Banner Image

Podcasting in the Age of AI

Podcasting in the Age of AI

  • Dec 11, 2023
Nathan Gwilliam
Nathan Gwilliam

Nathan Gwilliam: Hello, Incurable Creators, on today's episode, I'm joined by Jim Sterne. Jim is the President of Target Marketing of Santa Barbara. He's the board chair emeritus of the Digital Analytics Association and the founder of the Marketing Analytics Summit. He's also written a dozen books on digital marketing, including Artificial Intelligence in Marketing.

And that's what we're going to talk about today. We're going to talk about AI for podcasting. Thank you so much for joining us, Jim.

Jim Sterne: It's an honor to be here. Thank you so much for having me.

Nathan Gwilliam: So can you start off with sharing with us your journey as it specifically as it has to do with podcasting and AI?

Jim Sterne: My journey is pretty straightforward. I realized that I know a lot of people in the marketing digital marketing arena, and specifically marketing analytics. Knowing a lot about marketing analytics led me into machine learning, which led me into generative AI or the chat GPT kind of AI large language models that we know now.

I am a proponent. I am a consumer of a vast amount of newsletters and just try to keep up. And also because I know all these people. I started a couple of different programs of interviewing them. I didn't really want to start a podcast officially until I knew that I could do the production side that I'm comfortable in front of Zoom and a microphone, that I've got a good format for interviewing people.

And now that I've done about 20 interviews on the Marketing Analytics, Live online show. And about 10 on the data driven leader studio side. And I've worked out the kinks of the format. Now I'm comfortable. Okay. I know how to do the creation, the production, the distribution.

Oh, okay. Now I've got, and I'm a little queasy. So in the middle of this queasiness along comes this amazing flood of AI tools. We are, whether you're using Riverside to record or Descript to edit, or literally dozens of tools every day, it's now so easy. It used to be that I would go through a transcript, I'd do a Zoom recording, it would automatically be fed to, which would create a transcript, and I would go through the transcript and highlight things for my video editor to cut out. And, oh, and take this section where we talk about this and move it to the front and put the commercials in here. Then along comes the tools where it takes me less time to actually do it myself than to go through the transcript and instruct somebody else.

So I'm saving time. I don't have to pay the third party. And now, okay, great. This is working now. It's time for distribution about which I know nothing. So are there tools out there that can help me? Yes. And suddenly I've, I'm up against this interesting wall.

So I have started a new consulting practice of helping large companies create a strategy for adopting generative AI capabilities. I don't have a playbook. It's going to be different for every company, but I have a method for which they can create their own strategy. One of the tenants of that, then one of the main points, the things that a company needs to do is to create an AI council of people who are dedicated to keeping track of. What tools are good for what purpose to provide policies and training to the rest of the company. I feel like I am now the rest of the company and that's how we met. I was intrigued by the fact that you can help me with distribution and that I can count on you and your company to keep track of the tools and use them.

I really don't want to have to learn a dozen new tools, please. So if I can get companies that have been around for five years doing podcast recording and editing and production, that's great. But if I can come to a company that's specifically using AI tools, you can do it faster, better, cheaper than the old companies can and better than I can, because I don't have time to learn everything.

Nathan Gwilliam: And I love it. And it feels like everybody's in the middle of that journey right now. AI is so new. We're all just figuring this out and you seem to be ahead of everybody else, which makes you an expert in this space.

Jim Sterne: It amazes me because those of us who are in the middle of the bubble the one side are the data scientists who are writing the papers and creating the technology, and there are a couple of podcasts that I listened to, to just keep track of oh, there's a new term I haven't heard before, I better go look it up just so I can stay attuned to the language. And then there are the people who are really good at playing with this stuff and trying new things and saying, oh, did you know it can do this now? And now it can do this thing.

And that's great. I get a dozen newsletters everyday with new stuff. And I am consuming all of this information and I am behind all of these people who are moving at light speed. And then I turnaround and I talk to the C-suite of a large corporation. What are you guys doing with this stuff? The board is asking us for our plan and it's just, it's so it's moving so fast.

