Skip to main content
Three Reasons Podcast Creatives Should Embrace, Not Fear, AI

Three Reasons Podcast Creatives Should Embrace, Not Fear, AI

Listen to the podcast version

By Tom Webster

December 20, 2022

ChatGPT and other new forms of chatbot are already able to credibly write like humans, and it’s getting harder to tell who - or what - writes any given blog post. This week, Tom explains why this is a good thing for podcast creatives, and not the apocalypse.

Public opinion and risk tolerance can shift overnight. Brands and agencies are often concerned about how to approach brand safety and suitability when it comes to podcasts. Many of their decisions are built on assumptions from different times and different mediums. Do they even apply to podcasting?

Safe and Sound, the first ever study of brand safety and suitability from the listener’s perspective, aims to answer that. Check out the latest Sounds Profitable research, available now for download.

Recently, I wrote about brand safety and brand suitability in podcasting, the topics of our latest research project, Safe and Sound. I knew that I needed to come up with a solid distinction between the two concepts, because they are managed very differently. I gave it a couple of stabs, but brain-addled by RSV/Anthrax/The Grippe or whatever else is making everyone sick right now, I was struggling for clear writing.

So, I gave ChatGPT a shot. You’ve seen the term thrown around a bit, or even used it. For those who haven’t, ChatGPT is a new kind of chatbot – much more advanced than you have likely ever encountered. This AI-driven dialogue engine is vastly more powerful than “Julie” who takes your Amtrak reservations (sorry, Jules). No, ChatGPT is designed to do something that almost feels like magic, or actual thinking, or magical thinking. You give it a prompt, like “write me a term paper about the importance of public beheadings in England to the works of Lewis Carroll” (a thing I believe I actually wrote in grad school), and you will get something back that is either wondrously or terrifyingly like real, human writing.

I’ve played around with it before, so I was comfortable with using it for this query: “Define and describe the difference between brand safety and brand suitability in advertising.” What I got back was pretty good – four paragraphs, written at at least a high school level, and factually correct, if a little prosaic. I used none of it – the text, while competent, wasn’t quite Webster-level and certainly didn’t reflect my tone or voice. But, as I said, it was competent, and in one of the paragraphs was a turn of phrase that inspired me, which I distilled down to the phrase I actually wrote: “brand safety is about avoiding negative content, and suitability is about seeking relevant content.”

Now, I was pretty happy with that. I probably would have gotten to something like it over time, or over a few drafts, but ChatGPT definitely got me there quicker. I would never let it write my articles – you’d suss it in a heartbeat. But I am not ashamed in the least to say that it was an excellent writer’s aide to help get me where I wanted to go.

In the Sounds Profitable partners’ Slack channel, I also posted another fun ChatGPT result, just to see how everyone responded:

Me: “Create a true crime podcast about an unsolved murder in rural Maine.”

ChatGPT: “In the small town of Pine Creek, nestled in the remote forests of rural Maine, a terrible crime has gone unsolved for years. On a crisp autumn night in 2012, the town’s beloved high school English teacher, Ms. Emily Davis, was found brutally murdered in her home.

Despite an exhaustive investigation, the police have been unable to identify any suspects or uncover a motive for the crime. The case has grown cold, and for many in Pine Creek, the memory of Ms. Davis and the trauma of her unsolved murder have begun to fade.

But for her former students and colleagues, the pain of her loss and the unanswered questions surrounding her death still linger. In this true crime podcast, we will explore the events leading up to Ms. Davis’s murder, the investigation that followed, and the impact her loss has had on the community.

Through interviews with those who knew her, investigative reporting, and analysis of the available evidence, we will attempt to shed new light on this tragic case and bring justice for Ms. Davis.”

Top of the charts, here I come.

How do you feel about that? Amazed? Terrified? Nonplussed? Plussed?

For me, as a writer, I range from plussed to amazed. First of all, the AI would still have to take an incredible leap to offer up a phrase that used “nonplussed” and “plussed” together. I don’t think it’s going to take my job at Sounds Profitable. And there is a lot of distance between the competently-written treatment it spit out, and a compelling true crime podcast with a great script, immersive sound design, suspense, and quality voice acting. But it did make me wonder just how much time AI services like this could save a creative team if they embrace, rather than reject, these tools.

