REPLAY Oh the Things We Can Do With Generative AI

Doug Downs (00:02):

On my 12th birthday, my parents got it right. Man, did they get it right. They bought me my first guitar. I was in awe and I was hooked right away. I will never forget the first notes I ever played, and soon I was playing patterns and then playing chords. Eventually arpeggios, which are just individual notes of a chord, really easy. A couple years into playing, there was a song by the Beatles that I really liked called Birthday. I was playing it like this, getting all the notes right, but a friend of mine taught me how to slide on the guitar, changing birthday to sound like this. And then I learned hammer ons and pull offs. A hammer on is when you go from a lower note to a higher note without plucking the string a second time, and a pull off is the opposite. Going from one note to a lower note without plucking the string. These are important techniques to help you play more intermediate level songs like Crystal Gayle's Don't Make My Brown Eyes Blue. And with more and more practice and even more subtle techniques, I eventually became what I am today a pretty Intermediate level player still

(02:18):

Always knowing there are even more things you could do with this wonderful instrument to sound even better. Today on Stories and Strategies, this new instrument we have with Chat G P T and all generative AI tools, what can we do with them to climb that hill?

(03:05):

My name is Doug Downs. Okay, music. Off the top, we had House of the Rising Sun, a traditional melody, traditional lyrics made famous by the Animals. Birthday, the Beatles written by Paul McCartney. Don't it make my Brown Eyes, blue Crystal Gayle making it famous, but Richard Leigh was the composer on that song. And Solsbury Hill, the classic from Peter Gabriel. My guest this week is Brad Williamson joining today from right here in Calgary, Canada's Rocky Mountains. Hi Brad.

Brad Williamson (03:31):

Hi Doug. Thanks for having me on.

Doug Downs (03:33):

And Spring is here in Calgary Western Canada style. We've had some warm days. We've had some cold days. It's, it's sort of rainy right now. We might still get some snow You. It's a grab bag, right?

Brad Williamson (03:44):

It's all over the place, but that's Calgary for ya. Just that's what we live in. Just happy to see it warming up a little bit.

Doug Downs (03:50):

It's the price we pay for that view, which is awesome. Brad, you're the founder and lead strategist for Catch Digital, building marketing systems that are focused on return on investment. Your approach is systemic. It's to break down the client's customer acquisition journey, the sales process and re-engagement strategies to develop a scalable model that allows the business to focus on its operations. The work you do requires a mix of marketing strategy, user experience, designer, of course, UX as the lingo goes, user interface design, paid media content and automation and artificial intelligence, AI. Okay, first, in this episode, we might use the terms chat, G P T and generative AI kind of interchangeably here. Chat, G P T is kind of the Kleenex of the tissue world. It's one of the tools available in this wave of generative AI, but it's only one tool of thousands. It's the Coca-Cola of all those right now, but it's, it's not the only tool, it's just the really, really cool one that everyone's talking about. Brad, this interview process began with one of your team members, Stephanie, challenging me to accept that Chat G P T could write my blog posts for me and that they would actually be okay. And man, I was resistant. Stephanie was right though I gave it a try. It saves me a ton of time. Obviously, I can still tweak the writing. And the end goal, which she helped me realize really for my blog posts is search engine optimization. But that's the tip of the iceberg for what we can do with generative AI.

Brad Williamson (05:33):

It certainly is. So really at the end of the day, AI is still in its infancy, and we're seeing it transform industries, transform marketing from content production to data analysis and everything in between. So how do we keep that consistency for one given brand or one specific company? And that's a process that at Catch, we've kind of built something that we call mega prompting around. And I believe that this is something that people are going to use in the ingrained right into their brand guidelines, and it's going to span across corporate communications to their marketing messaging and everything in between. Now, this is speaking specifically about Chat G P T. And so if you're a business owner listening to this right now, you probably want to think about this as if you're building a brief, telling a content writer what you would talk about with your company, how would you break that down?

(06:35):

So that process, mega prompting is something that we do at Catch Digital, and basically for every single client, we'll have a large Google doc that'll talk about the tone. It'll summarize the company, the unique selling proposition, the audience, and it ends up being about two pages long that you copy and paste into every single time that you're writing a prompt. I feel that that consistency and depending on the level of people who, depending on the level that people have in creating prompts, if you have that central place to be able to get that key information out about your business, you're just going to have that consistency and you're going to be able to achieve what you want out of the AI that much faster.

Doug Downs (07:22):

It's going to sound more like it's coming from my company, it'll have my 14 herbs and spices within the writing and the flavour of it.

Brad Williamson (07:28):

And this isn't reinventing the wheel. This is something that marketing companies and market research firms look at in terms of looking at the audiences and the tone and the voice, and basically building that persona for a business. You just want to ingrain that persona into your prompts so that you're able to produce that content, whether it be blogs, Google ad copy, Facebook ad copy scripts for TikTok. This could even come into podcast itself. So the usefulness of that base is quite vast,

Doug Downs (08:06):

And the real power of generative AI you were telling me is the ability for it to personalize content. So here's an example. This is a pitch that I got yesterday from a PR handler who's pitching their executive to be a guest on this podcast. They said, here, I'll read it here. Hey, I love what you're doing with Stories and Strategies. I'd listened to the episode about sales funnels and found it really engaging and relevant content. I think blah, blah, blah my executive would make a really great guest on your podcast. So I would give this a three out of five. They knew the name of the podcast, they did dig up the title of an episode, but three out of five, because they went no further, they didn't listen to the episode. I can tell that they didn't listen to the episode. How might that, just using generative AI going forward?

Brad Williamson (08:58):

Well, you can stack a couple different tools together, and you could actually build an entire workflow around this if you wanted to. There are tools out there where you could scrape an episode or take a transcription, and it's really common today, especially in podcasts or video, for people to transcribe those and put them up as blog posts. So you could actually scrape that information, pop that into Chat G P T, say, write me a request to join X title of podcast, and then they would end up getting an output that would be so specifically personalized, but you can take this a layer deeper where you could actually implement tools like Zapier to tell a web scraping tool to scrape the blog and then find the email and then take that information, creating the prompt of that request, and then automatically send it at scale to hundreds to thousands of different podcasts.

Doug Downs (10:00):

Wow. Okay. Yeah, and we can use generative AI to actually pre-screen. You were telling me who gets through to my inbox, who I actually pay attention to that one I'm really interested in how do I do that?

Brad Williamson (10:17):

Okay, so this is two-sided, right? I mean, they might be reaching out to you using AI, and you could use AI in that same way. So I just talked about Zapier, and that's a very important tool in this workflow. Basically, if you use Gmail or Microsoft 365, if you receive an email and say that you had it go to a specific inbox like inquiries at stories and strategies, something like that, if it came to that specific inbox, then you could get Chat G P T to rank the ones that would be most relevant to your show and have the best pitches based on a specific set of criteria that you would find to be a really good pitch. Now, Doug, what would a good pitch be? What would you give a five out of five?

Doug Downs (11:06):

Five out five is someone who genuinely knows what Brad Williamson has talked about in this piece. In fact, he even asked Doug how he would rank a five out of five, and I think I can build off of that and introduce something that has another nuance to it, more legs to the existing conversation. That would be a five out of five.

Brad Williamson (11:30):

There you go. So you take that prompt and you say, take the transcription of this. Find questions that were asked in this episode. Reference that question and answer that question ahead of time so that your pitch is actually providing value to you, Doug, right? Wow. And then your AI would say, okay, it's relevant to a specific episode because it's looking at it versus the existing transcriptions of previous episodes, and it would be able to say, Hey, compared to this database, this is a match, therefore, it's a good pitch, therefore, you probably want them on your show.

Doug Downs (12:05):

Fascinating.

Brad Williamson (12:06):

And you could toss that right into a Google sheet or something like that. Again, using Zapier, so it would be Gmail to Zapier to open AI, back to Google Sheets, and then you've got this whole list that's basically built this entire workflow around, Hey, I get all these pitches and I just want their best ones. This is a lot.

Doug Downs (12:25):

I know when people write resumes now they're instructed to find the exact key words that are used on the website for the company they're applying to. It sounds an awful lot like that, where we're really cherry picking some key words here.

