We translate the language of AI on our third episode of Think Fresh with our first guest, Kathleen Perley. Kathleen is an entrepreneur, Fulbright scholar, and all-around AI oracle. We discuss the AI aspects everyone needs to know to leverage it to its best potential–from use cases and structuring effective policies to new tools and budding innovations.     

The conversation doesn't end here! Find us on Facebook, Instagram and LinkedIn, sign up for our newsletter, or send us an email at: info@thinkdenovo.com with the subject "Dear de Novo." 

Links and References:

The Code | A Dove Film | Dove Self-Esteem Project

Other Dove #RealBeauty Ads: 

Dove Real Beauty Sketches

Dove #RealBeauty Is Universal | Celebrate all types of beauty

Open AI + Figure Robotics Demo: 
Figure Status Update - OpenAI Speech-to-Speech Reasoning

AI Tools: 

https://www.perplexity.ai/
https://opus.ai/
https://julius.ai/

Transcript for Podcast Episode 3

The Language of AI with Kathleen Perley

00;00;00;00 - 00;00;20;16
RYAN
We're doing this. 

JEN
All right. Hello and welcome to... Were you talking? Oh, my God. All right, we'll try this again. Ryan, are you settled?

RYAN
I’m settled.

00;00;20;18 - 00;00;48;00
JEN
Okay. 

Hello and welcome to Think Fresh, a podcast brought to you by de Novo Marketing’s Collective Creative. Coming to you from our Ideas Institute and here to talk about all things marketing. Insights on new trends, innovative ideas and marketing tools you can use in your day to day life, and whatever else we deem relevant. I'm Jen Neumann, de novo CEO and your host. 


RYAN
…and I'm Ryan Shenefelt, Account Manager, Innovation and Education Lead, and resident nosy eavesdropper always looking to push the envelope.

00;00;48;04 - 00;01;11;24
JEN
He is indeed very, very nosy. 

RYAN
Today we're going to be talking about artificial intelligence.

JEN
…and we have a guest on the episode. So today's episode might be a little bit longer. We have our first guest, Kathleen Perley, here to talk with us today. 

RYAN
And I have to admit, I was fangirling a little bit during this episode. Kathleen's a wealth of knowledge, and truly one of the leading voices in this space. So, it was great just to talk with her about AI and hear everything she had to say. It was it was phenomenal. 


JEN
So sit back, enjoy this episode and nerd out with us and Kathleen Perley today. 

JEN
Today's guest is a friend of de Novo. She was a Fulbright Scholar. She started her own digital agency, Decode, in Houston, which grew to one of INC.’s 5000 top businesses.

After selling her company in 2023, she's turning back to her love of linguistics and is a professor at Rice University, where she teaches AI in the MBA program. And she is a badass. Everybody, it's Kathleen Perley. 

KATHLEEN
Hi. Thank you so much for having me. 

RYAN
Welcome, Kathleen, our first official guest on the podcast. 

KATHLEEN 
I feel so honored to be one of the first.

00;01;57;07 - 00;02;16;15
KATHLEEN
Hopefully not the last. 

JEN
Definitely not the last. And it might not even be  your only appearance on the podcast. Maybe you become a regular friend of the pod. 

KATHLEEN
That would be wonderful. 

RYAN
Tell us about your background a little bit. 

KATHLEEN
Yeah. So as a, and Jen knows this, we know each other really well right? I can be pretty stubborn.

00;02;16;15 - 00;02;44;29
KATHLEEN
And so I actually grew up with dyslexia and ADHD. And so I was often told that I shouldn't go to college or that I couldn't go to college. And so I struggled with language, and the English language, especially spelling and, trying to figure out how these words came together. And so as any stubborn child would do, I said, I'm going to own this and make it my - word that I probably can't say on this podcast.

00;02;44;29 - 00;03;11;23
JEN
Oh, you can say the words on this podcast. 

KATHLEEN
But just, you know, I was very, you know, I wanted to prove everybody wrong. And so I was like, I'm going to take the lead and I'm going to pick a major that is probably the hardest major I could possibly choose for me as a learner. And I just fell in love with it because language and words and as somebody who struggled with spelling and my dyslexia, ended up becoming almost like a math equation, and I just truly loved it.

00;03;11;23 - 00;03;32;19
KATHLEEN
And so I was doing that for a while. Graduated, ended up doing research with my Fulbright in Spain on phonetics and phonology. And then I had a true mom who was a big New Yorker. And I moved home after my Fulbright, and I had seven or eight months until my PhD program. And my mom said, “if you live under this roof, you have to have a job”.

00;03;32;21 - 00;03;56;22
KATHLEEN
And so I found a job in marketing and advertising. I was told they were hiring, it was an interactive, automotive agency, which I thought meant that I was going to be interacting with people. But no, my first task was to code the company's website. But I fell in love with it. And so I feel like I'm finally being able to get back to my roots again, which is nice.

00;03;56;25 - 00;04;25;03
RYAN
So your your first foray into marketing was website development and coding. 

KATHLEEN
Yep. Yes. 

RYAN
Wow. How did you do any coding before that or just dove right in?

KATHLEEN
Dove right in. In linguistics and a lot of stuff that we do, we do a lot of transcription work. And you end up writing in types of codes and patterns to identify either disordered speech or challenges, and things of that nature with aphasia or some of the language side of it.

00;04;25;06 - 00;04;42;18
KATHLEEN
And so kind of really aligned with what I had done in my previous life. And I just kind of found it interesting and fascinating. And so I dove right in and just did it. I also told my boss at the time, who was paying me a very low salary at the time, that, you know what? I have no clue what I'm doing.

00;04;42;18 - 00;05;09;18
KATHLEEN
I'll give it a shot. And if this doesn't work out, we can like both peacefully separate. But I was there for about two and a half years and loved it. 

RYAN
Wow. Yeah. So then what led you to. When did you start your, When did you start Decode? 

KATHLEEN
Started Decode in 2013. And that one was actually, I was on a plane flying back from a sales pitch, and ended up sitting between these two gentlemen on a flight.

00;05;09;18 - 00;05;37;26
KATHLEEN
They asked me a few questions, and by the time I ended the flight, they offered me a job, and they turned out to be the CEO and president of Ixia, which is a cybersecurity company. They wanted me to lead their SEO department worldwide and I ended up talking to their HR department the next day. And I told them that maybe if I could convince two gentlemen on a plane, you know, for an hour and a half to hire me, then maybe I could try this on my own.

00;05;37;29 - 00;06;01;29
KATHLEEN
And they actually became my first client. Which is wonderful. And they were so amazing. At our larger size, we were about 70 employees, and servicing the healthcare industry and, recently sold about a year ago. And now get to go back to my passion of teaching and sharing and the AI and analytics and linguistic world that I kind of started my career on.

00;06;02;02 - 00;06;25;25
RYAN
That's congratulations on the sale. Very, very much a celebration. 

JEN
It was so interesting to kind of have a front row seat to watching your, your sale of your company and like the emotions that must go with that are, it's just crazy. So I also think maybe we just ought to start a podcast on being ADHD in the marketing world. 

KATHLEEN
Oh, 100%.

00;06;25;25 - 00;06;46;27
JEN
It's it's, it's it can be a blessing. It can be a curse. It's what you make of it. For sure, for sure. So. But thank you so much for taking the time to be on with us today. And this is, for the people listening. This is such an interesting topic. And there are, there's a lot of smoke and mirrors around AI.