I’ve seen all the fear online about OpenAI, and how if we aren’t careful it will at best take our jobs, and at worst become Skynet. Maybe Skynet began with high-school level, prosaic essays about the scourge of humanity and then simply developed more style on its way to covering Arnold’s exoskeleton with lab-grown skin. Or maybe OpenAI will eventually kill us with a barrage of mind-numbing queries from college students faking their term papers. Dunno. I am not really worried about it taking my job. But it will change how people do their jobs, without question.

In fact, I see three main benefits for creative work (and society as a whole) from AI-driven content, which I have only recently begun thinking about.

The Writer’s Friend

The first, I’ve already talked about – as a writer’s aide. For those who focus solely on ChatGPT as something that takes writers’ jobs away, let me disabuse you of that notion right now. First of all, “writers’ jobs” makes me chuckle a bit. You think podcasters don’t make any money? Anyway.

I have a job/career/passion that stems from being able to write, and it will continue to depend on that ability for a good many years to come. I’m delighted to have some help in the process. I gave you one example already – the safety-vs-suitability definition the AI helped me craft. It didn’t make my content, and it won’t necessarily make your content better. It can make your content process better, and that thrills me in ways that other recent “gifts” from technology like NFTs and crypto have not.

I don’t always agree with Seth Godin, but he always makes me think, and I thought his recent article about ChatGPT really nailed it. One of my favorite lines was this: “It means that creating huge amounts of mediocre material is easier than ever before. You can write a bad Seinfeld script in about six minutes.” I thought a lot about this one sentence. You might even call it a “writer’s prompt.” You know, like the kind ChatGPT also gave me. This is how writing works. This is how it has always worked. This article did not spring forth from my keyboard in one continuous, laminar flow. Writers use many prompts and sources for inspiration, from Seth to Skynet. I even looked up “laminar” (in Webster’s, natch) because I thought “placid” was more relevant to a lake analogy, but “laminar” was the perfect word for a river.

The point is, I got help. And now I have more.

Writing’s Friend

Let’s get back to Seth’s point about “mediocre material” for a moment. The knee-jerk take is that AI will take writers’ jobs. The semi-spicy take is that it will cost the jobs of mediocre writers, but the good writers will always be in demand. I don’t fully agree with either of those. AI won’t be the death of writers, or even of mediocre writers. I’ll have more to say about mediocre writers in a moment, but one thing I can say is that even before AI, we already have a word for people who are employed specifically as writers who are mediocre at writing: unemployed. No one currently creating ‘huge amounts of mediocre material’ or ‘bad Seinfeld scripts’ are at risk of redundancy, because those people aren’t currently employed now, or if they are, it’s never been a long-term prospect, Skynet or no.

Seth didn’t say it would be the end of mediocre writers. It is going to speed up the demise of mediocre writing, however. Like my friend Ann Handley has penned so eloquently, everybody writes. And over the course of a thirty-year career, I can sum up the most important thing I learned about getting ahead in your career in four words. This is advice I gave my son, who is about to embark on his college adventure pursuing the sciences: learn to write gooder.

There aren’t so many mediocre writers gainfully employed as full-time writers. But there are plenty of mediocre writers in other positions, and that is going to hold your career back almost no matter what you do. My first paid job in writing was actually as a lecturer in rhetoric and composition at Penn State. I taught Freshman English, basically, while pursuing my own graduate degree. What I assumed was that I would get papers back that were competent, if uninspiring, and that I would help them become more persuasive. Instead, I left with the overwhelming and depressing feeling that I really couldn’t help many of them. Much of that was down to my own failings as a teacher, but that didn’t change the fact that many of these incoming students at a major University were shockingly bad writers.

My next job was as a tech writer for a consulting firm that employed mostly scientists. And I left that job with the same feeling. I’ll stop going through my C.V. now, but the point has stayed true across every level of my career – good writing is, at best, “uncommon’” in any company.

Today, if I were managing some of the teams I worked with earlier in my career, I would insist that they begin with tools like Grammarly, Hemingway, and yes, ChatGPT. Bad writing is where good ideas go to die. People have been too focused on the “writers will lose their jobs” canard, and not focused enough on the real import: non-writers will do their jobs better.

Which brings me to my last point…

Society’s Friend

The last sentence of Seth’s article was this: “Technology begins by making old work easier, but then it requires that new work be better.” Technology is not a threat. It’s a challenge. And here is the great thing about being human – we respond well to those challenges. Don’t we want new work to be better? I would like to be better at my job. If I am better at my job, I can help more podcasters and creative professionals be better at their jobs. And this in turn makes podcasting a more viable career for more people. That, in a nutshell, is what drives me. We need to be better – I need to be better. This isn’t about the hustle and grind. It’s about what evolutionary biologists call the Red Queen Theory of Evolution, named after the character from Alice in Wonderland. See, I didn’t drop Lewis Carroll at the beginning for nothing!

