Episode 278: AI’s Deep Impact, an Updated Chat with Paul Roetzer, AI Marketing Institute

Welcome to Skucast, the podcast for entrepreneurs in the promotional products industry. Skucast shines a light on our industry's best work, features maverick personalities, and discusses what's really involved in running a modern promotional products business. Skucast is the official podcast of Common Skew. Some of you might remember our friend Paul Raetzer. He joined us on the Skewcast 2 years ago to discuss this new thing called artificial intelligence, specifically AI in marketing. He warned that it was going to revolutionize our work. I remember the episode dropping and then hearing the echo from the cavern of who cares reverberating off the episode stats. But suddenly ChatGPT happened, growing to over 100 million users. Suddenly it seemed everybody cared, or at least were curious. So we invited Paul back to the Skewcast for round 2. Paul Reitzer is founder and CEO of Marketing AI Institute and founder of Ready North, formerly PR 2020, HubSpot's very first partner agency. He's the author of Marketing Artificial Intelligence, published last year., the Marketing Performance Blueprint, and the Marketing Agency Blueprint. And he's the creator of the Marketing AI Conference, MAACON. Paul's mission is to make AI approachable and actionable for marketers and business leaders, and he's consulted with hundreds of organizations. As we talked about in our newsletter, The Backpack, which if you're not a subscriber, you can check out our latest issue at community.commsq.com. Recently, Inflection AI raised $1. $1.3 billion, while ChatGPT's growth mellows a bit. And lately, Claude, C-L-A-U-D-E, is quickly on its heels, which some rank as one of the best out there. But all of these fast and new developments can be overwhelming. What I talk with Paul about today is immensely practical. We talk about the changes coming in design, the impact on agencies like us distributors, the most intriguing tools on the horizon, and the relevance of all of this to our sales and marketing work. Hi friends, I'm Bobby LeHew, Chief Content Officer at Commonsku. Hey, if you're heading to the ASI Chicago show, you don't want to miss all the Commonsku energy happening there. On Tuesday, July 25th, from 12 to 3 PM, we're hosting Commonsku University where you can spend a few hours learning about the power features on Commonsku and how to drive sales through shops and a whole lot more. We also have this community happy hour happening on Tuesday, July 25th You can register for all of this, plus grab a handy list of all the connected suppliers who are exhibiting at the ASI Show with their booth numbers in one convenient spot at commonscue.com/asishow. Now here's my chat with Paul Ritter. But I have to tell you a quick story real quick. Paul and I met— this has probably been a decade ago, Paul, when you and I met.

We—

I was serving on the board of AMA, and we brought in Paul as a speaker when released his first book about a marketing agency, because Paul and his company then were the first marketing agency, uh, connected part as a partnership with HubSpot. But we brought in— remember this promotional marketing campaign? Since we're talking to marketing professionals who are in the promotional, uh, product business, we did 100 of Paul's books and we co-branded those through the book company and we gave those away to attendees. And Paul, I think I sent out copies of those like all year long just to other marketing— because our, our clients were marketing agencies, right, and marketing professionals and things like that. So it was a lot of fun.

It's great to bring that full circle. It was. I still remember you picked me up and we went to like a farm-to-fork restaurant. Yeah, yeah, it was great.

We had a good time.

Yeah, good time.

Well, it's good, it's good to see you here. And, and my gosh, how things have changed. Let's get into this real quick, um, because one of the things, uh, let me introduce Paul officially as the founder and CEO of the Marketing AI Institute, founder of Ready North. I mentioned he's formerly PR 2020, HubSpot's first partner agency. He's the author of Marketing Artificial Intelligence The Marketing Performance Blueprint and the book I just mentioned, The Marketing Agency Blueprint. As a speaker, Paul is focused on making AI approachable and actionable for marketers, which is why we love talking to him today. Now, I want to start with these questions, Paul, that I thought was fantastic. You have an amazing chief growth officer named Cathy McPhillips. Cathy and you recorded an episode a few— a month ago or so where you have been hosting AI intro classes for marketers since the fall of 2021. And she took all 147 questions from this, ran them—

Just from the episodes from this year. That was only from this year. Yeah. All right. We've done that class 27 times. It was from like 6 of them. We combined those.

Yeah. Okay. So it took all 147 questions from those classes, ran them through ChatGPT, or 147 questions year to date, ran them through ChatGPT to consult them, consolidate those into the most essential. There were 15 core questions. Now, for our audience, since I know our audience well, I pulled about 5 or 6 that are real relevant to us. So let's start with that. Given this rapid change of, in marketing technology, I mean, what advice would you offer to us as businesses trying to navigate the influx of AI tools in the market? How can we, because you're talking to CEOs, you're talking to salespeople, you're talking to marketing professionals, how can we at this point build a coherent tech stack?