We're just, we're unbalanced. We don't know what to do first. So we've got this pressure from on high to come up with a plan. Oh, I see the disconnect. You've got people down in the bowels of the company using these tools. Without any direction, without any training, but they're so powerful. Nobody can resist and headquarters is not offering any support or training or even rules or guidelines. So it's this weird thing of the technology has come across along so quickly and it's so powerful. Everybody's using it and yet they're not using it to the best advantage because there's nobody there to help them.

So I'm embroiled in this stuff on a daily basis. And then I talked to my family and they go, Oh yeah, I tried Chat GPT once. And it just gave me a stupid answer. So I didn't go back to it. How do I explain how to use this new technology to make it more powerful? And then comes the moment where they call me up and go, oh, I asked it for a recipe and that was interesting. I can get that from Google, but then I asked it for their best advice for a dinner party of half of them are vegans and half of them are just vegetarians and what should I cook? And it didn't just give me recipes. It explained the difference and helped me understand. And wow, this stuff's great.

Nathan Gwilliam: You just were able to ask for the right information and the right way to get the really helpful information back.

Jim Sterne: It's, don't ask it for facts. Don't ask it for calculations. Ask it for its opinion. Ask it for its advice. Ask it to review a document and come up with some ideas about how to make it better, but don't download everything from Google analytics, upload it into Chat GPT and say, how many people came to my website last Thursday? It'll make it up. But if you say, what's interesting in here for my CFO. What's interesting here for my CMO, what anomalies should I look into further? And it's great for that. Fascinating.

Nathan Gwilliam: So let's break up the life cycle of a podcast. And when we talk about a podcast here, we're not talking about audio only. We're talking about kind of these next-level podcasts that the top podcasters do where they have audio and video and newsletters and live streams and blogs and social and all of those different components. If someone is a creator and they're starting in their life cycle, so they're getting ready to record their episodes. So the period of time before you record how do I best leverage AI to help me with that phase of my content creation?

Jim Sterne: Step one is research. What are you going to talk about? Let's assume that you want to be topical and newsworthy. In the good old days, we set up Google alerts. Tell me when there's something interesting out there about the subject.

As I say, a dozen newsletters a day, I can take those newsletters and upload them and say, give me a summary of the top stories in these newsletters. So if there's a dozen newsletters and five of them cover the same story, ah, that's an important story. Now go find out some interesting facts that I can talk about, and here is my guest go find out interesting things about my guest. Look at their LinkedIn and look at their blog posts. Look what they're posting on social and tell me what they've been talking about in the last week so that I can ask them about it. I want to ask this person questions about this topic. What are some good questions? I want to include something that's entertaining or fun about this topic. Are there any fun things on Twitter lately that that got a lot of likes that I should be aware of.

You can automate this process.

So set up a Google doc and have everybody on your team throw things in there when they find something interesting. And at the end of the week, put it into chat GPT and say, create an outline that will cover these topics in some kind of logical sequence. So that we're starting here and moving to there where we start talking about the hardware and then the software and then the implementation and then the results and find stories that fit each one of those.

So Putting together an outline, I don't want an actual script because Chat GPT output sounds like chat GPT wrote it and you can tell, so just give me the outline and now I've got something to work on to work with putting together an outline, I don't want an actual script because Chat GPT output sounds like chat GPT wrote it and you can tell, so just give me the outline and now I've got something to work on to work with. When I introduce my guest, when we have topics to discuss, and now I'm ready to hit the record.

Nathan Gwilliam: Yeah. And even finding the guests. If I want a podcasting AI expert, we can have Chat GPT help us identify who the best people are. Just think how much that can save our time as we, we have them research the guests, find the guests, find the questions, find the bios.

And not just save us time, but improve the quality as it can help us identify things to even talk about that maybe we hadn't even thought of.

Okay, let's go to the next period of time from, for the period of time that you're recording and maybe editing your episode. How can AI help?

Jim Sterne: Recording if you want to have AI participate, if you will so Zoom has a new AI chat bot that will listen along and you can ask it questions.

I'm shy of using that because when I'm recording, I want to be in the conversation, not running over to the keyboard and distracted by things. But it also will do a level, of course, transcription. That's automagic these days. That's been around for quite a while, but then summarization. It's intended to be a meeting summary.

At this point, these people were talking about this topic and then at the end, here's the to do list. Here's the next steps. And it's really interestingly good at that. It's not great. It's 85 to 90%, which is fine. But it can't be taken at face value. We know this about AI.