In Alice, the Red Queen was always running, even though she didn’t actually get anywhere. The running wasn’t pointless: if she stopped, she’d go backwards, and at least staying in place was preferable to losing ground. In evolution, the Red Queen describes the fact that species continually evolve not for advantage, but survival. Humans are no different. Civilizations that collapse do so often because they have failed to embrace the new tools, and clung stubbornly to the old. And these people didn’t just lose their jobs.

Technology like ChatGTP that causes some disruption in the short term ultimately leads to us all being better in the medium term. I think chess is a good example of this, by the way. Chess AI has progressed to the point that it’s impossible to beat for all but a literal handful of humans. So why bother to play? Did chess AI kill chess? Just the opposite: it’s never been stronger. Instead of a Skynet-style battlefield littered with disused chess boards and the corpses of Grandmasters, we have Magnus Carlsen, the strongest player in history, and the holder of one of the most daunting records in the history of competition, period: a 125-game winning streak in a game that, as infuriatingly as soccer, often ends in a draw at the highest levels.

Critics of Carlsen sometimes belittle him by exclaiming that he “just” uses computers to learn the rote, book moves of chess, so that he himself plays like a computer. Um, OK, Einstein – you try it. Chess AI has served as a player’s aide in the same way that ChatGPT serves as a writer’s aide. As a result, Bobby Fischer wouldn’t even crack the top ten today. I’m a mediocre chess player, so I suppose AI helped take my chess job. I apologize to my family.

But I’ll also say this – when I went to college, there was a booming economy for people to take longhand theses and reports and type them for you. The word processor replaced those jobs, but it didn’t replace those people. And, more importantly for me, it allowed me unfettered access to presenting my ideas in a legible and clear way. With my handwriting, if I were born 100 years ago, I’d be working on a farm. Now, I am not saying that I am any great gift to the world as a podcaster, but as a farmer? We go back to civilization collapse.

Even the smallest tools and advances throughout society have made our lives better, from disposable pens, to salted cod, to nails. Imagine what a tool that helps us craft better ideas might do.

Just imagine.

Market Insights with Magellan AI

This week we are highlighting the Gifts industry, which increased the number of ads it ran in November by 4X compared to October. While spend figures are still being finalized for the month of November, we wanted to highlight ad counts as another metric that can help us understand the podcast market.

By comparing ad counts in 2021 and 2022 (YTD) we see ad counts have been fairly consistent year-over-year, with ad counts slightly depressed in Q3′ 22. Despite the activity in Q3 ad counts have appeared to bounce back in November, which is typically when holiday spending starts to ramp up. It will be interesting to see how Gift industry spending compares for November, but this is a good initial sign.

Interested in more insights like this? Sign up to join the 15-minute monthly market update on December 22nd.

Anatomy of an Ad with ThoughtLeaders

Sponsoring brand: Book of the Month

Where we caught the ad: The Lazy Genius

Who else has sponsored this podcast? HelloFresh, ZocDoc, KiwiCo, Olive & June

Where else has this brand appeared? Add to Cart with Kulap Vilaysack & SuChin Pak

Why it works: Kendra, the lazy genius, is all about focusing on the things that matter and making the things that don’t matter as much a little easier. In this case, picking books to read. As Kendra Adachi puts it, “I love Book of the Month because they make it easier than ever to discover the best new books”, allowing her to spend more time focusing on the things that matter – family, work, etc. – and less time worrying about what should be her next read.

About the author

Tom Webster is a Partner at Sounds Profitable, dedicated to setting the course for the future of the audio business. He is a 25-year veteran audio researcher and trusted advisor to the biggest companies in podcasting, and has dedicated his career to the advancement of podcasting for networks and individuals alike. He has been the co-author and driver behind some of audio’s most influential studies, from the Infinite Dial® series to Share of Ear® and the Podcast Consumer Tracker. Webster has led hundreds of audience research projects on six continents, for some of the most listened-to podcasts and syndicated radio shows in the world. He’s done a card trick for Paula Abdul, shared a martini with Tom Jones, and sold vinyl to Christopher Walken.

Now playing on Sounds Profitable Podcasts

Or listen on your preferred platform: