Subscribe on

Enjoy this episode? Leave us a review.

All Episodes

Episode 91

The Truth About AI in Marketing with Shelly Palmer

AI won't replace humans. But it will dramatically change how we work. And brands that don't commit to AI will eventually find themselves struggling to keep up with savvy competitors, according to tech expert Shelly Palmer. So what’s really at stake here?

In this episode, Elena, Angela, and Rob are joined by Shelly Palmer, CEO of The Palmer Group and professor at Syracuse University's SI Newhouse School, to discuss AI's impact on marketing. Shelly offers an unflinching look at why marketers are hesitant to adopt AI, how the technology will transform creative work, and what the ongoing battle over AI and copyright means for the industry's future.

Topics Covered

• [02:30] How technology and music launched Shelly’s career

• [12:00] Why AI is transforming business productivity

• [17:00] The difference between human creativity and AI execution

• [21:00] Sam Altman's prediction about agencies

• [27:00] The ongoing battle over AI and copyright

• [32:00] Creative ways to experiment with AI tools

Resources:

2024 AdAge Article

“Think About This” Newsletter

Shelly Palmer’s LinkedIn

The Palmer Group

Today's Hosts

Elena Jasper

VP Marketing

Rob DeMars

Chief Product Architect

Angela Voss

CEO

Shelly Palmer

CEO of The Palmer Group

Transcript

Shelly: It's gonna get worse. Because right now it's doing the simple stuff, which is training off of other people's work. But the reasoning engines are going to learn to reason - nobody knows what's going to happen.

Elena: Hello and welcome to the Marketing Architects, a research-first podcast dedicated to answering your toughest marketing questions.

I'm Elena Jasper around the marketing team here at Marketing Architects and I'm joined by my co-hosts, Angela Voss, the CEO of Marketing Architects and Rob DeMars, the Chief Product Architect of Misfits and Machines.

And we're joined by a guest today, Shelly Palmer, an expert across technology, media, and marketing. Shelly is a professor of advanced media in residence at Syracuse University's SI Newhouse School of Public Communications and the CEO of the Palmer Group, where he advises Fortune 500 companies on emerging technologies, media, and marketing.

Recognized as one of LinkedIn's top voices in technology, Shelly regularly shares insights on CNN and Good Day New York and runs a very popular daily business blog. He's a bestselling author, award-winning composer, TV producer, and much more. So this conversation today could be wide-ranging, but we'll try to keep it focused on advertising and marketing. Welcome Shelly.

Shelly: I think I just need you to say that every time I go anywhere.

Angela: She's very good at that.

Rob: Now I have to ask a bit more though about the music, because I definitely have known you for all of your thought leadership that you've done. But the London Symphony Orchestra at Abbey Road Studio - I mean, how cool is that? And as a product of the eighties, I've watched more television than anyone else in this podcast. And it sounds like you've done a lot of TV jingles. Give me one, I know Shelly, give me one I know.

Shelly: They're going to be written on my tombstone, but I've worked on things like Meow Mix and Seagram's and Burger King. Also the CBS News theme, MSNBC, TNT, AP Radio. Going back, did all the music for Spin City, Regis and Kelly, Regis and Kathie Lee, and everything with Regis and Kelly, and now Kelly and Mark.

Rob: So how did you go from this music prodigy to technology prognosticator, if you will?

Shelly: I don't even know how I describe myself now. Look, mom and dad - they met at Juilliard, they owned music stores. So they were my first music educators. I grew up with parents who owned half a dozen small retail music stores and a piano shop. I earned my allowance tuning guitars on Saturdays.

Sometime around when I was maybe 12 years old, I got to see my very first ever Moog synthesizer in person. It was really a lot of money. My father said it would never sell - no one would ever spend that kind of money on a piece of technology that no one understands. And I started to play this thing, and I just had to have it.

When I was about 17 years old, I was walking through Radio Shack and I still have the book. It was 79 cents. It was a book about VCOs, voltage controlled amplifiers, oscillators, and filters.

And I asked the guy behind the counter if they were the same voltage controlled oscillators, voltage controlled amplifiers, and filters that were in my synthesizer. The reason I wanted to understand it is that these were integrated circuits on the cover and the synthesizer was all hand wired with printed wiring boards and component parts.

There were no integrated circuits - fully analog and I ended up buying that day about $50 worth of parts. And I took it all back to the music store where I dumped everything on the bench. And I said to Walter, we're going to computer control the synthesizer. He goes, how? I said, I don't know.

I bought a couple of books, let's see what we could do. And so that's my technology career in a nutshell. At 17 years old, I started to figure out a way to digitize the patches that you'd need to make an analog synthesizer have a sound and store it so you could get it back.

Ultimately my very first patent claims were written around that technology. And over time I was composing pretty regularly. Even as a 17 year old, I was writing music commercially.

As I got into the business, I went to NYU film school and got my degree in film TV production. My first job out of school was working for an amazing, unbelievably talented guy named Don Elliott, who was a super famous jazz musician and owned a jingle company. I started working, literally playing jobs for him as well as repping his reel - you'd go around to agencies trying to get work for him.