00;06;47;00 - 00;07;08;15
JEN
And I think Kathleen is a truth teller when it comes to this. And I think a lot of companies also, beyond using ChatGPT, have no idea how to embrace AI or in some cases don't even know that they're already using it. So, this is a topic that I think Ryan has really, dug into, and is really owning that space at de Novo.

00;07;08;15 - 00;07;25;17
JEN
And so, I am, for the most part, going to take a backseat while you two nerd out on this podcast today. So.

RYAN
Kathleen, what what, kind of started your journey? Why why did you get into AI? Did you start using it at Decode and wanted to keep expanding or what? What kind of made you start the process?

00;07;25;20 - 00;07;47;06
KATHLEEN
Well, so I've always had a passion for it and was doing a lot of what I'll call not true AI, but a lot more machine learning utilizing SQL databases. And really, up until the last three years, AI existed out there, it just wasn't really accessible unless you had lots of money to put behind it. And it really has become more accessible and more democratized.

00;07;47;06 - 00;08;12;13
KATHLEEN
And so we were doing things where we were writing code and scripts and algorithms that would take certain types of databases. One of them was we would take, the, clinical database of physicians at a health system, which is really kind of boring and not really consumer facing. And we actually built algorithms that would take what they got from credentialing and actually turn it into physician bio pages.

00;08;12;13 - 00;08;34;18
KATHLEEN
And so we would understand that, you know, the internal credentialing software might call it, “Oncologic Services”, but the average patient wants to understand,”can you treat my type of cancer?” They use the word cancer and they want to talk about specific types of cancer. And so we actually built code and an ETL process that would dynamically populate all of those things.

00;08;34;18 - 00;08;53;03
KATHLEEN
So while that was more about machine learning and some algorithms that we leveraged from that perspective, that really kind of started to be the foundation of my love of really seeing the possibility of what it could be. And then, you know, I remember vividly when, you know, when ChatGPT was released and, you know, I had my team working on it, like from November of that year. 

00;08;53;03 - 00;09;10;14
KATHLEEN
One of the jokes my dad and I have is, my sister, my whole family is in oil and gas. I live in Houston, which is not surprising that that’s the case.  But I want to say about a year ago. So I was playing around with it for about 18 months.

00;09;10;14 - 00;09;29;11
KATHLEEN
And then a year ago, today, my dad, we were at a family dinner, and my dad was like, “Kathleen have you ever heard of this, like, GPT thingy?” And my sister, who's also in oil and gas and finance - they typically control the whole conversation at the dinner table. I literally looked at my sister and I was like, “take a back seat!”

00;09;29;11 - 00;09;55;14
KATHLEEN
This is mine now.” And I was like, it’s my turn to shine. I'm finally relevant. Like, people finally want to talk about it. And, you know, my my love of linguistics and language has been a lot of the foundation of what AI is running on today. And before ChatGPT, the only time anyone was interested that I was a linguist was when they saw that alien movie where that woman, that linguist, was brought in to help decipher what the alien was wanting.

00;09;55;16 - 00;10;17;23
KATHLEEN
And so I feel finally relevant again, which is kind of nice. Like my nerdiness is cool. 

RYAN
Oh, yeah. Arrival with Amy? Nope. Is it Amy Adams, I think? 

KATHLEEN
Oh yeah. Yes, yes. 

JEN
Oh my gosh, our producer is over here saying “Arrival… Arrival!”

KATHLEEN
Yeah. I was like some alien movie because, like, people didn't know about linguistics.

00;10;17;25 - 00;10;33;11
KATHLEEN
Wouldn't talk about it and then they would see it and they're like, wait, isn't that what you studied in college? And I'm like, yes, it is. Would you like to discuss like the – you know, all that stuff. And so but I kind of feel like ChatGPT and this revival around like the democratization of AI has kind of brought that forward again.

00;10;33;14 - 00;10;57;22
RYAN
And it was interesting you said that you were kind of doing it wasn't necessarily AI, but it was still machine learning but like, you were doing it by hand, you were coding it. So it's interesting, our designers will talk about how they used to mark things up using wax boards, and now they can use Photoshop. It's kind of like how you can now use ChatGPT to not you understand how it works, but you don't have to then code it all yourself.

00;10;57;25 - 00;11;14;27
KATHLEEN
Yeah, it's it's actually one of the things that I love the most as somebody, you know, as I mentioned at the start of the podcast. Right, like I was told, you know, I probably shouldn't go to college. And I had some challenges. And I think what's so wonderful about this technology and what we're seeing with it, is that it's really leveling the playing field.

00;11;14;27 - 00;11;44;17
KATHLEEN
So people who may not have been traditionally, quote unquote, smart or intelligent are having those opportunities. And people who have been maybe ostracized, as in terms of leaders in the tech space like this is an opportunity for all of us because we're kind of starting on a flat playing field to really kind of step up. So whether that's women and girls of color or different orientations like that is what I get really excited about is because you don't have to code to be successful in this space, right?

00;11;44;19 - 00;12;10;29
KATHLEEN
Sam Altman has a wonderful tweet, where he talks about, like, the most prevalent and prominent coding language today is actually English, right? You you know, that is where it lies. And we have, you know, everyone has access to that. And so it is exciting to see where I think, you know, oftentimes our world has been stifled a little bit from an innovation standpoint because people didn't know how to code or they didn't know how to bring an idea to life.

00;12;11;02 - 00;12;34;20
KATHLEEN
And so it's going to be really exciting to see how people who have different perspectives and insights and are now able to kind of take that and really be an innovator in the space, kind of come to life and solve some maybe world life challenges. which will be fine. 

RYAN
If you had to summarize this, what's the difference between AI and that machine learning and or automation.

00;12;34;20 - 00;12;57;27
RYAN
Like what's the difference between AI and automation? 

KATHLEEN
Yeah. So what I’ll say is this: so when we think about automation, it's like following a very distinct rulebook and really strict kind of rules and parameters and formulas that it should follow. So like when I gave my example about how I created that ETL process to go from here to here, it was more automation.

00;12;57;27 - 00;13;15;22
RYAN
What is ETL. Sorry, Kathleen, what is ETL stand for? 

KATHLEEN
It's like edit transform and I don’t know what the L stands for now off the top of my head. But it's basically kind of when you think about it, when you take a database and you want to transform it so that it ends up being consistent. So that is, it's a process that it goes to.

00;13;15;22 - 00;13;38;06
KATHLEEN
You do like clean data into a usable format or into the end format that you want from a standardization standpoint. 

KATHLEEN
So and those have very strict rules. So they would have things like if you have the word street, based on Google's algorithm for schema and what they have for rankings for SEO, we want it to be St period, not street spelled out.

00;13;38;08 - 00;14;01;27
KATHLEEN
And so we would write that and maybe we would have it for street, road, drive. But if we didn't specify Avenue should be AVE dot, it wouldn't have caught that and it would have left Avenue. And that's what automation is, is it follows these very strict rules and guidelines. And it requires somebody like a coder or myself going back in and saying, oh wait, no, no, no.

00;14;02;00 - 00;14;20;25
KATHLEEN
If you see the word avenue, that then needs to become a period. But when we work with AI, you can say, hey, we want to abbreviate any type of word that deal with street, drive, to do things of that nature. And if avenue would have come up, it would have realized that and would have done that on its own.

00;14;20;25 - 00;14;40;03
KATHLEEN
So it's this idea when people ask, what is AI? AI is making machines smart.  And that's what the difference between automation and AI is, is it's able to say like, okay, you didn't give me this exact formula and there's a use, you know, an input that is slightly different or varied. But I can utilize critical thinking, which is kind of scary to say.

