Artificial Intelligence and disinformation comes hand-in-hand. While the tech has been used for political espionage, AI technology has been used in a much more lighthearted, but still problematic, scenario: an AI Joe Rogan podcast.
Listen to the AI Joe Rogan podcast
Released on Podcast.ai, a new AI Joe Rogan podcast was released. Lasting almost twenty minutes, the AI podcast sees Rogan interview deceased Apple CEO Steve Jobs.
Using Play.hi AI, the generated podcast uses digital clones of Joe Rogan and Steve Jobs. Rogan’s speech is based off his infamous podcast, The Joe Rogan Experience. On the other hand, Jobs’ vocals are based on an audiobook documentary.
This shows in the final product. Rogan’s speech sounds natural and fitting for a podcast whereas Jobs sounds theatricalised. Nevertheless, there are still moments where the conversation sounds rather natural.
The transcript of the podcast was generated by OpenAI’s GPT-3 text generator. This text is then fed into the speech generator to create an entirely new podcast.
At times, the fake Jobs echoes sentiments similar to his real-life counterpart. The generated Apple CEO even dogs on Microsoft at one point, saying:
“That's the problem I've always had with Microsoft. In many ways they're smart people and they've done good work, but they've never had any taste. They've never had any aesthetic sense."
Is this an issue?
The creation of AI generated podcast content is not necessarily an issue, but can be an issue in the future. Currently, the technology is used to create proof of concepts such as this Steve Jobs interview.
However, some point in the future, this technology could become problematic. As the technology continues to evolve, it could be used to make celebrities, YouTubers and more say whatever they want to.
As AI is already out in the wild, this future is likely unstoppable. Hopefully, it will not be used for wrongdoing, but that does seem inevitable.