Nathan Gwilliam: Still have to have a human review it and edit it and make it publishable.

Jim Sterne: Absolutely. But the actual recording part for me is the least amount of automation. Because it's all humans. It's a conversation.

Nathan Gwilliam: Okay so then talk to me about editing. There's some cool things in editing that AI can do now where you can get into the transcript and see the text and you can edit the text and then it can automatically edit the videos and the audios for you.

I'm even hearing about somewhere you can add text in there. If I introduced you wrong and said you were. You're part of the, you'd produced a summit and I said, you'd produced a challenge course or something. I can go and change the word in the text and then it can create my voice and re put it into the video and audio. Even with my mouth saying the words.

Jim Sterne: Yes, and so not only can you edit what is said and then it will replace that will overdub it in your voice. It will change how your lips are moving to match those words. And now it can also translate into, I don't know how many different languages. So you can produce your podcast in multiple languages that you choose, and it will take this conversation and have both of us with the same video, but move, changing the mouth to match the words in the different language, the same emotional content. In German and Spanish and Hindi etc. So you're now an international speaker.

Nathan Gwilliam: Can you imagine how just a game changer that is? Right now, I bet 99 percent of podcasts are only done in their native language, but that you take these top podcasters with all this information and translate it into 140 different languages. And all of a sudden people in Brazil that don't have access to this kind of information being published in another country, another language do.

Jim Sterne: My marketing analytics summit is coming up in the middle of November in Berlin. So about three months ago, I recorded a, what I thought was humorous.

I took German classes in high school and my grandparents grew up in Germany. So I've got really good pronunciation. Grammar, yeah, that's a problem. But so I created a script and recorded it in German. I don't really speak German, but you should come to this conference, even though I'm going to give my presentation in English and I did it all in German.

And as of two weeks ago, I didn't need to do that. I could just hit a button.

Nathan Gwilliam: Okay, talk to me about eye contact. One of my, the biggest things that I get criticized for when I podcast is I like to look right at you instead of looking at the camera, or I have to look at my notes to get the next question right.

And and I'm not looking at the camera and it's distracting to do it that way. And and people are sometimes not as kind to me as I would like them to be because I'm not looking at the camera like I should be. There's now AI technology. That, that solves that. You want to talk a little bit about eye contact?

Jim Sterne: So this is something that's, there was a moment in time, a historic let's call it two years where we got used to the fact that we're on Zoom. And if there's five of us you can tell because my eyes are stable that I am looking at Nathan and then somebody else talks and then I'm listening to them.

And even though I'm not looking at the camera, you can tell I'm looking off, I'm looking at somebody else. And then somebody else speaks and oh, okay. And it's like being at the table, but what I'm, but I shrink my window down and I put your face right under the camera. I am not looking into the camera, which I am now. No, I'm looking at Nathan. We got used to that.

Now comes AI where, yeah, I can read my script. I can look at myself and the video will make it look like my eyes are always looking at the camera. And then we'll forget that it was ever like this and like that, and we will expect people to always talk into the camera from now on.

Nathan Gwilliam: That's a really hard thing to do in real life. And I'm grateful for AI to help fix that weakness of mine.

Jim Sterne: It's interesting. I think the next level, haven't seen it done yet. But if we have five people and they're across the screen that AI is going to figure out that when I'm talking to person number three and looking over there.

You will see me looking in the direction of that person on your screen so that if I say wait a minute, I want to ask, you can tell that I'm going to go ask Susan or Henry or somebody, because that's the thing about real life. In the middle of you talking, I can go and you know that I am now going to address Crystal over here.

Nathan Gwilliam: So sometimes we may need to turn the eye contact off. Is that what you're saying?

Jim Sterne: I don't think it's turn it off. I think it's going to get smarter.

Nathan Gwilliam: Yeah, smart enough to do that on its own. That's, true. Okay, let's talk about the phase of creating the content, publishing the content and syndicating the content.

What are some of the coolest things AI can do to help us in that phase?

Jim Sterne: So here's direct experience. I did some consulting for an Austin firm that has launched. The ability to create advertising for Google, Facebook, and Instagram automatically so here's my landing page. Here's a description of the audience that I'm interested in, and it will take the images and the text from the landing page and create an ad in the proper format for Facebook and Instagram and Google and publish.