00;14;40;08 - 00;14;57;17
KATHLEEN
But, critical thinking in a way to resolve it. and so I think that's really how you differentiate the two. I often also use this analogy of, you know, going on a road trip with your kids, and yeah, you can use a GPS that will tell you, you know, how to get there. But if you're like, okay, I need to go.

00;14;57;17 - 00;15;14;16
KATHLEEN
My kid’s shouting in the back row, “I need to go potty, mommy, I need to go potty”,you know, one thing, as a parent, like, you know, you asked the GPS and it might give you a couple locations, you know, ten miles up the road and it'll give you, you know, most of your locations, like a mile behind you.

00;15;14;16 - 00;15;35;09
KATHLEEN
And as a parent, you know, like we do not turn this car around by any means. Whereas AI will say, okay, yeah, you're here. We know you're going here. We're going to actually take you. Instead of waiting ten miles, get to the next gas station that's just off the freeway. We're going to take you two miles and exit, and you're going to go a mile and a half further off of the freeway to get gas.

00;15;35;11 - 00;15;54;29
KATHLEEN
And so I think that's really this idea of it really becoming almost a copilot. And I know that word is used in terms of a lot of the packaging of the software, but it really is more of a copilot being an assistant to you, not just taking your instructions and being a doer, if you will. 

RYAN
So when you, you use the word assistant.

00;15;55;02 - 00;16;17;03
RYAN
So would you say that a lot of people already are using AI like you mentioned, maps like GPS and then assistants like is Siri and Alexa, do those also do those count as AI?

KATHLEEN
Not quite yet, you know. And you can tell right? Like how many times have you tried to ask Siri or Alexa or something, and they completely botch it?

00;16;17;05 - 00;16;45;19
KATHLEEN
I have a five year old who is like the biggest space nerd ever, and he'll be like, “mommy, how do astronauts go potty in the space station?” And like, and I ask Alexa and they're like, please visit the Johnson Space City website, dictates that the astronauts do use the restroom on the International Space Station. And you're like, that's not the question.

00;16;45;21 - 00;17;04;19
KATHLEEN
Or they'll send you a stupid link and you're like, I am cooking dinner and doing eight other things, like, I don't have time to like, I want you to tell me what it is. I don't want you to send me a link to my phone that I have to then read again, right? Yeah. And so I do think those, those two things are going to get disrupted very quickly.

00;17;04;21 - 00;17;29;04
KATHLEEN
Obviously Open AI has their whisper technology, which is incredible. And Inflection Pi, had really, really good promise. Now they've, they're in in talks, you know, Apple is in talks with Google and Gemini to voice to power their, iPhones moving forward. Because they, you know, Gemini has had struggled a little bit from the standpoint of getting distribution.

00;17;29;08 - 00;17;48;00
KATHLEEN
Yeah. But, you know, I actually really think Apple should be working with Inflection Pi because their technology is so great. 

JEN
Isn't it shocking, though, how flatfooted Google and Apple came out in this? They should have been ahead of it just based on the voice technology. But you're right. I mean series a piece of crap most of the time.

00;17;48;00 - 00;18;07;26
JEN
Like she and I are in a fight. Actually, I changed my my, Siri voice to a man and it got worse. So I'm just saying. But it's it's the perfect application for it, but it doesn't work. 

KATHLEEN
Yeah. Well, and that's I mean, I actually think when we think about kind of where we're going to be in the AI space as we move forward is right.

00;18;07;26 - 00;18;26;07
KATHLEEN
Like your, AirPods and voice is going to be where we're going, right? Where you're just you're not actually having to take the time to type stuff in, and you're just like, “hey, I'm, I'm showing signs of X, Y, and Z symptoms. What are some of the conditions that I could be exposed or that this could be?” and they would tell me in my ears what's going on.

00;18;26;07 - 00;18;50;27
KATHLEEN
And and I think these really when you think about us as humans, typing is not a natural. Like that's not natural for us. Right. So speech and dialog is much better. And so I think you'll start to see that come to fruition very quickly over the next couple of years. I don't know when I say next couple years, I think like six, two, eight, 6 to 6 months to a year.

00;18;51;00 - 00;19;09;22
RYAN
Yeah, it'll happen quick. I truly think that all of these devices, these separate devices, are going to be swallowed up by an Apple Watch. It's going to be your phone that you already have. It's going to be your, like you said, your headphones. It's incorporating all these into existing hardware that we already have versus making people buy new.

00;19;09;24 - 00;19;37;19
JEN
And Apple is very good at making things look sexy. Right? Like nobody wants to wear something that makes them look unattractive. The AirPods were a social status before they became such a regular part of our lives. The Apple Watch is, you know, it's it's pretty elegant looking. That's where I think Apple has the opportunity to really get into this and incorporate it into the things that they're good at, which is product design.

00;19;37;22 - 00;19;52;29
KATHLEEN
Exactly. And I think that's what we're starting to see. Yes, Apple is coming out and they're talking that they're working on their own foundational models. But I think they've also done a great job saying we're going to be partnering with the best in the, you know, what are we trying to figure out how we can partner the leaders in this space and focus on what we do best.

00;19;53;01 - 00;20;26;23
KATHLEEN
And I really commend, you know, I will say like and it changes, you know, I feel like, you know, right now, my favorite model I use on a regular basis, on an everyday basis is Opus, which is Claude Anthropic’s latest three version at the highest level. And that one's my favorite right now. But I know it's been over a year since they've made any major releases on four, so I know it's like one of those funny things where I feel like I'm cheating on ChatGPT a little bit because of Opus, but I know they're gonna win me back as soon as they release the next iteration.

00;20;26;25 - 00;20;54;21
KATHLEEN
But, you know, they had put a lot of time and effort into robotics at OpenAI and, they kind of struggled with it. And I think they're starting to realize, you know, we're really good at software, like, and they've kind of made an announcement that they're partnering with Figure, which is a robotics company that's like the leader in the space, and they're going to be utilizing they're going to be using, OpenAI software and their models and their AI models into these robots.

00;20;54;21 - 00;21;25;11
KATHLEEN
And there's some demos out there that are truly incredible. And it's actually really interesting. It's a very simple -  it's a gentleman talking to a robot, and he has a few, items in front of him, and the man asks, that makes a comment, you know, “I'm hungry. can I have something to eat?” And the robot is able to then look down, looks at the table, picks up an apple and hands it to him, and then the the gentleman, the real human says, “you know, can you help me understand why you did that?”

00;21;25;13 - 00;21;49;18
KATHLEEN
And he said, “well, I looked at all the items on my on the table. And I realized that there's only one edible item that's on the table, which is an apple. So I provided you the apple.” And what's so interesting about this is it is embedding, right. It's embedding that meaning behind I'm hungry. I would like to eat something and using that to make and discern and critically think about what's going on in front of it.

00;21;49;18 - 00;22;14;27
KATHLEEN
But then also in this demo video, you can see we talk about a lot about this idea of moving towards more of a multimodal interface. And this is a prime example, like the robot is listening and viewing what's around him. So it's utilizing kind of both like both kind of more language models as well as computer vision to achieve the outcome.

00;22;14;27 - 00;22;33;01
KATHLEEN
And so it's a very simple use case that when you really understand what's happening behind the scenes, you're like, this is going to be amazing. I tell people all the time and like, this is our generation's moon landing. And like, we're so awesome and lucky to be like, here to see it happen. 

RYAN
Yeah. And like you said, it seems so simple.


