know, because we don't know as a collective, you know, culture of what ownership means. If things are trained on, you know, data that is real, ownable data from other people, and then I'm creating approximations of those, you know, those works through AI. Shouldn't that be attributed to the person who created this style in the first place? And I think that there is absolutely merit to that in being able to do that.
[00:17:32] Franklin: But what I, I, I think the challenge I have just personally is I think that if you create too many guardrails, too fast, then the applications of the technology never get fully realized, and so you end up with something that gets stymied in its growth before it really has a chance to shine. Now, I know that the exact opposite side of the coin is.
[00:17:59] Franklin: If you grow too far, too fast, you might blow up the world, you know? And not necessarily like reality, but figuratively do that in a way that you can't come back from. And so I do think that growth needs to be measured and growth needs to be like constantly monitored and we need to have a legitimate feedback loop in what we're doing.
[00:18:21] Franklin: I think. Uh, so Adobe has their new product, Firefly, and something that I love about how they're going about it is they're saying, Hey, we're doing all of this very measured. You can't do everything today out the gate. You can do some things. You can't do everything, and you can definitely give us feedback about if the things that you're seeing are good, bad or otherwise.
[00:18:41] Franklin: And I think that that's very important. Let's get people to be part of the process for how we grow this new thing so that we all feel comfortable and confident because I would say like the skepticism in whether or not we're doing it right, A lot of it stems from people just not knowing, not understanding what is happening in this space.
[00:19:02] Franklin: And you hear a bunch of things, especially from news networks and depending on what you listen to and how you listen to it, it's, you know, it can be very frightening and that's hard. So I think becoming very educated on this, if this is something that's interesting to you, it's probably the first step in deciding where ownership should lie. And then also where this should evolve over time.
[00:19:23] Jerilyn: Yeah, I a hundred percent agree with this because I still feel like this process has human oversight. A lot of these models are not a hundred percent accurate. It is trained by people. We all have biases it, and it reflects on that. So I still, I think the process is still early and when I'm using these AI tools, I still think I'm kind of the owner be, well, mostly because I'm coming up with the ideas.
[00:19:45] Jerilyn: I'm, I am inputting it in, but I feel like it's still like a collaboration. And if I ignore AI, I still use tools in my real life for my work. So I am the one producing the work. I'm still the artist. And that's the way before I will use any AI tools. Like I might get research, I might cop things together.
[00:20:03] Jerilyn: I'm putting this whole process together. I am creating, so when I use a tool, I'm still bouncing ideas. It's never the final for me. The AI generation is not the final work. There's still sometimes retouching that needs to be done. If it's artwork, they're still refining in the words. Um, I'm still reading through things that Chat GPT gives me.
[00:20:21] Jerilyn: Sometimes it's inaccurate. I'm still fact checking to make sure it doesn't give me something wrong. So yeah, this conversation about ownership is gonna be very interesting again in upcoming months, in the years that, that this exists. But you still need humans. You still need humans to, to, to make sure this is accurate.
[00:20:36] Franklin: Yeah, I would say the one thing, and this is just more with large language models than like generative AI art. When you think about the Chat GPTs, something that concerns me that I think that we just all have to be cognizant of is, I would say that things that get written by Chat GPT sometimes get just accepted as truth.
[00:20:58] Franklin: And so being very aware of what you don't know and doing what Jerilyn you just said, the fact checking. Because it's one of those things, right? So I, I come from development background and so understanding, you know, creating of a, of a product through a programming language is one thing. And you can use Chat GPT to be able to create something, to be able to like build a website for example.
[00:21:22] Franklin: Now the beautiful thing about that is if you ask Chat GPT to write some code for you and it writes some code, you can run that code. And if it doesn't, It basically checks itself and it says, Hey, well the thing that you brought me was not right. And I think that that's kind of like a clear parallel to the fact that what will happen in regular language as well, not just programming languages, the things that you expect to be truth, if you don't know it's truth, don't just turn around and say those things because you might be saying things that you know, based on the training data or based on how you asked the question or how you led the machine might give you something that.
[00:22:02] Franklin: It's a proper response, but not a factual one. I would just say make sure that that part is part of your interface with specifically the large language models more so than the AI generated art.