Yeah, for everyone, I always say it starts with understanding and education. Like, those are the critical things. So you really need some people in the organization, people on the team that have a strong comprehension of what AI actually is. Because once you have those people, then you can move forward trying to figure out what are the best use cases where you could be looking at infusing technology. How do you go out and find a few of those players that you can integrate in and start piloting within the organization? So an example, that would be like if, you know, AI copywriting, say you do a ton of copywriting for emails and newsletters and landing pages, whatever, then finding an AI writing tool might be your initial pilot project where you're going to focus on. So that's how you really want to get started, is get the understanding, whether it's coming to an intro class or doing a webinar like this. Just get some fundamental basics. You don't have to spend months on this. You can do it in a day, learn the fundamentals, and then you become more of an educated buyer and consumer of the technology and the use cases.

There are now thousands of apps being released. I mean, how, how can we discern which ones to invest in for our company's needs? It's overwhelming.

It's really hard, honestly. I mean, we've been tracking these companies since 2016 when I started the Marketing AI Institute, and it's really hard to keep up with the technologies. The thing I often tell people is I start with the core tech you already use. So for example, Microsoft and Google. So if you're a Microsoft 365 or a Google Workspace user, they are building generative AI capabilities to write, to create images, to create content right into their platforms. So you will have those capabilities just through the existing companies. If you're a HubSpot user, they're gonna have ChatSpot. If you're a Salesforce user, they have Einstein GPT in their AI cloud. Like, use the trusted providers you already have because every SaaS company in the world is racing to infuse these AI capabilities right into the software. So start with the companies you already trust and that are already in your— have already gone through procurement. That's gonna be the easiest path forward. From there, you can find those use cases where you're gonna get a ton of value and then go search AI for newsletter writing, AI for landing page development, AI for advertising, whatever it may be. So that's how I would advise.

I, I love that advice too because, if we're not careful, we're gonna build this Frankenstein of a tech stack. And, and there's, and the proliferation is happening so fast that these tools are gonna be built into a lot of our most trusted providers. A lot of us work with designers and we either have designers on staff or we contract with designers. Can AI replace or supplement roles like graphic designers? I mean, how would you talk to entrepreneurs in this space about it? I remember you saying— I went to the Writers Conference, which— yeah, which AI for Writers Conference went— you had like 1,000 attendees or something like that, 4,200 registered. Yeah, so, so this is like 3 months ago or so, uh, and I remember you said this at the Writers Conference. You said AI won't replace writers, but writers who use AI will replace writers who don't. Is that the same thing you think with graphics?

I'll let you— Yeah, any— replace any knowledge worker in that and it probably works. So designers, illustrators, coders, whatever, whatever it is. So I think that the way what's gonna happen is professional designers are going to become even better at what they do and do things more efficiently and do things in entirely new creative ways. People who aren't designers like me will have some basic capabilities now. So for example, if I wanted to do a logo design, I am not going to create the final product myself as a non-designer. I'm going to work with a graphic designer to develop this logo. But now I can go into DALL·E 2 or Midjourney or whatever, and I can start building out some concepts. So rather than me sending a brief to a designer and saying, here's what I want, here's what's in my head, and I'm terrible at explaining this to, to artists and creatives, I can now go to them and say, listen, I played around with it. Like, here's the prompts I used to explain it. Here's kind of the outputs I got. I really like the direction here. Can you kind of now take and do what you do? So it's almost like it's enhancing my ability to give a project brief to a creative person. Interesting. I don't see the need to not have the creative person. Yeah. Now that creative, that designer might be able to do 3 to 5x the amount of work they used to do. So the question eventually becomes, well, do we need as many designers if you're a graphic design firm? My hope is, yes, you do. You find other things for them to do, or you just increase your productivity as a firm. So yeah, I don't— but I do think designers who don't infuse these capabilities are probably gonna get left behind in the near future.

How do you see this impacting agencies who are doing work on behalf of clients? I mean, that's a whole big topic, right, on copyright and different aspects, but also just staying on top of things.