00;22;33;01 - 00;22;48;29
RYAN
Like when you're talking to someone, it's like, yeah, I could choose an apple too, but this is a machine. This is a computer that is choosing an apple. And there's a whole lot of there's a whole lot more that goes into it than just being like, oh yeah, it's obviously food. Choose the food. It's like, no. Yeah, a whole lot more.

00;22;49;02 - 00;23;12;28
RYAN
We talked a little bit about AI in kind of general, but as companies are looking to incorporate AI into their business model, what things should they be thinking about before they jump in? 

KATHLEEN
I think the first thing they need to realize and acknowledge is they need to stop with the delusional thinking and keeping their head in the sand that no one in their team is using AI, because it is happening.

00;23;12;28 - 00;23;31;17
KATHLEEN
And so, you know, I think when we talk through that, one of the things I always tell when I do a lot of implementations are working with companies is you need to think about, okay, first, do a survey, understand what your what your audience and your employees know about the technology, what tools they're using, how they're using it.

00;23;31;20 - 00;23;59;27
KATHLEEN
But then I think you really need to focus on not necessarily what I'll call initial guidelines. I really think it's important to focus on how your organization wants to approach it from a philosophical perspective because, right, like, I feel like if I miss a Monday and like, you know, two Mondays ago would be a prime example. I mean, Databricks, you know, released their Open model and you're like, oh crap, I missed Monday and now I feel like a year behind in terms of what's happening in the space.

00;23;59;27 - 00;24;35;00
KATHLEEN
And so you can't - it's hard to create guidelines and policies for something that's changing so quickly. So when you develop those types of documentation, focus on what is the philosophical foundation of that and how you want that to be used. And I think the core one that I'm a big proponent of is transparency. Transparency between your employees and their managers on when and how they're using AI technology and what tools they're using. Transparency from leadership down to the employees in terms of what their vision is for how I can be used.

00;24;35;02 - 00;24;58;24
KATHLEEN
I know there's a lot of talk around, this idea that I can help replace all these humans and create all these efficiencies and and yes, it can. I'm not going to sit here and say it can't, but I think there's such an opportunity that comes to the world and in life in terms of those companies who see this strictly as an efficiency play and cut a bunch of their staff.

00;24;58;26 - 00;25;28;28
KATHLEEN
And when I say a bunch, I mean like large numbers of their staff, I know there's going to be some of that. They're losing out on the opportunity of what we can do as it pertains as humans to leverage those tools, and what would be better invested to figure out how do we make our organization leverage AI in a way that is human centered and really helps humans achieve what they're best at, and their knowledge base of an industry or or experiences working in those industries and use the AI to help them solve those.

00;25;29;01 - 00;25;47;23
KATHLEEN
Then seeing as a straight, you know, kind of efficiency play across the board. So those are the two main things I would focus on there. And then I would always say, you write your policy and guidelines, keep it high level and then do an appendix with every tool. And I loved in college that it was a really big trend. We had those books.

00;25;47;23 - 00;26;12;19
KATHLEEN
I was like, eat this, not that. And so make your appendix and say, here's the AI tool. This is how you can use it, and then use it for this. And you have a column with all the ways that they can use it for and then not this and you and then you're constantly updating just the appendix or, you know, of that or the exhibit of that document and not what is the foundational belief as an organization.

00;26;12;19 - 00;26;38;06
KATHLEEN
Because oftentimes I feel like, too, right, this is changing so fast, it could be really easy to get distracted by the shiny objects and not think back. Okay. As an organization where did we want to stand on this and how did we want to integrate it in our company? And how do we want to handle bias that's innately built into how these models have been trained, because of the the, you know, the Internet of Things and the data that was collected has bias in it to begin with.

00;26;38;08 - 00;26;59;12
RYAN
Yeah. When you said the policy is really about beliefs that stood out just because you can't say, oh, we're going to use ChatGPT for X, Y, and Z because you even said as different models are published, one becomes better and one becomes worse, essentially. So if you don't have every single option, then yes, you will just use ChatGPT for everything.

00;26;59;12 - 00;27;16;24
KATHLEEN
But you know, like that, you just I feel like you're constantly like in this you in a way, you almost feel like you're like the hot chick on campus and everyone wants a piece of you, and you get to date around on all these different tools, which is a lot of fun. 

RYAN
Jen, I think Kathleen just told me that I was the hot chick on campus.

00;27;16;26 - 00;27;41;08
RYAN
Way to go! Oh my God. I got to work with him for the rest of the afternoon. 

RYAN
I'm going to be strutting around. Would you compare? Is it fair to compare? I, I think that I was a big boom, probably. And I mean, still very popular, obviously, but it was almost like monoculture. Everyone was talking about it, and there was a lot of discussion around how AI is going to be replacing everyone's jobs, and there was a lot of fear around it.

00;27;41;10 - 00;28;04;29
RYAN
Would it be the same as like when personal computers became a thing, when email became a thing, like all of those people were like, this is going to completely disrupt our industry. And I think it probably changed some things, but it didn't necessarily upend anything. And completely eliminate people. Right? 

KATHLEEN
Yeah. I mean, I will say this and I know, I know, I just called you the hot chick on campus, and now I feel like I'm about to bring you down.

00;28;05;01 - 00;28;29;15
RYAN
No! 

JEN
You know, that needs to happen. 

KATHLEEN
But I do think, like, you know, I'll be honest. Right? Like, yes, this is very much so. And I would say it's very similar to kind of the invention of the internet. Right. Like it is completely changed, right? Like most transactions happen online. Right. This huge transformation of how we work, is going to occur.

00;28;29;15 - 00;28;57;25
KATHLEEN
I and I think it'll be very similar to that or more of like kind of the huge adoption of like the iPhone. Right. And in that huge boom. But the rate of the, adoption and the innovation is so much faster. So while we've had over our past where you want to think about it, compared even back to the industrial revolution, we typically have a generation where you can kind of have this change occur and allow for adoption and adaptation.

00;28;57;27 - 00;29;17;20
KATHLEEN
This is happening at a much faster rate, which is slightly terrifying. And, and I'll be honest, like, I do have some concerns about, you know, what that looks like. And, and I think if anyone tells you that they know, like, yes, I think, you know, we're seeing it happen already. I mean, I don't know if you know about Klarna, which is obviously a fintech company.

00;29;17;20 - 00;29;43;18
KATHLEEN
I mean, they laid off 700 of their customer service support tasks. like customer support agents, because they developed an AI assistant that not only was able to handle the requests of their customers, they actually were able to do it faster and had the same quality metrics as the real humans and had, you know, less about 20% less, like reengagement of the same issue.

00;29;43;18 - 00;30;07;14
KATHLEEN
And so, you know, I think that there is that that is going to happen. and I don't know where it's all going to land, but I do think that there are going to be new jobs that are created. And maybe, you know, I have this maybe it’s too altruistic and maybe I blame it on the like, early retirement phase of day, day 15.

00;30;07;14 - 00;30;27;03
KATHLEEN
But maybe this is an opportunity as a, as a society for us to solve some really big challenges, that we've been facing that we haven't been able to really compute, just because of the vast amount of data that we have to analyze. And maybe this is an opportunity for us to really create that work life balance.

00;30;27;03 - 00;30;53;15
KATHLEEN
Like, is it a bad thing if we don't all have to work the grind forever? I mean, I know there's I have some, you know, I feel that way. And, you know, maybe that does mean that there's some type of other financial support that needs to come in play, obviously, as a result of that. but I also worry on the other side of that, I'm not only from the financial side, but just what does that mean for so many people who find such purpose and meaning from their career?

00;30;53;17 - 00;31;28;20
KATHLEEN
And I think, you know, I studied, I mean, I studied and worked overseas in, in Spain, and I would ask people all the time like, “oh, why did you choose to become a teacher?” And they're like, “because it pays the bills”, or things of that nature. And you're like, oh, okay. Like it's not as inspiring as I thought it would be, but they will tell you, like, I, I work to live, I don't live to work like and you know, so I think like as a society I actually I'm a little bit worried about you know, the United States because we are we've had this huge culture, this like grind, the hustle culture, and

00;31;28;20 - 00;32;00;07
KATHLEEN
finding meaning and value of people's worth and their jobs and their careers. And, you know, I hope that maybe AI can actually help alleviate some of that and help course correct from that perspective. Yeah, I think we've got some time. If you're leaning in, if you're leaning into it and you're are wanting to learn and and be a leader and bring that your knowledge or your industry sector, your niche to the application of AI, you will be successful and you will have a job.

00;32;00;09 - 00;32;23;28
KATHLEEN
Yeah. for now. Yeah. But, you know, I think those people who are a little bit more resistant are going to struggle. 

RYAN
Yeah. When we were talking about an AI policy or talking about introducing it at your company, we talked about the the human centered piece of it and how your organization views AI. Talked a little bit about adding in the appendix at the end.

00;32;23;28 - 00;32;52;02
RYAN
That's where you have like your do's and don'ts. But when it comes to the rest of the policy, like who goes into creating that, how do you determine even what's in what's in that policy? 

KATHLEEN
Yeah, I think a few things like, you know, it's interesting. Right now I'm seeing a lot of CTOs own the AI space, or I was just talking to a big company just earlier today, and they were like, oh, yeah, our legal department is owning our AI implementation

00;32;52;02 - 00;33;18;26
KATHLEEN
And I it's like, well, it sounds like, I'm sure it's very progressive based on that. But I think, you know, when you think about who should be on the governance council and putting some of these documents together, I truly believe it to be, you know, somebody with a very technical data security background, somebody with like a legal or compliance mindset, an executive of the company who understands the business and what they want to achieve in it, and somebody who has a passion for AI.

00;33;18;28 - 00;33;41;28
KATHLEEN
And then it's almost like, you know, a cage match and you fight it out and figuring out where you guys can land in the middle and all be comfortable with and constantly revisit it. And, but no, I mean, I think it's something I've always said this and I believe this when I, you know, as a saying and a culture of Decode was: talent produces something good, but collaboration produces something that is truly great.

00;33;42;01 - 00;34;03;25
KATHLEEN
And I think that's so true, even in this instance. And yes, while like in my past career, my CTO and I sometimes would bicker like an old married couple about, you know, him being too cautious about something and me being too like shooting from the hip. And but you know, the truth is, the right answer was really probably between the two of us.

00;34;03;25 - 00;34;26;10
KATHLEEN
And so I think those are the individuals that should be leading those conversations and helping you implement it. and I really do think, you know, and maybe it's just because I have a like I was a digital agency and I'm a digital nerd at heart, too. Is, you know, they understand complex challenges. They've had to deal with legal and data privacy issues in the past.

00;34;26;12 - 00;34;56;11
KATHLEEN
They've built entire strategies around Vine or Thread and then only to have it destroyed within like weeks. And so I do think those individuals who've been focusing on digital transformation for the last decade or so are really ones who, if they want it, could really be a huge asset for any organization that's going through this. 

JEN
Kathleen, how long do you think, before the government starts to regulate this?

00;34;56;11 - 00;35;21;25
JEN
Because they've taken sort of a like we're going to look the other way approach with a lot of technology around social media, which I think is is coming to a pretty big friction point. They have looked the other way on data privacy until very recently, where they're starting to finally make some noise. I mean, that looking the other way is what allows for innovation, but at the same time, it's what allows for abuse.

00;35;21;25 - 00;35;44;22
JEN
How far do you think AI goes before the US government says, that's too far? 

KATHLEEN
Yeah. I mean, I'll say this and I actually got this question in class the other day when we were going over AI ethics and legal, and I was like, you know, I do hope, you know, and I do think, you know, the way that we control this is through regulation.

00;35;44;25 - 00;36;05;03
KATHLEEN
and they're like, you're telling us that you think the US government and the individuals, based on the demographics of those individuals, are capable of understanding this technology and implementing it? And I said, well, just like I would love to be bikini beach ready by summer, some of these are just wishful hopings. But I think, you know, when it comes to that, you know, I think there's two things.

00;36;05;03 - 00;36;25;10
KATHLEEN
And I think what, you know, I don't have a lot of faith just because I don't think they truly understand it. And as much as I see a lot of people in the industry wanting to lean in and and help, put those regulations in place, there are some ulterior motives to that in terms of creating a moat for new up and comers to get into that space.

00;36;25;12 - 00;36;48;00
KATHLEEN
I do think, you know, my thought in terms of how we finally get true regulation in this that makes sense, is we have to make it economically incentivized both on the company and the purchasers of this technology to have those regulations and I think that is where it will really actually stick and work. And by that I mean, right.

00;36;48;00 - 00;37;12;03
KATHLEEN
And how do we do that is we have to have an educated and informed audience and community. And so like, I feel like when I teach at rice or when I talk at a lecture or I go to, like, I’m going to speak next week, it's about like continuing to spread the word and create awareness around AI and helping to ensure that more and more people have the language to talk about this.

00;37;12;05 - 00;37;32;15
KATHLEEN
Because if we can start demanding it from the AI companies and it then becomes an economic reason for them, like, I think that's when we will start to see change. And obviously with this being an election year, I'm like, for the love of God, everyone learn some things about - I understand that like deepfakes can happen and you will not know the difference.

00;37;32;17 - 00;37;52;02
KATHLEEN
And there are times where I'm reading staff and I'm seeing things I’m like “deepfake” and then I'm like, oh no, actually real. And even scarier. 

JEN
It's hard to separate the reality and the fiction right now. 

KATHLEEN
But yeah, and you think like, I think about my mother in law, who is a wonderful and kind woman, but like, she believes everything on her Facebook feed even five years ago.

00;37;52;02 - 00;38;14;12
KATHLEEN
And it wasn't even that good at faking it to be, like, legitimate. And now that it can look real, like, oh, crap. Like as a society, we just need to get way better about questioning things and not just reading what we want to read and assuming the truth. 

JEN
Well, we need better bullshit detectors at every level 

KATHLEEN 
Yes, we need better bullshit detectors.

00;38;14;12 - 00;38;35;09
KATHLEEN
Now, that would be an AI tool that would be fun to make. 

JEN
That's right. Let's make that next. 

RYAN
I think that some schools are, kind of adding a media literacy piece into a lot of their curriculum, even at a, like a middle school level, which is really important just because, like you said, we don't know. 

KATHLEEN
Well, even like, I don't know if if you remember this when the internet was invented.

00;38;35;11 - 00;38;56;02
JEN
Yes, I was there. 

KATHLEEN
Right? I see when I was there, and you do remember when you were taught and it was like, here's how to do a boolean search. And it was like, if it's a.org, you have this credibility. And if it's a dot this, you have this type of credibility. And we were taught like to like understand like how to utilize like search.

00;38;56;03 - 00;39;20;09
KATHLEEN
I mean search was new like Google didn't exist. And so, you know, like understanding that and when it did exist, like we were taught these different have parameters and all that stuff. And so I do think there's going to be a whole new round of literacy that's required. And it doesn't matter if you want to be a computer engineer, you know, computer engineering, tech founder or a doctor, you have to know and understand these things because it's going to infiltrate your lives.