[00:22:15] Jen Sain (Host): Just to go back to something that we were discussing earlier, you know, frankly, you had said that, you know, you used that great example of, you know, the Model T and then turning, you know, from horse drawn carriages and how it created jobs and roles that were unexpected, like mechanics and manufacturing and all that good. So not looking so much as broad, you know, jobs and job creation, but for individuals, do you see implications for generative AI, for example, disabled individuals or other communities or groups? Any thoughts about that?
[00:22:45] Franklin: I absolutely do. I mean, I think that, If we can create the input, whatever that may be, if we can find ways to make sure that the differently abled can literally get information into a generative AI system then they then have access to all of the benefit that this can provide. And those benefits, I mean, are so varied. I, I feel like I could just start rattling off a list of, you know, if I can't speak and I wanna be able to actually. Put my thoughts into words using eye tracking that I might be able to create some sentences, but to get the feeling, the emotion behind that that may not exist without being able to like create an image that is representative of the thing that you want to say as well.
[00:23:37] Franklin: So, there are so many like implications for being able to say more, do more by putting in a little bit of effort into the thing. And I think that that's probably one of those, those elements that I almost imagine, you know, using like a Chat GPT being able to say something very. And then asking the model to, you know, make this more eloquent, for example, and just being able to like get that version of you or maybe training the model on your voice, your tone, your ability to say things the way you want to say it, but you don't have the ability to be able to make that just more kind of efficacious for you in delivering that message.
[00:24:21] Franklin: That's amazing to me because it helps to, to keep your sense of self and who you are. And I mean, I'm not sure that you can put a dollar value on that for them. You know, we probably can in, you know, in in industry, but for them, I think that the values are measurable.
[00:24:40] Jerilyn: Yeah. I think generator part AI has so much potential to help stable people. Like when I think about like improving lives, cuz we do focus a lot on accessibility as experienced designers and just being able to customize things for user's needs. Like, I think there's so many possibilities, like assistive technology, for example. So maybe it is voice recognition to, to help people with mobility impairments, um, to control the devices with their voice.
[00:25:06] Jerilyn: I think of things like prosthetics. Um, if you think about it, we can probably use generated AI to help like create prosthetic limbs that are more comfortable, durable, functional, getting that data like, and constantly improving on it, help design it. Also, I think about communicating like we, we are able people right now on this podcast, but maybe there's predictive software, like predictive text software that anticipate what the user is gonna say or talk.
[00:25:35] Jerilyn: And it can suggest words like, actually Gmail does that a lot like, and now I'm writing emails and it's completing my sentences. I'm like, oh yeah, that sounds about right, that's what I was trying to say. So based on their input, like it just helps them work faster. And I think about also, maybe automatically generating alt texts. Uh, we, we do, we tend to do that a lot for our emails, for web, um, so maybe generally generative AI can see what's in that picture and generate the right caption for the user, and it can help read that aloud to screen readers. For, like, for people with visual impairments. I think also about virtual assistance.
[00:26:10] Jerilyn: We, we do a lot of these here, so they're designed to assist people with cognitive impairments. So maybe you can provide reminders or prompt them for the task that's gonna happen in their lives. So, yeah, I, I only see benefiting people and hopefully people listen to this can, can understand that it's not all like a scary world. There's so much potential for us, especially in our field to help the user.
[00:26:33] Franklin: So with that, I will just add. The key to all the things that Jerilyn and I just said is access, and we need to make sure that those communities have access to those tools and that we, as you know, the folks who are building and iterating on those tools are doing that with those communities in mind, because if we're not creating tools for them, then. I mean it, it's not great and we won't be able to actually get the benefit of what tool those tools being used in those communities looks like, cuz I think that the benefit will be larger than the communities that they serve.
[00:27:13] Jerilyn: Yeah, I, I agree. And again, I love that you said, you talked about that Franklin again, it's, it's only based on the type of people using it. And that's why I want everyone to try at least get to understand what this technology does because someone is going to use it. I mean, I'm gonna use it. So am I the one who's gonna be training this models? And that's what ends up making them bias. So just make sure. Yeah. Like people have access and we train them well.
