Over the last six years, artificial intelligence has found its way into several aspects of creative life. With ChatGPT, you can request a story, a song – even a blog like this – in seconds. Sorry Chat, but I like using my brain to write blogs.
One of the growing epidemics today is AI-generated podcasts. They’re popping up a lot on free platforms like Spreaker. I spent half an hour listening to 2 of these podcasts with the intention of uncovering whether or not AI could approach a topic I like in a way that’s engaging. Spoiler: it couldn’t.
Podcast #1: Classic Short Stories
I’ll admit I liked that this podcast told each story in its full length. I mean, the episodes can be thirty minutes to an an hour long!
However, I didn’t enjoy the lack of emotion and emphasis in the voice while it read. Sure, the voice is somewhat believable but there is a substantial absence of inflection.
Thankfully, it isn’t Gemini-esque.
(Image credit: Polina Tankilevitch on Pexels)
When it comes to pacing and flow, the voice is a bit too fast for my liking. I found it quite hard to keep up at points. And, I never heard a single background element. The volume levels were standard, but quite hollow and machine-like.
Ultimately, a human producer could have added a light music bed that corresponded with the story being read, as well as some SFX where applicable (bells ringing, wind howling, etc.) At the end of the day, it needs a human reader who is expressive and can provide more emotion when reading lines that require things like excitement, sadness, fear – anything!
Podcast #2: Indie Music AI
As someone who loves music history, this podcast made me really mad. The only thing I can give it credit for is having different content to cover each episode (if you can call it content).
For starters, I didn’t like the long convoluted intro. The episode I listened to was about Prince and he wasn’t mentioned until 2 minutes in. What the heck?
(Image credit: Rahul Pandit on Pexels)
Both narrators, who I’ll lovingly refer to as Mr. and Mrs. Gemini, did not seem knowledgeable about Prince himself. They focused way too much on saying big words about the parts of a song. They also lied about Stevie Nicks “plagiarizing” Prince at around the 20 minute mark. There’s no credibility in that…
The biggest takeaway is that neither of the voices are believable. If anything, they are a dead indicator of AI use. There was zero inflection. I mean zero! They always sounded like they wanted to interrupt one another with a bunch of nonsense. No music or SFX either.
Like the first podcast, a human producer could have recruited two humans with a real background in musicology that can establish who Prince was, and organize each song that is being referenced in order of release date. Being a conversational podcast, a well-produced intro would have been nice at the beginning.
Conclusion
Life is far too short to ask artificial intelligence to spit out a piece of media. Get your hands dirty. Both of the podcasts were sloppy from a production standpoint, and at times, they made up information to sound smart. It’s right for us to be scared about losing a human touch – but AI is a long ways away from recreating the same thoughts and feelings we live with every day.
Instead of giving these podcasts any more clicks than they already have, consider checking out a human-made podcast like this one by a NAIT grad.
Happy listening ~