00;39;20;11 - 00;39;46;09
RYAN
What are some ways that companies can figure out how to use AI at their company?

KATHLEEN
So I have a framework that I use, where I look at different pilot programs, but I often say, you know, it's okay to start small, right? You don't have to start with something that's like super advanced, like, and out of the box that figure out where the friction points in your in your business, whether it's internally or with your customers.

00;39;46;11 - 00;40;09;02
KATHLEEN
And, and start there and say, okay, here's this problem. Or maybe on the other side, here's this opportunity. And what is the challenge? I'm struggling with it from this standpoint now. And how can you use AI to like, if I could do this, even if you don't know, it can, write it down, think it through researching and sometimes I have like a list of things that I've been thinking about and doing.

00;40;09;05 - 00;40;25;24
KATHLEEN
And there were times where it was like this was impossible nine months ago or a year ago, but I go back to my list and like, oh, actually, it's now possible. And so I kind of move that forward. And, I had a student, they just finished their capstone projects. He actually was like, okay, I have a he was working with a local plumber.

00;40;25;24 - 00;40;41;02
KATHLEEN
Right. So nothing sexy, nothing exciting. And he said, you know, one of the things that we struggle with is lead capturing. And sometimes they would fill out the form and he would get calls all the time. And there are certain things that he did and other things that he didn't do in terms of services or plumbing.

00;40;41;04 - 00;41;04;00
KATHLEEN
And so this student created an AI, chatbot that just understood what services they offered. Some basic like plumbing and sewer, knowledge, if you will. I'm not very well versed in either of those, and then connected it so that his customers, when they went to their website, could ask a question and say, oh, it does sound like you have an issue with a plumber.

00;41;04;02 - 00;41;35;27
KATHLEEN
I do offer services for sewer work. Do you want to you want me to contact you or would you like to call me now? And he actually had the chat bot integrate through a Zapier connection to actually call the kind of owner of the business immediately, if they asked for that, or if they said no, have them call me at this time or that time, they would send a text message to the business owner that said, hey, here's the name of the person, 

00;41;35;27 - 00;42;05;07
KATHLEEN
here's their issue, here's the conversation we had. They want you to call them at this number at 4 p.m. on Friday. And he has seen a huge increase in his business, already in terms of leads. And this student built it in a weekend. 

RYAN
Oh, wow. 

KATHLEEN
And so I think, you know, when you think about, you know, what's possible, make sure on your list of different pilots think about, you know, what is the impact of change management, what is the cost, what are some of the risks and kind of what I like to categorize.

00;42;05;07 - 00;42;28;17
KATHLEEN
What type of pilot is this? Is this a quick win? Is this a sweet spot for us? Like something that is a little bit more challenging than a quick win, but like we could easily get this done? Or is this like a moonshot, right. And having a mix of those constantly going on, I think is really important because you don't want to be focusing on moonshots, that you're not able to make that incremental improvement and actually feel the success of AI.

00;42;28;18 - 00;42;47;20
KATHLEEN
Sometimes it can be really frustrating when you're like working on a moonshot and it's not working yet. And so I think keeping that mix alive is important. 

RYAN
Have some quick wins to keep you going. 

KATHLEEN
Yeah, exactly. 

RYAN
So it sounds like make a list of your the challenges that you have in your business. maybe some tasks that just seem repetitive or take too much time.

00;42;47;22 - 00;43;09;16
RYAN
And see how you can use that. 

KATHLEEN
And I always say, you know, it's this idea of you have thinking and thunking, right? So thinking is actual critical thinking about stuff and thinking is that mundane, stuff that you have to grind through, take that away from it and find the friction points. Like everybody has something in their day that they don't like doing.

00;43;09;19 - 00;43;27;25
KATHLEEN
And sometimes you start there, right? I can be really scary and intimidating, but if you take away the things they love, they'll hate it forever. If you take away the things they don't like and you show them how they can make their life better, where they can focus on what they do love. You'll get people buying into it left and right.

00;43;27;27 - 00;43;45;24
RYAN
Yeah. Don't. Yeah. Don't take away that piece of their job that they really do love because then -not not fun. Not going to be at their job very long then. But yes very good explanation of the thinking versus the thunking. Like don't thunk, use your brain to think. There's other things that can help you do the stuff you don't like.

00;43;45;27 - 00;44;12;21
RYAN
Would you say that, the AI boom is over? Or do you think we're still very much in it? 

KATHLEEN
I think we're in the infancy of it. I think we're at the very beginnings of it. I, I think in the next 18 months we're going to be like completely transformed by it. 

JEN
But, Kathleen, do you think that all of the AI advancements that are yet to come will be coming from, you know, maybe the big three?

00;44;12;21 - 00;44;37;23
JEN
Or do you think that we're still going to have some, you know, dark horses come out, into the field that we haven't thought of yet? Or do you think that now that, you know, Open AI, Anthropic, etc., they, they've kind of established themselves as the ones that are going to lead the charge. You think most of that's going to come out of them, or do you think we will still see some really interesting startups come out of this?

00;44;37;25 - 00;45;14;26
KATHLEEN
So I think you know, this. You know, I would like to say it's like a 70/ 30 rule, right? Is we're realizing that foundational models are really hard to build and they're very expensive. And it's gonna be very hard to compete with organizations like OpenAI and Anthropic and Gemini and Google to compete from a research and innovation standpoint. So I think, yes, I think 70% of the innovations in the movement that we're going to see is going to be coming from these big players in the space, but I think you're going to have about 30% of these dark horses that come out of nowhere, and some of those dark horses might not be building foundational models.

00;45;14;26 - 00;45;34;12
KATHLEEN
They just might have a unique way and a unique perspective of wrapping the foundational technology that they're leveraging from one of those big three, from that perspective. So I do think that, yes, we're going to see we're starting to see consolidation across the board. And so yeah, I don't think we're gonna see big, you know, changes from that perspective.

00;45;34;15 - 00;45;54;14
KATHLEEN
But I do think we're going to see some dark horses that are going to come through and do some really awesome work. 

RYAN
Yeah, it's kind of like, Facebook, like the big leading social network. I've, I've heard it on a few podcasts that Snapchat should really get a paycheck from Facebook because they come up with all of their ideas and then Facebook eventually steals them.

00;45;54;16 - 00;46;13;17
RYAN
So you need those places that are coming up with new ideas, new technologies. And, you know, maybe they do either get acquired by OpenAI or they get acquired by someone else or they might not get acquired. They just might become feature ized and they might steal their idea, but still very much need those free thinkers. 

KATHLEEN

Yeah. Or acqui-hires.

00;46;13;17 - 00;46;30;06
KATHLEEN
Right. You'll see that. I mean, that's kind of what happened at Inflection right? They didn’t buy them, but they took all the talent and said, wait now come over here. Yeah. We want you on our team. 

RYAN
Yeah. That was that was a rough one. I was nervous when all of a sudden there was stuff happening with Sam Altman and it's like, oh, he's going to get hired by Microsoft.

00;46;30;06 - 00;46;52;12
RYAN
And then they're going to hire a bunch of the OpenAI team. It's like, wait, nope. OpenAI calmed down a little bit. They calmed down and they brought everyone back. What would you say your personal favorite AI tool is if you had to choose just one right now? If you could only use one platform, one AI tool for the rest of your life, what would your one be?