[00:27:42] Jen Sain (Host): So to kind of go back where we started back to, you know, the advertising industry kind of as, as this, as generative AI, you know, starts to emerge and gain traction and perhaps even reach critical mass, how would you advise your clients, you know, Franklin, you had said earlier you would advise any individual, and I assume that goes for any, you know, business as well to, to research and, and learn and you know, really get a grip and understand what this is. Um, beyond that, what would you think would be a viable and measured next step that you would advise a client?
[00:28:17] Franklin: So for me, I actually think that there are ways and approaches to it. I think it's, it's so broad, right? The, the applications are so broad and you have to tether them to be around what is the business need or the challenge that you think that you have. And then how would AI, whichever version of it you want, how would that impact, influence or change for, change your problem for the better, right? I think about the idea of like force fitting technology. This is an exercise, right? You just force fit a technology to the problem that you have. So if you want to be able to do better attribution models for, you know, whatever you're doing online, to be able to take large language model tool and be able to use it to be able to do that more positively, proactively, what do you want?
[00:29:07] Franklin: Whatever you wanna say. I think. That may work or may not. I think the problem is you have to know what you want your output to be. You have to figure out whether or not, whichever model or tool that you're using is actually getting you to that thing. That's no different than what we've been doing in advertising and all these other industries thus far.
[00:29:29] Franklin: I'm just saying do it with a new tool that's current and something that you may not know how it's going to perform with what we want to do today. But I don't have any specific like, yes client, you should definitely use Chat GPT today to do all these things. The book on that hasn't been fully written, and so what I say today may be completely irrelevant tomorrow.
[00:29:53] Jerilyn: If I focus on the creative part, I definitely have strong opinions on specific ways that the generative art is being done. For example, I might generate an image just for comp, just to get client buy-in, but I will still hire the model and photographer to take that actual picture. That's me as an artist, that's what I do. I use, um, the, like for example, I use Mid Journey a lot. So I use Mid Journey to come up with different ideas and then I will take that as inspiration and that will never, never ever be the final input for me, output for me. Chat, GPT I probably get some, craft, some ideas, but like I I mentioned earlier, I still take that content, I refine it, and then I actually end up using other AI tools like Grammarly to, to make sure the, um, it's, it's right, or there's a plagiarism tool in Grammarly to make sure that nothing I'm using is plagiarized.
[00:30:46] Franklin: Lean to that point really quickly. I mean, how often are we talking about the tools specifically as aside, unless they are very specific to a requirement that either a client or a vendor has, right? So it's like, Hey, use Adobe XD to create this. Generally it's not because they care about the platform, but it's more about the delivery of the assets to the, uh, to the third party vendor, right?
Franklin: So I. At no point are we really going to get caught up in, are we using these tools or these tools to be able to make it better? What if I give you a comp or a mood board, whatnot, and everything's been generated by AI, but you're able to then follow the story more clearly and we can make assessments about where to go.
Franklin: It should literally be irrelevant, right? That's the goal. The goal is clarity of communication between two parties or more, and being able to make the next decision of what we create. And that's okay. So I think, again, I'm thinking about these tools as exactly that tools not the end all be all. You just do my work for me and then I'll sit back and relax.
[00:31:50] Jerilyn: And you know what, I think that's what makes a lot of creatives fearful about this at all. Cuz I, I do hear writers, um, I, I absolutely adore writers, like good writing. And when you put something in Chat GPT, if it's very, very simplified and you, it produces something, it generates something, it does. Have substance, like it's terrible work. Like I'm just, this is just me being objective. So you actually have to sit here and recraft it and set the tone and make sure it's right. Again, I told you that, that's why I bring it into Grammarly to to help refine it and make it better so, I am again, never gonna put something Mid Journey and think that's a final outfit.
This is, this is, now, we're in version five. Version four used to show like the fingers were terrible. You can spot if the, the, it's a generated image immediately. I mean, it's definitely getting better. But even an image I generated last week was, was some food. It looked horrible. It did not look appealing. I would still need to go in and retouch it to make it better. So yeah, we need to consider, That it's not, it's never gonna be the final up and, and as Franklin said, it's a tool. It's just helping me get there. I'm hoping that the client trusts us enough that we're gonna make this look great and cohesive because these tools are not doing that.
