The AI-generated podcast caught the attention of Rogan himself.
On a recent podcast episode, you can hear Joe Rogan ask OpenAI founder Sam Altman whether society should be concerned about fake AI content. “How will we be able to tell what’s real and what’s not?” It was a particularly fitting question for the episode — since the entire hourlong conversation was generated by AI.
Depending on whom you ask, AI is either an existential threat to every creative industry, a groundbreaking technology that will advance human civilization, or a kind of fun, dumb toy that is ultimately a pale imitation of the real deal. The creator of The Joe Rogan AI Experience, a sort of AI-generated fanfic YouTube podcast that depicts imaginary conversations between Rogan and a new guest every episode, kind of believes all three. Or at least, he brings up all three over the course of a conversation that Hot Pod had with him earlier this month.
Hugo, who asked us to only use his first name to protect his identity, is a creative director for a VFX studio and ad agency in Sydney, Australia. Hugo is quick to admit that his YouTube podcast is just for fun and not intended as any real threat to the real Rogan or actual flesh and blood podcasters.
“This is purely fiction — just for fun — so don’t go around making your little TikToks and make the world believe I said things I never said,” the AI clone of Joe Rogan AI warns in the first episode of the podcast, which Hugo published a month ago. In it, the fake Rogan and an equally fake Altman have a ChatGPT-scripted conversation. The resulting podcast was eerie enough that Rogan addressed it himself.
“This is going to get very slippery, kids,” Rogan — the real one — wrote in a tweet.
The idea for The Joe Rogan AI Experience came to Hugo back in February, when social media discourse around ChatGPT seemed to be at an all-time high. He had been regularly following AI advancements over the past year and watched “a couple of YouTube videos about voice cloning” when the idea popped into his head.
“What if I tried to generate a whole The Joe Rogan Experience podcast with ChatGPT and then managed to use one of those voice cloning platforms?” he thought.
After some Google searching, he found a number of AI voice software companies with demo samples of Joe Rogan’s voice. Unfortunately, a lot of them weren’t that great or available for public use. Finally, after a couple of weeks of searching, Hugo found a text-to-speech platform that can clone any voice based on snippets of audio. He used that platform for the voice of Rogan and other guests on the podcast, which have also included Donald Trump and Andrew Tate.
Supporters of AI say it will be faster and cheaper to produce than human-made work. But the work of prompting a usable script on ChatGPT, throwing it into a text-to-speech program, and editing audio — all takes hours of human work.
“The first […episode] took me a week, the second one took me a day and a half, and the last one took me more than three weeks,” wrote Hugo in an email. “It’s very dependent on many different aspects but the main one is my own critical opinion of the episode. If it’s not good enough in my eyes, I will spend a lot of time making it right.”
Hugo spent a long time gathering sample audio and perfecting his AI-generated voices. While creating the first voice clone of Rogan was easy, Hugo spent hours refining it to perfection in later episodes. Even then, there would be moments when his speakers sounded robotic or AI-like. Editing the script to create something that sounds like a normal human conversation also took work.
“[…ChatGPT] is really not that fantastic yet, really, with replicating the speech patterns of someone,” said Hugo. Even though someone like Rogan has countless hours of voice recordings available on the internet — ChatGPT only trains on written information. Even feeding it transcripts of interviews, according to Hugo, isn’t foolproof.
“When you really listen to a conversation between two people, you’ll see that they talk over each other a lot. They answer each other — with very brief answers sometimes — and it’s just a constant back and forth,” he said.
Making an AI-generated podcast of Joe Rogan no doubt takes Hugo longer than the actual Joe Rogan just doing it himself. So why make an AI-generated podcast of one of the most popular podcasters at all?
“I wanted to create a version of the Joe Rogan podcast that would never happen, or hasn’t happened yet,” Hugo said.
Which is true — Rogan has yet to interview Altman or any of the other guests that are depicted in The AI Experience. Hugo believes it’s the show’s main draw — to stage hypothetical conversations between Rogan and other people. He eventually wants to try creating episodes that depict dead guests, such as Steve Jobs or Abraham Lincoln.
I spent an hour or so listening to segments of The Joe Rogan AI Experience. While the quality of the voice cloning is uncanny and Hugo’s editing is expert-level, the actual content leaves a lot to be desired. The dialogue is wordy and bland in a way we now associate with AI-generated text.
“To be honest personally, I think we’re not anywhere near close to replacing real podcasts or real conversation yet, but I can understand the fear of people. Because I don’t think it’s super close — I mean — it is close. But if you’ve watched a lot of Joe Rogan’s podcast, you can feel straight away that […The AI Experience] is not real and lacks a lot of emotion and correct intonations,” said Hugo.
The first episode of The Joe Rogan AI Experience wracked up close to half a million views (likely due to Rogan tweeting it), but viewership has dropped in subsequent episodes. That’s in contrast to AI remixes of songs by Drake and other artists, which have quickly gone viral, but the jury is still out on whether humans will actually prefer AI recreations to the real thing.
There are other reasons to be concerned about the rise of AI in podcasting. Digital fakes could make Rogan or any other host say something objectionable. And Hugo says he’s already seeing how AI tools could replace certain jobs in the VFX industry.
But that doesn’t mean they spell the end of live-hosted shows. At one point, I brought up a Vulture interview with Comedy Bang! Bang! host Scott Aukerman, where he notes that podcasts stand out because so much of entertainment today is prescripted. Podcasts are rare in that you can hear people laugh and “be joyful in the moment.”
Hugo doesn’t believe AI can replace that spontaneity.
“You know, the thing when you’re with a friend at night and you’re talking, and an hour later you end up in a completely different place from where you started? And you realize AI can’t really do that. The way the human brain works — it’s very unpredictable.”