And then when I got work for him, I was allowed to write a demo and was able to throw in my own to see if I could get any work, which I did. It was kind of fun. I left Don and got my own production company.

Now we're in the early eighties. So by '86 we had put the very first ever totally tapeless recording studio online.

And it was really expensive to do and it took a lot of tech, but I learned my tech. One of the things that's fascinating about the early part of my career was that we were working on sampling - it didn't exist at the time. It's just a fancy name for digital recording.

So by '86 I was lucky to do a lot of really good work for a lot of big agencies. One of the agencies had their research department. We had a very big control room and we had a really good audio chain. And I had invested in an extremely expensive television monitor. It was a 16 by 9 cathode ray tube. Now this is before anyone had a flat screen.

We're talking the 80s now. So this was a 16 by 9 format, which no one was using - everything was 4 by 3 back then. Everything was analog. This was a color correction monitor, a Sony color correction monitor that literally cost $40,000.

If you were an ad agency, the best place in New York City to see your spot finished to show your client was in control room A at Creative Audio. Because the monitor was the best TV monitor you could buy for money. And the soffit mounted Westlake speakers were the end of days from a sonic perspective.

We used to let the agencies come for free at night to have playback parties for the finished spots. There was an argument going on between one of the agency leaders and their research group about some optimization. And I laughed and said, "Oh my God, that's just an easy regression."

We have enough computational power in the machine room to do that right now. We're doing that a couple thousand times a second to make this work on each of these machines. They're like, what? It's like, yeah, we could definitely do that. It's just silly math.

And so we did this media optimization algorithm for them, which I literally coded while they were sitting there. I got them their output and printed it out for them and they said, "Oh my God, this is amazing! We need this!" You just need a computer. You literally just need a computer. You shouldn't be doing this on a spreadsheet. You should just code this. It's easier.

We became the darling of that whole group because the math is the same math as we were using for sampling - we were regressing lines and looking for R squared. It's just what it is. So anyway, it was really a very funny moment - a bunch of people who were not mathematically gifted talking math in a room filled with people who did math 24 hours a day.

Angela: You just blew their minds. That's amazing.

Shelly: It's just a different thing. So anyway, that put us into one place that we didn't know we were ever going to be because all of a sudden we started getting much deeper into the mathematics and the mechanisms of distributing advertising and marketing materials.

And what really took it over the edge was right around 2000, I had patented a lot of my tech and the Disney organization decided they wanted some of it. And they, in a very stealthy way, tried to buy it without letting us know it was Disney.

So that's when my career really changed because I'd been doing music since I was four years old. But when Disney came knocking, here was an opportunity to see a patent that I had created in '93, which taught this thing called the Open Cable Application Protocol, OCAP, and it basically turned set-top boxes and headends into client-server networks and allowed you to do data distribution, interactive television, basically on set-top boxes.

So Disney had this thing they wanted to call Enhanced TV - this ability to do things like play Who Wants to Be a Millionaire in sync with the broadcast through your set-top box or on your PC. My patent covered all of that. And it was the earliest patent in the space and they wanted this patent because they didn't really want me to sue them. They wanted to buy the patent. So we made a deal and ultimately I went to consult for them and I couldn't resist.

I closed my production company and I was really excited to go consult for the Imagineers and watch my technology get reduced to practice. I thought that was super cool. And so I wrote a book called "Television Disrupted: The Transition from Network to Networked TV" and started to get on the speaker circuit.

In less than a decade, it went from no one knowing what any of this was to fully invented. I've just been enamored with and immersed in all the tech around advertising and marketing, and we've been using artificial intelligence - what they're calling AI today. We've been using machine learning since the late seventies to do the work that we're doing. Real neural networks probably for the last 15 years or so, transformer networks maybe since somewhere in there. Everyone thinks AI was invented on November 30th, 2022, but it really wasn't. It really wasn't.

I'm not scared of artificial intelligence. I'm scared of artificial control. So is everybody else. No one wants to be controlled by this nonsense. Nobody.

Rob: Is that why people are freaking out as much as they are? You see Toys R Us or Coca-Cola spots and people love to hate on them. Is that just because they're early adopters or people that aren't early adopters freaking out?

Shelly: You know, Rob, I think that's still human arrogance. I think that's still human beings thinking "A machine's never gonna be as good as us." Again, it's a point of view and I respect it. People say it to me every day. Every day, someone tells me, "Well, you're so high on AI, but it's never going to do this." This is like, A, I agree. And B, it doesn't have to for it to have a dramatic financial impact on how everybody I know spends their day in business. Like my entire client base, everyone that we work with, it does not need to be much better than it is right now. Data sets need to get better.

So workflow and process needs to be orchestrated and plugged together in ways that they aren't now. Like, if I say to you, "Let's search the web," you kind of know what I mean. I say, "Let's Google something." You know exactly what I mean. Well, there are no phrases like that anywhere in the world of AI, but when we get there, when I can say, "Let's AI something," what that's going to mean, and we all know it as well as we know what "to Google something" is, we're going to be in a different place. We're nowhere near that, but when everything is wired together, if it was just as good as it is right now, the economic impact would be stunning.