00;46;52;14 - 00;47;15;20
KATHLEEN
I would say right now it's Opus just because of where my love affair with it right now, in terms of everyday use, the one that I have the most like I think is the sexiest and have like this deep passion for is actually, Hume. And it's utilizing - It's basically like an emotionally intelligent AI voice, which is really cool.

00;47;15;20 - 00;47;37;00
KATHLEEN
And the reason I love it is because I'm a linguist into at my heart is just amazing because they're utilizing different types of patterns and pauses and intuition to understand, the feelings of the communicator and analyze what the feelings are behind the speech, and then have the AI respond in a way that is reciprocal and matches or is appropriate for that type of sentiment.

00;47;37;02 - 00;48;02;04
KATHLEEN
And they're doing it from a very mathematical approach. And so I don't use it every day. But I like really think it's cool. 

RYAN
How would you describe what opus is? 

KATHLEEN
Oh man. So Opus is I think, you know, it's very similar in terms of where it can land in terms of ChatGPT for comparison, but I do think it does a better job of understanding your voice and, and your tone a little bit better and can sound more like you.

00;48;02;04 - 00;48;26;22
KATHLEEN
And I am right now liking the output of it just a slight bit more. I also love, as a linguist right there, naming conventions. Like naming conventions in AI are just atrocious overall, it like bothers me. And it's confusing. But their model for Claude 3, they're you know, they focus on they have sonnet or no haiku sonnet and opus.

00;48;26;27 - 00;48;43;23
KATHLEEN
So it kind of follows from like haiku is their more basic model. And sure, it's like more condensed. And so it's just beautiful in terms of how that all comes together to you as a linguist. I'm like, I love that you're naming your lines with like the use case and data size and all that, and it's easy to understand.

00;48;43;23 - 00;49;06;28
KATHLEEN
Then, you know, some of the other ones that have like Pro and 1.5 Pro and plus and you're like, choose which ones, which ones better?

JEN
How many pluses? Seven pluses. 

KATHLEEN
Yeah seven pluses. 

RYAN
Yeah. So you use opus kind of to create output. And you like it because it, it can better reflect your, your style of writing or do you use it for other things as well?

00;49;07;01 - 00;49;26;28
KATHLEEN
I use it for other things as well. And I also really like Julius I it's a, it actually utilizes Python, to. So there's, I think a little bit less from a hallucination standpoint. Er, and it's really good at financial and data visualization. So I think, you know, right now you're constantly like dating and dancing with a lot of folks.

00;49;26;28 - 00;49;52;12
KATHLEEN
But I really do like, those three. And listen, I still love ChatGPT for I like, I think it's, you know, and one of the best all around players, if you will. Right. Like it can do the code interpreter and data analysis. It can do the Dall-E three. So it has a little bit more functionality and a little bit more an all around function, if you will.

00;49;52;15 - 00;50;14;11
KATHLEEN
but you know, and like I said, they're going to really releasing their next iteration probably in the next I, I my bet is this summer. And then I'll be dumping Opus to go back to GPT. 

RYAN
You said hallucination. Can you explain what a hallucination is in the AI space? 

KATHLEEN
Yeah, it's just this, hallucinations when AI gets it wrong.

00;50;14;14 - 00;50;37;26
KATHLEEN
or kind of misrepresent something, and and it's not intentional. I think, you know, a lot of the AI is is in trying to please the user and, and satisfy the request. So if it doesn't have the answer, sometimes it will try to find an answer or come up with something that's not accurate. Most of the best models that we're seeing right now have about a 3% hallucination rate, which, when you think about it, is not that high.

00;50;37;29 - 00;50;57;09
KATHLEEN
And I think, you know, it's funny. You know, in my class we were talking about it. My dad sits in my class and he said, why do we have like, yes, why do we have such high expectations? You know, humans hallucinate probably more than 3%. And so why do we have why are we placing different expectations on AI than we are each other?

00;50;57;11 - 00;51;16;26
KATHLEEN
which I thought was kind of funny. And I'll say, like, I guess that's true too. 

RYAN
I love that you referred to or you mentioned your dad, because when I talk about AI and when I talk with my students, I call them Vernons because my dad, his name is Vernon. He will confidently state something. And I thought my dad was the smartest man in the whole entire world because he would confidently say anything.

00;51;16;26 - 00;51;33;19
RYAN
So when you said, your child was asking about how, astronauts go to the bathroom, my dad would just make something up and say it so confidently that I would just ask, well, yeah, of course. And yeah, AI just says things confidently. And you do believe them. So there is. Yeah. You do have to check things sometimes, right?

00;51;33;19 - 00;51;54;01
KATHLEEN
Yeah. that critical thinking that that bullshit detector always gotta be on now 

JEN
We need a custom Vernon GPT. It's just going to state things confidently, but be complete…It should be instructed to be wrong. Yes. 

RYAN
And it'll just follow things up with, whenever you ask a dumb question, it'll say, “you've got to be shitting me”.

00;51;54;04 - 00;52;32;10
KATHLEEN
Oh my God, that amazing Vernon ism. Yes, yes. 

RYAN
Not even from a business standpoint, but what is the, the AI tool that you would recommend everybody check out today? 

KATHLEEN
Perplexity. AI 

RYAN
Perplexity. Can you explain what that one is?

KATHLEEN
It's like Google on steroids and the best thing ever. You know, one of the things that it really does is it takes, you know, just like search, you can ask the question, it actually gives you a response like it would be talking to a friend, but then it actually will source where it's getting the answers from with like little ones and twos and dots and stuff.

00;52;32;13 - 00;52;56;12
KATHLEEN
So you can actually, look at it further and then it pre prompts like your next questions and answers those for you. It's free. It's an incredible tool. You will never use Google again. 

RYAN
Perplexity AI a free tool okay okay 

JEN
Kathleen this was amazing. Thank you so much. The work that you've done in this field is really it's really great and we really appreciate you being here with us today. Thanks so much for joining us. 

00;52;56;18 - 00;53;17;12
KATHLEEN
No thank you. And Ryan, it was a pleasure to nerd out with you guys. 

RYAN
Thank you. 

KATHLEEN
Now I now crown you AI evangelist and spread the word so that we can help with the bullshit detector that we all need to hone over the next several years. 

JEN
He's going to spend the whole weekend making a freaking crown now.

00;53;17;12 - 00;53;28;04
JEN
And it's going to say AI evangelist. He'll make you one, too. So, thanks so much. Take care.

00;53;28;06 - 00;53;52;11
JEN
I need a beer. After all that. That was an incredible amount of information. And, she is so smart. But it's just a lot to take in. 

RYAN
That's a lot to digest. Yeah. Cheers, Jen. 

JEN
Cheers. So for the Power of Three segment Ryan, what are our top three takeaways from what Kathleen talked about and some of the things that people can implement to reduce that friction?

00;53;52;14 - 00;54;11;03
RYAN
Yeah. She talked about a lot of great things, but if I've got to summarize it into three points, a little bit of theory and also a tool, the first one is going to be creating a company policy for how you and your employees use AI, how you want that to happen. And really, it's your philosophy behind AI.

00;54;11;03 - 00;54;31;10
RYAN
How do you want it to help your business? Establishing that first will be a really good step as you as you progress into your usage of AI 

JEN
And and to not be afraid of it and to allow your employees to use it. As a creative agency, we have to think a lot about how do you use it for maybe ideation, but you know, things you can't copyright that or things like that.

00;54;31;10 - 00;54;51;19
JEN
Like, that's not that's not a proper use of it, but it has a place in the company. 
 