And now I'm using those platforms to do the bidding and how much do I want to spend per day and that sort of thing. But this is automatically from just a product description page, creating advertising, now apply that to podcast. I want to be able to push a button and say, okay here's the audio version that goes on Spotify.

Here's the version that goes on Apple. Here's the one that goes up on SoundCloud and here's the video that goes up on YouTube. And. What's the one I saw Opus, I don't, I want to say Opus dot clip, but I think it's Opus dot pro or something.

Nathan Gwilliam: Yeah I have it written down here or I had it written down on a notepad. I think it's Opus dot US maybe.

Jim Sterne: Where you take your hour long video or however long it is and it will chop it up into little bite sized snacks for social. It's probably 75 percent there, but it will add in the subtitles and it will make it short enough that each one is a little snack of video that will then draw people into the whole show.

And here's the proper version for YouTube and here's the proper version for TikTok and etc. So that's the point where I'm taking a step back and saying, yes, I could learn how to do all that stuff, but that's not my primary business. My business is doing the recording and I want other people to use those technologies on my behalf to efficiently and effectively and inexpensively make all that magic happen for me.

Nathan Gwilliam: It is coming along so quickly. What an amazing opportunity for business. Some of these now seem so simple. One of the AI features, the very first AI feature we put in to PodUp allowed you to start with a video or an audio file or a transcript, answer a few questions, click a button, and it writes your blog post for you.

Nowadays that seems like that's not magical anymore. That seems so basic. But if you think about it that's so amazing. That was probably the most time-intensive part. Of the whole content production piece was writing the blog post.

Jim Sterne: Let me give you an analogy because I'm of an age where I saw the advent of computers being used in business.

So I think this is the difference between learning how to write a basic program to do a calculation versus using a spreadsheet. I don't need to know code. I don't need to plow through trying to create something that works and to see if my calculation was okay. I've got an Excel spreadsheet. I just pop in the numbers of the relationship and it gives me the answers.

And that's normal now. We've got a new normal.

Nathan Gwilliam: So it's changed the bar, the basic standard that's there. Okay so then let's go to the marketing phase. How can I use AI to promote and grow and market my content, my show?

Jim Sterne: The show gets chopped up into bite-sized bits. Tweets are written. LinkedIn posts are created. Scheduling can happen. This is, what formats do you want and what outlets are you aimed at and let's just automate it all. Let's just say hit the button and the word spreads. And then you get the notices back that, hey, over on this platform, this particular version of the short video is getting a lot of attention.

You should go participate in that conversation. Now, do I expect that I can have my own agent participate in that conversation on my behalf? Eventually I will have to train it carefully. My tone, my frequency what I do want it and don't want it to talk about. So if I do a blog post or a podcast about a particular topic and somebody makes a comment that starts to be political, I don't want to participate in that.

I, that's, I am not using that format and that outlet for personal opinions about politics. That's not what I do. I do business. So I will be able eventually to train my agent to act on my behalf. And as long as I am transparent about it, it's the Jim GPT is responding to you, not the Jimster.

Nathan Gwilliam: Yeah we have a tool in PodUp that allows you to bring together all of the engagement that people are doing on all of your content from all time on all platforms. At least the ones that allow us to do a APIs and put some in one inbox. So you can easily see what, maybe imagine me doing my show and I was publishing everyday and let's say I was syndicating to 20 different channels.

And then someone responds to something that I posted two months ago, right? I never would have seen that. That would have slipped through the cracks. So AI bringing to the top of the surface of these are the comments on all of your different platforms that you need to go. Engage with right now.

Jim Sterne: And that then becomes training data for the agent that can host on my behalf. Typically, when somebody has made this kind of comment, Jim Sterne has replied in this way. Therefore agent follow suit.

Nathan Gwilliam: That's right. I love how it can learn. It doesn't have to do everything from scratch, but it can follow.

You were talking a minute ago about tone and I love that tone. You can set the tone. Sometimes you want to be business professional. Sometimes you want to just be fun and personable, right? And I love how it not, it doesn't just give you raw data, but it allows you to infuse personality into that as well.

Jim Sterne: So this is one of the joys of a large language model is that I can tell it what tone I want it to respond in. And ChatGPT plus the $20 a month version has something called. Custom instructions where you can write in two parts.