It's a big— it's a huge challenge and an even bigger opportunity, I like to think. So the challenge side is a lot of agencies still function in a billable hour type model where they're basically getting paid for the output, the time it takes them to build an output. Either they charge by the hour or they're factoring hours in when they're doing their quotes for their work. So this is incredibly disruptive to that because if I was charging to write an article and I was charging $1,000 for a 1,000-word article, just for rounding sake, and that used to take me, say, let's say 7 hours for all in, plan it, write it, edit it, whatever, get it approved. Let's say that I can now use AI to assist me in the development of that, and maybe it now takes me 3 hours. Am I still charging the same I was for the 1,000, or am I changing my whole billing model? So there's these fundamental issues around billing and operations and efficiency and teams. But the opportunity to me is Every company, every size company in every industry around the world is trying to figure out artificial intelligence right now and they have no one to turn to. So if your agency can become a resource for them and actually help them solve this and you can build some services and consulting around all these challenges they're gonna face over the next decade, then you can flip that from being afraid to being like, wow, this is a huge opportunity.

Let's go. Yeah, let's back up just a little bit for those that arrived today and they're like, you know what, I trust CommonsQ. I like this Paul guy. I'll give 30 minutes of my time to figure out why I should pay attention. And let's just ask this question for those folks. I want to get started today. What are some steps I should like immediately learn or identify ways I can jump in and just start doing it without getting overwhelmed?

I would just look at the things you do every day as a professional, as a practitioner, as a leader. So, you know, let's say that you're on the content team and you're doing a lot of writing or you're doing the newsletter, you're doing the ads. I would just look at the things you already do and say, okay, I'm spending 20 hours a month on this and 30 hours a month on that. And I spend another 10 here and 5 there and 50 here. And I would just start with the things where most of your time goes. And I would look at it and say, okay, is anything here repetitive, data-driven, making a prediction or generating content of some sort? Like those kind of the 4 variables I look at. Yeah. And if the answer is yes to any of those, like let's, so the example I like to give is the podcast. We have 21 steps that go into doing the podcast planning and production every single time. About half of those steps we now use AI in. And so it took our podcast from, you know, let's say 15, 20 hours per episode, probably down to, you know, 7 or 8 hours an episode. Didn't replace anybody's job. It actually freed Kathy up to go and spend more time on one-to-one relationships with our community members and engaging with people. Like, that's how I see the value of AI is redistribute that time to something uniquely human.

Yeah, I think that's phenomenal. And I love how you say start where you're at. Like with the podcast, we've been using Descript. Writer, writer we've been using as well. And there are 2 or 3 other apps sort of in our content arsenal right now.

It doesn't have to be anything like totally transformational. It's just like you save an hour or 2 here or there and you don't have to do the repetitive thing you didn't really enjoy anyway.

Yeah, yeah, exactly. And as things iterate, it's just going to get even more, even more better as, as we sort of move toward an increasingly AI-driven world. You kind of mentioned it here, how How do you answer folks that are threatened by AI? I think that was the initial reaction of a lot of folks. They were like, there was so much doom and gloom in the media about the headlines. How would you encourage professionals today?

Yeah, it's still a very natural reaction. I deal with it every day when we talk with some of these bigger companies or we're dealing with AI for Writers Summit. That was the reason I launched it, is I knew a bunch of writers were afraid for their jobs. Like, okay, we gotta have this conversation out in the open. So, I mean, I think the best way to deal with that fear and anxiety is to develop a comprehension of what the technology actually is, because then you start to realize like, okay, there's ways around this. Like if I can use it to enhance what I'm doing and it's not really gonna replace me, then you start to get rid of that fear. But I think fear comes from uncertainty. And so, so often corporations don't know what to do about this or how to do it. And they're not communicating with their teams and their teams are maybe using ChatGPT and they're not even sure if they're allowed to be using ChatGPT. And so there's just all this uncertainty because there's a lack of communication and education. And I think if we just solve for that first, then we can start to break through these barriers like the fear and anxiety that comes with the idea that maybe I'm not going to be needed anymore to do my job.

Yeah. Switching gears a little bit, you outlined 5 major categories of multimedia content that AI will impact the most: language, image, video, audio, code. So let's play a little game, Paul. You analyze so many tools. This may be a dangerous game to play. What do you—

what—

when I mention each of these, what— which platform or tool comes to mind? If I mention image, what's the most progressive or amazing thing happening right now? Is it Midjourney? What comes to mind in terms of apps?

Yeah, I mean, certainly DALL-E 2 kind of created that space, like, you know, and that again, that was only April of 2022. It's kind of crazy to think back. Midjourney, from a quality of output, seems to be quite a ways ahead of everyone else. But again, going back to the tech you already use, Adobe, it came out with Firefly, they did generative fill. So if you're an Adobe user, you're probably just focusing on trying to take advantage of all the things Adobe has in mind. So yeah, there's a lot of players there. Midjourney is definitely seems to be in the lead in terms of the quality of the output, but there's lots of players.

We mentioned audio with Descript. Are there any others that come to mind?

So audio, another one thing I think about there is RunwayML.com does audio as well. But the ability to create music is a really kind of emerging area. So like music ML, I think Meta, Google, there's a number of players that have these capabilities and they're evolving. So I think about them also on the audio side.

How about on language?

Language is, yeah, there's so many players, but the way to think about that one is you have language models and that's like OpenAI, Cohere, Anthropic, Google, Amazon, Microsoft. They're building their own models.. And then on top of those models are the software as a service companies or the application layer companies like Rytr and HyperWrite and Jasper. So that's kind of the two areas there, language models and then the application layer above them.

Speaking of like with Rytr, video is the next category. And with Rytr, you know, I'm able to take a conversation like this, drop it into Rytr, and it's able to pull out highlights from our conversation that would've taken a videographer, you know, an hour or two to like clip and cut and splice and all those kinds of things. So what else comes to mind when I mention video?

Yeah, so video, I mean, definitely Descript. Runway is doing text-to-video. So you're starting to see the ability to actually create video based on this stuff. So those are the first that jumped to my mind.

Yeah.

And finally, code. Yeah, code. So again, if you're into code, you have GitHub Copilot, you have Replit has Ghostwriter. The— there, that's a major area because you're seeing massive increases in productivity and coding where that's just there as an assistant. So yeah, those are the two that immediately come to mind though.

How do you see— how do you see AI changing search and SEO?

It's a really— I mean, it's like the trillion-dollar question. Certainly Google is already starting to evolve. You're seeing the infusion of GPT-4 into Bing. So it's already starting to change it where it becomes more of generative AI output sort of replaces the 10 blue links, I think is the most obvious way this affects search moving forward. I think the bigger question is what is, what is the consumer appetite for information become and how do they go about seeking it out? So for me, navigating through 10 blue links feels old already. Like it feels like an outdated way to find information. I much prefer going into ChatGPT and asking a question, or Google Bard, or Anthropic, or Claude, or whatever it is, and asking a question. But I also know that those models get stuff wrong all the time. So I can't trust them the way I would trust clicking on a blue link from Harvard Business Review or MIT Tech Review and reading it and be like, okay, this is a legitimate source. And I'm curating this information myself and learning it. Whereas ChatGPT may give me something that sounds completely author— like, high authority, high confidence level, and it may be completely wrong and made up. Yeah. So that's the challenge we're at right now is these language models aren't trustworthy yet. And so they can't replace the way we have traditionally used search. But as soon as you solve for that and these things become 99% accurate or whatever that number needs to be, I could see the way we've historically done search just disappearing.

Since you are speaking to an audience who bridges the marketing or in the marketing world, and since you bridge the marketing world and AI, is there anything we can be doing right now to change? Should we be doing anything to change our approach when it comes to SEO?

The thing I've told people is I, I think you have to assume a drop in organic traffic. So anybody who's involved in driving organic from the SEO team to the paid team to the content team, you have to assume there's going to just be this natural drop in traffic over time and probably a continuous drop as people start searching for things in different ways. Nobody seems to know if that's going to be 5 percentage points or 50 percent points. Like we just, we just don't know the drop. So the thing that seems to me that is going to be sustainable is what I would say is more human content, the stuff that's not easily fake, the stuff I know ChatGPT didn't write and you're just dropping on your blog. So that's things like podcasts, webinars, in-person events, videos, editorials with strong points of view from leadership, like things where you know a human is behind it, because I think that people are going to crave stuff they know that humans made. When all of us can create AI synthetic content for nothing, when it's basically commoditized, we're gonna wanna know someone put the time and energy and thought into creating a piece of content.

Yeah. We are seeing some suppliers and manufacturers in our industry use ChatGPT to write copy for behind the scenes search engines that are proprietary search engines toward our industry. So searches a million products. So it's really interesting. We have an episode out there about that. But how are you and your team using AI daily? Are there any other tools that you and your team use on a regular basis?

So we test a lot of the AI writing tools. So I don't personally use it that much in my daily workflow. Most of my writing is LinkedIn posts. I usually try and do a post a day and it's oftentimes 300 to 500 words of just commentary and tie, you know, kind of context around something happening in AI. I don't use AI at all for that. I like the process of writing. I think by writing. So it hasn't replaced anything there. But I do use it for really interesting use cases, like for example, for our conference, like drafting an abstract for a session. So I gotta come up with an abstract for it, you know, give it a prompt and see what it comes out with. I use it a lot for outlining, for strategy development, for planning. We do use it for summarization of transcripts, which you kind of alluded to earlier. We use it to summarize those transcripts and turn them into blog posts. So there's all these individual use cases, but they aren't a writing replacement for us. They're, they're all these others that support writing. And I think that's how we continue to look at it moving forward is as an assistant and making a lot of the other areas more efficient.

I had a young entrepreneur that was speaking and I asked him for a bio and he said, man, I don't have one, hang on. And a minute later comes in an email and he said, I just asked ChatGPT to create my bio. That was great though. It put him on it like it just— it's something he would have agonized over, you know. It would have been a long project. Talk a little bit about this end of the world sort of doom speak. There was a great article in Financial Times that you linked to on LinkedIn where this constant doom saying is a massive distraction. This was from this Financial Times article, particularly for those of us who are trying to make AI tools effective for our use.

What's your perspective on Yeah, it's definitely captured a lot of the mainstream media headlines in recent weeks and months because there has been some high-profile people in the AI research community who have sort of stepped forward and said that they fear the work that they have done is sort of leading to a potential existential threat to humanity. I mean, that was the exact words they put it on par with pandemics, nuclear weapons, and AI. My, my general feeling on this, and I've thought a lot about this, I've read everything you can read on it, and I've studied most of these people talking about this stuff for the last decade. Um, I think it's real. I think that they truly do believe there is an existential threat to humanity from these models if they advance the way that we think they could. If they become kind of super intelligent, superhuman, uh, then it is a realistic thing we should be thinking about. But so is an asteroid hitting Earth and wiping out humans. Like, right, it's not that it's not possible and that it isn't something people should be thinking about. But the stuff we know is here today is disinformation, synthetic media, you know, affecting our democracy and threatening the next election campaign, bias, the ability to, you know, spread disinformation rapidly online. Like, all of this is like, it's here that we don't even need to advance the current models to deal with that. So my general feeling is I'm really glad that there are some AI researchers and ethicists that are thinking about this long-term potential impact that may or may not happen. I feel even better to know the vast majority of them are working very intensely on trying to solve for the issues we are dealing with right now around deepfakes and synthetic media and elections and democracy, because that's the stuff that we know is real, and it doesn't take a leap of, of faith or technology for it to affect us.

Um, you seem so fascinated by this still, something you've been working on for years. This isn't something that you just started doing. You've been doing this for a while now. You ran a successful agency, you were HubSpot's first agency partner. You've been on the cutting edge of some radical concepts. Is there something else you're keeping an eye on that's not AI-specific in Paul's world?

Um, I don't know that anything moving forward that'll matter isn't— doesn't have an AI component to the story. I feel like it almost becomes the underlying architecture that enables a lot of things, but quantum computing is the thing that has been on my kind of horizon for a while. And I do still pay attention to that. All the same players who are investing heavily in AI are also investing heavily in quantum computing. And I think over the next 5 to 10 years, that may start becoming like a real thing. So that's the thing I will read a little bit about when I have time. And then I just, Space exploration is fascinating to me. It's what we're learning about the universe and the origins of time and all that stuff.

I love that stuff. Uh, our community loves this question. Last question for you, Paul. Give us a, you know, you read a lot. You're, you're, you're a research hound. What's a book, article, or documentary or podcast episode lately that's really impacted you?

Uh, I'll, I'll say, um, there's this professor, Wharton Business School, Ethan Mollick. He's actually going to be one of the keynotes at our conference this year. He has a blog called One Useful Thing. Honestly, guy, like he's, he's the kind of person, like every single thing he tweets or puts on LinkedIn, I could retweet it. Like it's just, it's packed with like value and unique insight every single time. And so he writes this phenomenal newsletter and you could just go back and read the last 10 or 15 articles and your mind will just like melt. Like it's, it's so good though. And it's so approachable the way he does it. So I would say his, his newsletter would be a great one.

Awesome. Paul, thank you, my friend. Thanks for joining us again today, yet again for another episode, and we appreciate your time.

Happy to do it anytime. I'm sure we'll do this again in a few months.

We can answer these questions in all new ways. That's exactly right.

Thanks, Paul. Take care of yourself, my friend.

Thanks so much for tuning into this episode of SKU Cast. Be sure to keep up with our latest content by subscribing to SKU Cast on iTunes or to our blog at community.commsku.com. Until next time, friends, thanks so much for listening.