RYAN
Right. The other thing, as you as you create your policy, adding an appendix of a do's and don'ts.

JEN
I thought that was brilliant. 

RYAN
Oh my gosh. Yes. Use this platform for this. Don't use this platform for this because it will help people save time.

00;54;51;19 - 00;55;11;11
RYAN
Maybe you've learned that this platform doesn't work for it or there's more hallucinations. All of that's very helpful. That saves everybody a lot of time. 

JEN
And your operations director will thank you for not having to rewrite the employee handbook every five minutes. It just references a link to the appendix, and you can update that every five seconds as the AI tools change.

00;55;11;13 - 00;55;32;01
RYAN
Yes, yes. And the last one. The last one in our power of three. Perplexity AI. This is something that Kathleen has talked a lot about over the months. I've started using it as well. It truly is Google with superpowers, and Google has made the world so much better. Perplexity has made it even more advanced and way more tools for it.

00;55;32;04 - 00;55;52;09
JEN
All right. As power three. 

RYAN
Okay, now it's time for our Creative Brief segment. We just spent a lot of time talking about AI today, and in line with that, we have a new commercial by Dove. This commercial is called The Code. I recommend everyone go to the show notes and watch this video. This one really stood out to me.

00;55;52;09 - 00;56;13;19
RYAN
And the commercial starts with kind of an ethereal, dreamy quality of sound, so it makes you stop and pay attention. It uses the song Pure Imagination from Willy Wonka and really sets the tone for the ad. According to the ad, it starts with some text that says by 2025, 90% of ads will be generated by AI.

00;56;13;22 - 00;56;39;07
RYAN
And then it goes to a young woman who is searching or prompting images in a image generator tool like Midjourney. Saying “imagine a gorgeous woman” or “imagine perfect skin”. And in in a lot of, AI tools, you start with “imagine”. So you're telling Midjourney to imagine something and then it comes back with it. So then when they type it in, it's showing what the AI has been trained on.

00;56;39;07 - 00;57;00;27
RYAN
It's showing graphics of, a blonde, white, tall, very thin woman. It's pulling in pictures of people that’s skin clearly is photoshopped, all of that. So it's, it's pulling in what it's been trained on 

JEN 
…and scantily clad, if I may ad, scantily clad. 

RYAN
Yes, of course, in a bikini on the beach. That's what a gorgeous woman is according to according to AI. 

00;57;00;27 - 00;57;27;29
RYAN
Then they add in the words “according to real beauty ad or Dove's real beauty ad”. So “imagine a gorgeous woman - according to Dove's Real Beauty ad”, that's the name of the campaign that they've been running for 20 years. So it has images, videos of the Dove Real Beauty campaign. And the images are still beautiful women, but it's real beauty.

00;57;28;06 - 00;58;01;16

RYAN
It's what people actually look like. 

JEN
It's not just tall, thin, blond women. It's inclusive. The first image generated ad is a woman in a wheelchair, for that matter, not just a tall, skinny woman on a beach. 

RYAN
And AI works by being trained on all the content that is out there. So, if AI is trained on a bunch of content that says a beautiful woman is a, tall, thin, blond woman with absolutely no texture on her face, that's what AI think's a beautiful or gorgeous woman is.

00;58;01;16 - 00;58;24;29
RYAN
And that's what it will say when somebody asks what a gorgeous woman is. But at the end of the commercial it says, what kind of beauty do we want AI to learn? Because that's truly what we're doing. We're teaching AI with every social media post, with all of these, commercials, ads, magazine clippings. We're training AI. So really, what do we want AI to come away with?

00;58;25;06 - 00;58;53;04
RYAN
And I think that this commercial really, does a great job in saying, hey, let's not distort. Let's not distort things. 

JEN
It's interesting that if it's true that when they add in that qualifier of, “according to the dove Real Beauty ads,” that it gives them something very different, much more realistic, way more inclusive, then that shows that they've trained the world in some ways.

00;58;53;06 - 00;59;14;09
JEN
And I just like, that's just amazing. I gotta take my hat off to to Dove on this. It’s really good work. 

RYAN
Right. And I think that Dove has become one of the, the pillars in the advertising world. You know, it's they create amazing work, but it's that work that really hits at the human emotion because they point out something that you inherently see is wrong with the world.

00;59;14;11 - 00;59;34;23
RYAN
It's like, oh yeah, we we aren't doing right by ourselves and by our children, by really putting this type of skin, this type of look on a pedestal. So it, it kind of shines a light back at yourself or makes you look in a mirror and be like, is this how we should be acting? 

JEN
I think this is important work and I'm sure it sells soap.

00;59;34;25 - 00;59;57;00
JEN
I'm sure it does. But I think it also contributes to the betterment of society in general. So that's I'm gonna I'm going to give a two thumbs up to the creatives at Dove…Cause they're listening to this podcast. 

RYAN
Oh yeah. Number one subscribers. They love our podcast. But the commercial or the or the campaign ends with a pledge almost.

00;59;57;02 - 01;00;15;25
RYAN
and it says Dove will never use AI to create or distort a woman's image. I am going to consider that a pledge from them. I hope that they do take that to heart. They've talked about photoshopping and how they would never do that in their campaigns to distort a human's image. And, I hope that they follow the same with AI.

01;00;15;26 - 01;00;33;06
JEN
I kind of forgot about that, how they they made that pledge. So, it is interesting, you see a lot of ads with these very, very fake looking images and not all just people. I mean, I saw one on a landscaping company's website and I'm like, that's so clearly fake. I think that's what people are looking for when they are looking for their garden.

01;00;33;06 - 01;00;54;28
JEN
But, it's kind of everywhere and it'll be really interesting to see how, obviously fake imagery is accepted or not accepted by, by consumers. So that's our that's our creative brief for the day. 

RYAN
Yeah, definitely check it out in the, in the show notes. We'll put some other videos within this real beauty campaign that they've done.

01;00;55;04 - 01;01;19;10
RYAN
Like we said, it's been happening for 20 years and they've got some truly amazing work. 

JEN
Get your tissues. 

JEN 
That's a wrap on our third episode. special thanks to Kathleen Pearly for joining us. That was so fun. She's so smart. 

RYAN
Yeah. Kathleen. Truly amazing. A wonderful conversation. She blew me away. But my favorite part, she called me the hot girl on campus and crowned me an I evangelist.

01;01;19;10 - 01;01;34;26
RYAN
I am going to go into my evening on cloud nine. That was amazing.

JEN
That is the number one violation of rules at de Novo is to not compliment Ryan, because now we have to live with it. 

RYAN
Yeah, my head's too big to get out of this podcast studio. I'm stuck here.

01;01;34;28 - 01;01;57;09
JEN
Thanks for listening. We hope you enjoyed this episode of Think Fresh. 

RYAN
And remember, the conversation does not have to end here. If you liked what you heard today, be sure to follow us on Facebook, LinkedIn, or Instagram. Review our show on wherever you listen to your podcast now, or share all your marketing trials and triumphs by shooting us an email at info at info@thinkdenovo.com with the subject line “Dear de Novo” so we don't miss it.

01;01;57;12 - 01;02;20;18
JEN
And while you wait eagerly for our next episode, you can get your fix by checking out our blog, Fresh Thinking at Blog.thinkdenovo.com. Stay tuned for more engaging conversations, laughs, and of course, marketing brilliance and me making fun of Ryan in the next episodes to come. 

RYAN
Here's to fresh thinking, sparking creativity, and never being boring. Bye Friends!

01;02;20;20 - 01;02;23;00
JEN
Are we swearing on this or no?