[00:33:00] Franklin: I will say this, and I'm just gonna play devil's advocate because it's fun. It is what it is, but version one of all of these. Not as good as version 2, 3, 4, 5. And as we continue to move, they're getting closer and better and all of those things. Right? Hands not so good today. Hands great tomorrow. You know, not being able to make a cabin one day. Now that cabin has snow and you know, is on the backdrop of a mountain, like we are getting better. So I think right now we are really focusing on the fact that, you know, Jerilyn as an author, you understand crafts, right craft in like the art of putting pen to paper and creating prose that really work for the audience that's gonna read them. And I think that right now for some large or most large language models, it doesn't capture that version of craft. It's like the generic version of whatever we're, you know, putting into it. I don't know how that's gonna evolve in the future.
[00:34:00] Jerilyn: I know that it's getting better. Um, me? No. It is an art. It is definitely an art. And, and this is why I'm really lasting that you said this, like, I'm fine. It's very humorous because as I, cuz it's learning, it's learning over time. It, it is definitely getting better when I'm having conversations quote with Chad g p t, it is getting like better and better. Like, I'm like, wait, you forgot this point that I made a few sentences. It's like, oh yeah, yeah, that's right. Let's recraft this and it's, and it does better and better. So Franklin do not scare me like that.
[00:34:30] Franklin: Hey, if, if there's anything that I am, it is a realist, and it's not that I want to, you know, slow down the pace of, you know, progress. I just want to, as always, go into it with my eyes open.
[00:34:44] Jerilyn: I, I did test, um, a tool, um, recently, so I went to the first generative AI conference held by Jasper in February, and Jasper's a great tool. It's kind of similar to Chat GPT, and what you're able to do with Jasper is basically train your entire brand on it. So imagine being able to, like Jasper, understands the brief and then it generates all the content for you. So it'll learn, it'll learn that eventually. It might be the crafting all your emails that you won't have to do anymore. Cuz it understands the tone, it understands your writing and yeah, that's pretty scary, isn't it? But, but you are, right? It's moving to that direction.
[00:35:20] Franklin: Let's be honest. Th those are things that we want our clients want that. If you think about like the efficiencies that our clients are looking for, Hey, we wanna be able to do the same quality work, but in half the time. Those are the benefits of being able to feed some of these, you know, some of this data into these like models because I can imagine the day where you get to feed all of your brand assets that you've created over the course of the last year or two years or whatnot, into just that, like that corpus of data that it can start working off of.
And you can start thinking about, okay, so can you just make sure to. What my style guide is. Boom. No problem. Of course I can. Hey, can you find any inconsistencies in the work that I just created based on the style guides that we know are identifying this brand? Yeah, this doesn't work. This copy is, you know, doesn't feel stylistically correct, like. And here are the recommendations of how to update those. Those are already kind of happening.
[00:36:18] Jerilyn: Yes, I was, I was gonna tell you like, Hey Franklin, didn't, you know this already popped off yesterday? Oh, oh, no. Yeah, yeah.
[00:36:20] Franklin: No, that, that's when I, and, and that's what I'm referencing. So for me, it's, you know, all in beta and early preview right now. But when it's good enough and when we allow access to our data in those ways. The, I mean, the future changes. It changes. I'm not gonna say good or bad, it just changes. Sorry, went off on a tangent.
[00:36:45] Jen Sain (Host): No, tangents are welcome here. No, that's great. And I think, and I think everything that you're saying, you know, kind of, you know, weighing the pros and cons and seeing where it's used as a tool, but also being aware that the tools. Perfect and change over time. Um, and I almost think that's where, you know, this concept of generative AI kind of differs from the metaverse because I've heard them in, in the same space and I feel like, you know, everyone was all in and it was the next best thing, you know, metaverse. But then it kind of, you know, not to say it's done or fizzled, but I don't think anyone really got on the same page with the definition or really looked in a measured way at the applicability.
So, I mean, and feel, I mean, feel free to argue with me, but I, I'm just seeing that this has more traction in the sense that, you know, I think, you know, while definitions change and you know, can vary from person to person, I feel like there is a commonality and a way I think folks are really looking how to practically implement it. Would you agree with that or no?