Angela: Yeah, it's just that transition and maybe fear of what's to come not knowing, but then there's different sides of this debate, right? There's ethical, there's legal. You've got some artists like Imogen Heap telling the world she's going to empower her songs to go make love with other people. She's going to create an AI twin.

Then you've got artists that are just absolutely losing their minds over what's going on. And I think you've called this the fight of the year, the fight over IP and copyright. What do you think is gonna happen here? How do you think the world of copyright and IP are gonna change?

Shelly: It's going to get adjudicated in the courts. And so it doesn't matter what I think. It matters what the judges who are involved think. Look, this is the problem by itself - what's ethical? Who do you pay? And if so, what gets paid? And the problem is, the data is not stored in a way where they can tell you flat out that it's learned from you. Here's something you can do as a test: You ask any of the models, ChatGPT, the paid version, who Shelly Palmer is - it'll tell you. Ask it if it can write like me.

There's one or two answers it'll give now, depending on which version you're using. It'll either say, "While I can't write directly in his style, he writes in the following way, and I can write in that style," or it'll just start writing in my style. When it does, depending on the version you have, I can't tell that I didn't write it. Who do I go see about that? And I'm not famous, but like, if I was, I'd be pissed. I've written 2.2 million words, according to ChatGPT, that it's read. And it literally can write in the style of - I write a blog every single morning, as you know, since 1996, literally five days a week, I write a rant.

I write a thought leadership piece on Sundays. Saturdays it's a recap of the week. Ultimately, I write over 4,000 words every single week, closer to six and it's read them all, and it can write in my style. Now, for me, I don't think anybody's really going to go do that, but what if you're Michael Crichton's estate? What if you're Aaron Sorkin right now? What if you're Michael Lewis? I mean, these guys are alive, they're kicking, they're amazing writers. And when you ask ChatGPT to write like Aaron Sorkin, it kills it.

So who do you go see about that? And what would you do to stop it? And I don't know the answer to any of those questions. And I think the courts will decide. The Copyright Office tried listening sessions and they invited everyone to comment, and Shira Perlmutter, the Register of Copyrights, they were really good about asking everybody to come on down and give their opinions.

I was asked by Paul Williams to moderate the AI symposium at the ASCAP annual members meeting. And I was basically booed off the stage for telling the truth. Like, nobody wants to hear this. Nobody. It's gonna get worse because right now it's doing the simple stuff, which is training off of other people's work.

But the reasoning engines are going to learn to reason, and nobody knows what's going to happen. There's one line in the Constitution that talks about IP and it's granted to humans, not to machines. So now I can use a machine to infringe on your copyright, but I can't copyright the work product that's infringing. So I just built something that you can sue me for, but what I just built, I can't copyright because the machine did it.

Oh my God, this is a hot mess. And it's going to be a hot mess. I think I could fairly argue both sides of this cause I'm on both sides. And it's terrifying to me because it's not going to end well.

Elena: Yeah, let me move us to a lighter note to end the episode, but we appreciate you giving your point of view on that. I know you said you didn't have an answer.

Shelly: I have questions, Elena. I literally don't have answers. I have a billion questions about it and I have a point of view cause it's my lived experience, but that doesn't make me more qualified than anybody else to talk about it. It just means I have a deeper point of view, but it's no more valid than anybody else's. And this is going to be a fight. Trust me.

Elena: Well, to wrap up the show, Shelly, we have kind of a fun question for you. Outside of marketing, what's one world-changing task that you're most excited about AI solving or AI helping with in the future?

Shelly: Outside of marketing?

Elena: Yes.

Shelly: I love to debate with it. It is fabulous to argue with. So you give it a personality. You're acting as Stephen Hawking and I want to dispute your assertion that Hawking radiation exists. And then you have an argument with it. "You are acting as a group of physicists, including Marie Curie, Isaac Newton and Albert Einstein. I want to talk about..." And you fill in the blank and this thing will go at you. It is so fun.

Like you invent your own games. So you literally can invent any scenario. I did one the other day that I thought was really fun. I was trying to build a playlist building tool for Spotify. I wrote a prompt that said "you are taking on the persona of a genre of music known as bebop."

Do you know what bebop is? And I said, yes. Now I want you to become bebop as a character. And we are going to go listen to jazz in the 1950s. And we're going to talk about what this music means and who's there. And all of a sudden I'm talking literally to the genre of music.

Angela: That's amazing.

Shelly: And we were having so much fun. To me, that kind of approach isn't like a synthetic companion - it's turning this into things that can't exist in the world. How could a genre of music be alive and have a personality and talk to you? Except bebop has a point of view musically and music is a language. So it interpreted what it felt like that would be as a personality.

Angela: What's your go-to platform to do that on?

Shelly: You can do it on anything. You can do it on Meta.ai. You can do it on Claude - Claude's good at it. GPT-4 is good at it. It almost doesn't matter. Each one's slightly different. I play with all of them. There's about 150 foundational models out there.

You can go to Hugging Face and pretty much get access to all of them. They're all over the place and each one will do something slightly different. But my favorite thing to do with AI is push its boundaries to push my boundaries. To think of something that literally cannot exist in the universe without this tool.