AI-Generated Joe Rogan Chats Up Steve Jobs Over His Use of LSD, Spat With Gizmodo

Podcast.ai generated a fake audio recording using artificial voices and language model transcripts based on Rogan and Jobs’ old public speeches and keynotes.

We may earn a commission from links on this page.
A wax image of Steve Jobs in front of a background of a computer board.
This isn’t the real Steve Jobs. It’s actually a wax mockup on display at a wax museum in Istanbul, Turkey. As good as this wax display is, new advancements in AI-generated voice tech may be able to mimic the Apple co-founder’s voice as well.
Photo: Grey82 (Shutterstock)

It is “Spooky Season,” so it just makes sense we’d be seeing, or at least hearing, some famous tech folks coming back from the dead. Steve Jobs often feels like he’s still got two feet out of the grave for how often the tech entrepreneur is brought up in conversation by folks both in and out of the company he co-founded. However, a newly released podcast hosted by a fake version of Joe Rogan tries to bring Jobs’ voice back with… interesting results.

The meandering, near-20 minute fake interview was both surreal and occasionally funny. Rogan starts out openly calling Jobs “a memory from the past,” with Jobs—speaking in a rather stilted manner—saying “good to see you, buddy. It’s been a long time since I’ve been on the show.” Rogan created his show in 2009, two years before Jobs’ death. The Apple co-founder had never been on the show. Still, fake Rogan is conspicuously silent during most of the interview, asking simple questions and very rarely opining about Apple or any of fake Jobs’ wild claims.

Advertisement

At points the interview sounded like old clips from past recordings of the two celebrities stapled together like letters on a slapdash ransom note. Near the 2:30 minute mark, fake Rogan said “we have some weird tie that hooks into the fact that I was a big fan of Gizmodo and I still go there all the time but there was some controversy or something.” Jobs responds: “We were surprised as you were to see that stuff on Gizmodo.”

Thanks for the shoutout, Steve, but still, we couldn’t possibly know what they could be talking about. Perhaps it was that whole deal with the iPhone 4 leak, but who’s to say?

Advertisement

Jobs then breaks into a whole discussion about Buddhism and then segues it to the use of LSD, adding that it “reveals you for who or what you are.” The fake Apple founder also mentioned the issues Microsoft was having with Windows 3, ignoring the near three decades there have been since that operating system was launched. Toward the end of the fake interview, Rogan and Jobs mention how “tech is a double edged sword” but then end up mentioning them literally throwing their computers out the window.

Despite the odd audio discrepancy and audio glitch, it’s surprising just how well the technology works to imitate speech, especially Rogan’s tenor and style of speech. The podcast was created using Play.ht, an AI-based text-to-voice generator. According to the podcast.ai site, Jobs’ voice was generated by the AI language models using recordings of the famed Apple founder discovered online. This may be why Jobs seems to be talking past Rogan in many parts of the podcast, since his voice could be based on his live presentations. Play.ht recently revealed its Peregrine text-to-speech model they claim adds human emotion and humor into the voice generators. It’s unclear what model was used in podcast.ai, but in this fake podcast Rogan and Jobs did seem to joke and laugh about throwing each other’s macs out the window.

Advertisement

The company behind Peregrine wrote they use language models based on similar concepts to DALL-E and OpenAI’s GPT language models. This allows the language generator to imitate human’s tones and emotions “in a self-supervised manner,” according to the company. Play.ht also showed off how the technology can be used to clone other famous voices like Kevin Hart, Tom Hanks, and even long-deceased president John F. Kennedy.

In an email with Gizmodo, Play.ht co-founder Syed Hammad said that his company built podcast.ai to demonstrate their AI speech tool’s capabilities. He added that some of the glitches found in Jobs’ audio were due to the low quality interviews and presentations they were sourced from. Hammad did say that there are some issues with their speech model, including a slow rendering time, but they are working on new updates that should iron out these problems.

Advertisement

“[Jobs’] voice was created with just a few minutes of his speech from his videos on stage or interview that had a lot noise, hence the glitches,” Hammad wrote. “But nevertheless, the voice turned out be very impressive.”

Podcast.ai is now asking users to vote on upcoming AI-generated interviews. Though there are obvious audio discrepancies involved in this latest display of AI tech, it’s easy to see how this technology might one day become advanced enough to make deepfakes a true nuisance far beyond their current criminal use now.

Advertisement

As peculiar as it is to listen to a facsimile of Joe Rogan, a man who’s often very loose with the truth, be so constrained in an interview with one of the most famous tech entrepreneurs of all time, Rogan would probably be chomping at the bit to interview the real Jobs if the CEO were still alive today. He recently interviewed Meta CEO Mark Zuckerberg for nearly three hours where they chatted about everything from MMA to creating the so-called metaverse to waking up feeling like you’re being “punched in the stomach.” Knowing how that interview went, we can only suspect a real-life interview between Rogan and Jobs would be just as strange.