[00:37:41] Jerilyn: Yeah, absolutely. Also, I feel like generative AI is more common in our. Existence right now. Like it's been on like late night talk shows, like everyone's talking about it. The Metaverse, uh, I mean, Franklin, we talk, you talked about access. Like I, that's one of the things I was like, well, good luck with that one because I, I'm from a very, very small village in a country that is like not a first world country. So they don't care what, they, don't care what the metaverse is. But I can tell you a lot of my family knows what generative AI is.
[00:38:16] Franklin: Yeah, so it it, it's interesting, right, because I was gonna tap on the access thing too. The biggest for me, and I'll say this as a person who is infatuated with like AR, VR, metaverse things, I can see the future that has this incorporated as another channel by which we communicate and connect with other people. It makes kind of perfect logical sense. The biggest issue on that side I think has always been and continues to be the hardware, more so than the idea of it getting hardware, one into people's homes. That's one. Getting hardware that people actually want to use, getting hardware that people want to use for extended period of time. Getting hardware that looks good enough that people want to wear it in front of other people. Like there are all of those, like those are all like real problems with getting people to even and, and like that's not the metaverse, that's just hardware that gives you access to the metaverse-esque element.
So for me, that's its own problem. But I don't think that the metaverse and AI generative. Are actually the antithesis of one another. They can very easily work together. I can see a world that assuming, you know, the metaverse even is not adopted by everybody, but the generative, you know, uh, worlds that it can create. Ideally, if we have the processing power to do so, where your experience and my experience in the same space are completely different. Being able to, you know, speak into reality the thing that you want to see. I wanna see a, a phoenix flying overhead and then generative art and video, create this 3D model, put it into this space, and now it's flying. And I create my worlds not by holding onto handsets and creating these new things, but by literally speaking them into reality. I can see that world and I want in on that so badly, you know? So I think that at the end of the day, there will be a merging of those things as a person who, you know, grew up on video games. Video games were very much just. You know, the Marios or the Prince of Persian, it's very linear, but then you get like the no man sky and they have generative worlds that are being created that literally just as you continue to move forward in space, these worlds are, are made on the fly. Like there's, there is a world that we are going to live in that I can see all of this kind of merging and melding together to just create things that are so experiential, so immersive and so accessible to people that, you know, people or some people will try and live in those spaces.
[00:41:02] Jerilyn: Yeah, because, and you're right cuz I, I also, I mean while we were talking, I was thinking also about the type of user, so accessibility. I, I do think about that a lot. So is that, are those hardwares accessible? Yes. By, you know, like the person being able, able afford it, but also the person being able to use it.
[00:41:20] Franklin: Hundred percent accessibility needs to be the kind of the underlying element that bonds all of the things that we create. Because if it's not accessible to everyone, it's just not fair. No one's gonna use it. So embody in on, on these experiences moving forward, you know, everyone can love or hate it, but be able to have an opinion.
[00:41:41] Jen Sain (Host): I actually think that's a perfect note to end our conversation on. I could talk to you both endlessly. This is fascinating. I love your thoughts on this. Truly, really, this has been such a great conversation. Thank you both so much for being here.
[00:41:50] Jerilyn: So I do have a question. Do you fear? Or are you embracing it?
[00:41:56] Jen Sain (Host): Me?
[00:41:57] Jerilyn: Yeah.
[00:42:00] Jen Sain (Host): Oh, I have to say it changed from this conversation, and I'm not just saying that I do think I got, you know, kind of, I bought into the media hype of AI's coming for your jobs and we're gonna be ruled by the bots. But I feel after this, I mean, just hearing, you know, both of you just articulate so well, both sides and such a measured approach. And also that the beautiful application of it, both from creativity and imaginings to, uh, you know, access and just kind of democratizing creativity and tools. So, you know, I, I, I definitely have more of a measured hopefulness than before we started this conversation.
[00:42:36] Jerilyn: Fantastic.
[00:42:37] Franklin: Well, it's a good day.
[00:42:38] Jerilyn: Yeah absolutely.
[00:42:40] Jen Sain (Host): well, thank you both again, and I hope we get to talk soon.
[00:42:42] Franklin: Absolutely. Our pleasure. Thank you so much.
[00:42:46] Outro: Thank you for listening to the Collective Intelligence Podcast. For more marketing insights and ideas, please subscribe to this podcast or visit intelligence.interpublic.com.