Brad Williamson (12:40):

Yeah, that's exactly it. And you're seeing this already happen in that world as well, and you're also seeing it in sort of spam, right? You're starting to see a lot of form fills, go on websites and cold emails, outreach that's talking about, oh, I love the systematic approach that you have to digital marketing. I'd love to show us, I'd love to show you our software, et cetera. That's something that we would get at Catch Digital, and it's really talking about the content that's on our website. It looks like someone did the research, and it's really presenting as a specific person. And if someone's really good, they'll talk about using sort of back to the mega promptt, that specific tone and voice that represents their brand. And the more that it feels natural and a real person, the more that people are going to be tricked. Now, AI can typically pick up whether something is AI or not itself, and there's tools specific to that as well, but if you're just talking about purely the content and it's seeing the eyes of an actual human being, chances are you're going to think, oh, this is legit.

Doug Downs (13:42):

When I get an email and it's from, say, Brad, I'm more, well, now we've met each other. Well, let's say we hadn't. If it's from Brad and not just Hello user, I'm more likely to read it. Generative AI can actually send those out now and make it look like it's really from Brad and it's not.

Brad Williamson (14:00):

Yeah. Yeah. So that cold email outreach, I mean, it's going to become stickier and stickier, and there's the bright sides to that and the dark sides to that, specifically with people you, you've seen this already happening where there's scammers out there that are putting Trojan horses on. Oh gosh, yeah. Fake emails that are going out to firms that look like it's coming from the C E O, and it's analyzing all the messaging that they've put out there. Maybe it's just pulling everything that they've talked about on LinkedIn and they wrote themselves, and then it's an email that looks very similar too, and it's like, download this. You know, PDF turns out it's not a pdf, it's a virus, and then they've got issues within their entire organization. So cybersecurity on this side is also quite the issue, but where there's good, there's bad, and this is an early, early industry. You're talking about how this has progressed since November of 2022 to May, 2023. Imagine how it's going to progress in by 2025 and what that's going to do to the marketing industry, PR and business in general. It's insane.

Doug Downs (15:19):

Well, and there's more dimensions to this. It goes beyond just written communication. Audio is in full swing. There was that song created recently that used Drake's voice. It wasn't Drake singing the song. Drake's not very happy about it. It was not Drake's plan, but the technology is there to do that, to repeat it as close to perfect as the human ear could possibly detect and video, I would call video in beta mode. I've seen some efforts by generative AI on the video side. What does the future look like?

Brad Williamson (15:52):

Well, it's crazy. You can actually take an idea that starts with two lines and end up with a full creative production out of this. And there's so many layers to unpack behind this, but number one is today a practical thing that someone could do is they could have an idea for a video and use G P T four to write the script, and then they could take images of that video, or you could ask G P T to give you prompts for images even, and then create those images in mid journey. Then you could use a tool called Runway ML to turn those images into video and then add a voiceover using a tool called 11 Labs. A company called 11 Labs will do the voiceover, and then you can create the background music with Sound Draw, and all of that would be royalty free because it's your prompts and it's your creativity that created these elements. The outcome would be a full

Doug Downs (16:52):

Right, and it's not animation, no,

Brad Williamson (16:53):

It's live video,

Doug Downs (16:57):

And they're using publicly accessible video, which exists already, and publicly accessible photos exists already.

Brad Williamson (17:03):

Yeah, you do have to be cognizant of the source in which you're pulling those from. There's a lot of lawsuits going on back and forth between where that initial source came from, but that's what gives companies like Adobe a bit of a competitive advantage here with their new tool Firefly, because they're using Adobe stock to train that model, which means that they don't have that issue. So it's going to be interesting to see exactly how that plays out, and that's just the tip of the iceberg available today. In the future, we're starting to see advances in gaming and full world production where basically each scene that you enter, say you have a little guy running around on the screen, the next scene is completely created via ai. Imagine when that makes its way into full video production and movies where it's like, I want to be the star of a movie that you're watching on Netflix, and your co-star is Marilyn Monroe, right?

(18:04):

And you had a bad day, so you want it to be a romcom, and you just want a feel good movie, and you could build that prompt to entertain yourself. It's crazy. It's changing the game across the board and the way that that's going to work, its way into the audiences and stuff like that is just insane because it's going to be so specific to that person that relevancy is going to be crazy, and that relevancy is a key thing that we think about from a marketing side as well. When you could think about how web is going to change and when you click on a Google ad campaign and the ad that you clicked and the content was auto automatically created based on your search term lands on a landing page or a webpage that's specifically tied to your needs as the user, the more data that it can capture on that type of environment, the more effective marketing's going to be. That's where I'm really excited about it. From a strategist side of things.

