The fake Joe Rogan interviews the fake Steve Jobs in an AI-powered podcast

Enlarge / Which Steve Jobs is the real one? Ars-Technica

A Dubai-based text-to-speech company has released a fictionalized podcast interview between Joe Rogan and Steve Jobs using realistic digitally cloned voices of the two men. It takes place during the "first episode" of an alleged podcast series called "Podcast.ai", created by Play.ht, which sells text-to-speech services.

In the interview, you first hear a replica of Rogan's voice created by voice-cloning technology similar to the one we've already covered on Ars. Deep learning technology has enabled AI models to reproduce distinct voices with a high degree of accuracy, as in the case of Darth Vader in Disney's Obi-Wan Kenobi TV series .

To achieve this effect, someone must first train the AI ​​model on existing samples of the voice that will be cloned. Rogan is a prime target for AI voice training through deep learning models because large amounts of his isolated voice exist on his podcasts. In fact, The Verge covered a PR stunt of an AI company called Dessa synthesizing Rogan in 2019.

Where this example of AI deception gets more interesting is that Play.ht also featured the voice of late Apple CEO Steve Jobs. His voice, sometimes robotically choppy, is reminiscent of his Apple keynotes and All Things Digital interviews from the late 2000s. And Play.ht claims the interview text was also AI-generated, possibly from a large language model (LLM) similar to GPT-3.

"Transcripts are generated with fine-tuned language models," writes Play.ht on the Podcast.ai website. "For example, the Steve Jobs episode was trained on his biography and all of the recordings of him that we could find online so that the AI ​​could bring him back to life with precision."

In keeping with his LLM roots, the 19-minute interview doesn't make much sense. After a while, parts of the fictionalized interview begin to look like conceptual mashups of common Jobs talking points, including aesthetics, groundbreaking products, competitors such as Google, Microsoft, and Adobe, and the triumphs of the original Macintosh.

For example, during part of the interview, the fake Jobs delves into criticisms of Microsoft that are very similar to what the real Jobs said in a famous 1995 interview for Triumph of the Nerds, but it's not a carbon copy - and you can tell the voice is synthesized if you compare the two. "That's the problem I've always had with Microsoft," says Fake Jobs. "In many ways, they're smart people and they've done a good job, but they've never had taste. They've never had aesthetic sense."

It remains to be seen whether it is legal to use the voice likenesses of Jobs or Rogan in this manner, particularly to promote a commercial product. And despite the advertising nature of the podcast, the concept of entirely fictional celebrity podcasts caught our attention. As text-to-speech becomes more widespread and potentially undetectable, we envision a future where media artifacts from all eras are likely to be completely fluid and malleable, shapeable to fit any narrative. In this particular fictional world, Jobs is a big fan of Rogan.

"It's nice to sit in the car and listen to you fuming," he says.