Part one is who am I? So I gave it a custom instruction saying. I am Jim Stern. Here's my LinkedIn profile. Here are the books I've written. Here are the presentations that I've given. So this is what I know. So when I ask you a question about these subjects, understand that I actually know stuff about this. If I'm asking about theoretical physics, explain it to me like I'm five.

And then the second part is how I want you to respond. I want you to respond business professional. And then I, for a while I said, respond like a children's birthday party clown, respond like a stand-up comedian and it's painful, but it's interesting. And then no, let's go back to business professional.

And then, oh I've taken a little further. Anytime I give you a prompt, I want you to improve the prompt, make suggestions about how I might further improve the prompt and ask me three questions that I can answer. In order to improve the prompt and then show me a new version of it and ask for my opinion and we will carry on this conversation until I give you the word done and then execute.

So this is now prompt engineering. I just ask it to engineer it for me.

Nathan Gwilliam: Okay, let's go to the last phase, the monetization phase. How can we leverage AI to make more money from our content and our shows?

Jim Sterne: The first place my mind goes is analytics. Which of my podcasts, what topics or what guests or what how emphatic I'm being or gentle I'm being or professional I'm being. Which of my podcasts is getting the most attention, which ones get the most response, which ones get shared the most.

And this is market research. This is telling me what the audience wants. And as I grow my audience, I suddenly hit the point where now I can have sponsors. And of course there's the natural call to action at the end of every podcast where you say, hey, if you enjoyed this, go, and follow and, but also here are the things that you can acquire from, this is the services I offer, here's where to go to buy from me.

And let's do the analytics there, which of the shows are getting the most people to click and follow and to purchase.

And now I have this feedback loop of, oh, machine learning that says, hey, we recommend you do more shows like this and this for this audience and more for that audience. And when you want to sell more of this product, have your shows more in that style. It is market research closed loop.

Nathan Gwilliam: Yeah, I love it. That would definitely help increase monetization.

A couple other ideas that come to my mind. Tell me what you think about these is the creation of digital products. Let's say that I wanted to sell an ebook, right? I could very easily go in to to AI and use AI to, to create an ebook on a topic. Or at least create a first draft that then I need to read through and make my own.

Another one that, that's probably a bigger revenue stream would be creating a course. PodUp is just about to rollout an AI that will allow you to create your course. And you give it the outline and then it goes in and you still have to record the video. It can't do the video for you, but.

Jim Sterne: Oh, it will. That will happen. This is the thing. it's like being in 1994 what can you do on a website? What can't you do? We didn't know to call it web 2. 0. We didn't know to call it e-commerce, but it was obvious that was going to happen. And I was one of those people that said sure you can sell books. But you can't sell shoes. People have to try them on. You can't sell cars. People need to do a test drive. Like how wrong was I? But if you can imagine, and I can, right now, you can have your voice cloned.

We're already seeing that happen. You can create an avatar that will take a script, apply your tone, match your movements with an avatar. And guess what? You just automated a whole show. What topic? And so when I've talked about machine learning and again, my book, Artificial Intelligence and Marketing, I wrote six years ago, it's ancient history. But even then, oh my God, it's going to take my job.

No, the human is still needed for three reasons. What problem are we trying to solve? So what topic do we want to discuss? What guests do we want to have? How, what format do we want to have? Number two, what data might be useful? Here's all of the posts that I've made, here's all the articles I've written, here's all the conversations I've had on LinkedIn, and then does the output, does it pass the smell test? Automate as much as you can, but absolutely, positively review every word before it goes out the door. Now, if you choose the topic or the problem to solve, you choose what data is useful and you review everything. Everything else can be automated.

Nathan Gwilliam: We are definitely entering into a new era of operating a business and creating content.

All right, Jim. If our audience has enjoyed this interview, like I have, they want to learn more about you and your services. What are the best ways for them to do that?

Jim Sterne: Thank you for asking. Easiest way to find me is targeting. com T A R G E T I N G. That is my personal professional location from which you can find me. And by all means, connect on LinkedIn.

Nathan Gwilliam: And thank you for sharing your time and wisdom with us today.

Jim Sterne: My pleasure. Thank you.

Subscribe now to the free Podcasting Secrets newsletter, and we send you our Ultimate Podcast Monetization guide at no charge!