Doug Downs (19:04):

That's the true personalization. Wow. Yeah. Last two questions here. We have about three minutes, the ethical requirements. For example, if I'm currently writing a blog for a client and I'm charging them two hours for each blog post that I write, which I think that's pretty fair, look it take me less than five minutes to do it. Now, am I ethically bound to tell them and charge them accordingly? Second, with the ability to create so much more content so quickly, do we risk a world in which we create too much noise?

Brad Williamson (19:41):

Excellent. Excellent Questions. Number one, the ethical requirements, your example with the blog post. So when you're talking about the results of what that blog post can do, and if you are really, really good at it right now in the market, there's a bit of an opportunity to leverage these tools and softwares in order to create a better blog post using ai, implementing the keywords that you need, driving better, SEO. So if you're specifically being paid to write a blog post, that's one thing, but if you're being paid to increase the SEO and blog posting is the method in which you're using it. That's kind of a different value tier. So if you hired me to write a blog post, for example, I'm going to look at the analytics and I'm going to say, well, here's the gap in the keywords, and this is what we need to produce the production of the blog post itself, the writing, yes, that is getting devalued in the market, and what's going to happen over time is a market adjustment because more and more people are going to catch on to this, and business owners are going to realize, well, I can do this myself.

(20:47):

So the cost of writing a blog post in two hours is going to go down, but the effect of what that can do for a business and the critical thinking and the strategy that comes behind it, I believe I'm a firm believer in using the tools that are readily available for you. So the ethical discussion around that is, oh, you're overcharging for that two hours of time, but if you're turning what that two hours actually does for a business and you change that to something that's more effective and you can increase your value as a result of it, then I'd say that you're fine charging that proper amount paid for

Doug Downs (21:21):

The outcome,

Brad Williamson (21:22):

Assuming that the outcome is there. And you've seen this for a long time. There's an argument of that in the design world as well. It's like, well, why are you going to hire a fancy agency to do a full brand guidelines, logo, colour palette, voice tone, audience research, et cetera, when I could go to a cheap website and get a free logo? It's like, well, you could do that, but are you going to have the depth of the strategy and the outcome that's actually going to help your brand and help your business grow? Right? So that's answer number one. Now, number two, with the ability to create so much content so quickly, are we risking making too much noise? Well, I think there's going to be a shift in what an impression really means. Back in the day, I'm talking six months ago,

(22:11):

An impression. An impression was something that we used as a leading factor of whether a post had success or it was a measure of virality. Right? Now, we're going to see a measure of relevancy, and I believe that although you're not going to see a TikTok video potentially blow up to be 6 million views or what have you, you're going to see more tailored audiences who are much more ingrained in what you are offering or what value you're providing. And so for that reason, smaller audiences more relevant. If you take a step further beyond that metric of impressions and you're like, well, I'm actually trying to drive website clicks. I'm trying to drive form submissions, I'm trying to drive conversions. That's stuff that we talk about in the marketing world. That's a step layer. It's a layer deeper than impressions. You would say, well, it won't feel so noisy if the content is directly relative to that very specific person, and that's what this is unlocking. So short answer is absolutely, we're pumping out content faster than we've ever been able to before, and these tools are just going to exponentially increase that. But at the same time, people are going to find it that much easier to find exactly what they're looking for and those specific, that specific value that they're trying to find,

Doug Downs (23:38):

All the more powerful social media is monitoring my habits as well as I'm convinced my phone is listening to me, and all the more powerful that's going to be to inform the big companies of who Doug Downs really is. So thank you for this, Brad. I really appreciate it. Thanks for your time.

(23:54):

 Yeah, thanks for having me, Doug. Really appreciate it.

(23:57):

If you'd like to send a message to my guest, Brad Williamson sent him an email. It's in the show notes, stories and Strategies is a co-production of J G R Communications and Stories and Strategies podcasts. If you like this episode, do us a favour, give us a five star rating in the directory that you are listening on. Obviously that influences the algorithm somehow. It's a powerful message. Thanks for listening.