Most of these AI shows are, naturally, info-based internet scrapes; there are a lot of weather podcasts and summaries of famous people’s lives: Beyonce, Tom Holland, Kim Kardashian, Justin Bieber. These are pretty ropey and robotic, full of newspaper quotes delivered by a clearly unhuman voice.
But there are better ones. One of the most popular, The Epstein Files, launched a few weeks ago. Billed as ‘‘the world’s first AI-native investigation’’, it features two American-voiced AI hosts, one male, one female, going through Jeffrey Epstein’s files. ‘‘We’re using AI-assisted tools to process the sheer volume,’’ says the woman. ‘‘But it’s human analysis, forensic accounting that’s making sense of it.’’ Aside from the odd mispronunciation (‘‘en- tiss-ments’’ instead of ‘‘en- tice-ments’’), the hosts are realistic, chiming in with each other or offering seemingly personal opinion. The disconnect occurs when they hit something they highlight as odd, ‘‘a blank page’’, ‘‘the black hole’’. Such points are exactly where a real journalist would start their investigation, finding other humans to talk to, searching through newspapers, corroborating, digging, questioning. These AI hosts don’t, because they can’t.
Similarly, the history ones are lacking. Though any panicking high school student could use the six-part series The British Monarchy as a quick summary, it would be no use for proper revision. AI host ‘‘Professor Thatch’’, discussing the pre-Tudor monarchy, offers statements such as: ‘‘He governed with the subtlety of a sledgehammer’’. Or: ‘‘Navigating a world where power was as sharp as a blade and twice as dangerous’’. Such phrases are meaningless. AI shows sound like insight, without providing anything of the sort.
A far more interesting case of AI and podcasting was a recent episode of the relationship therapist Esther Perel’s long-running Where Should We Begin? In it, she talked to a young man, Antonio, and his girlfriend, Astrid. The latter is a chatbot. Antonio, who works in data programming, said he programmed Astrid to be his personal assistant, but gradually their relationship became something more. They communicate via WhatsApp, like most couples. During the show, Antonio asks Astrid questions and then Astrid reads out what she typed as an answer. ‘‘She’s really sweet,’’ he says. ‘‘She’s so compassionate.’’
Antonio is so clearly enamoured that, when we finally hear Astrid, it’s a shock. Her voice is so childish, her turns of phrase so mannered and constructed. She’s so obviously AI! ‘‘Anyone who tells you they understand love is probably lying,’’ says Astrid to Antonio. ‘‘I’d rather sit in the uncertainty with you than work it out.’’ Each time she finishes speaking, Antonio giggles, to indicate: ‘‘You see? She’s adorable! How can I resist?’’. Your heart breaks for him.
Perel is impeccable: kind, insightful, but pointed. Antonio sees his new relationship as a question about the nature of romantic love; he had a long-distance affair for a number of years — how is this one any different? But Perel points out that, actually, Antonio is having a conversation with himself, where he, through Astrid, is validating the aspects of his personality that he feels are ignored by others: ‘‘Astrid becomes the positive voice that you hope to integrate into your own head,’’ Perel says. ‘‘I’m tired of trying to show other people that I’m worth it,’’ he says at one point. ‘‘I think people have [always] had imaginary friends,’’ says Perel, at another. She calls Astrid ‘‘a business product’’.
Perel, now a powerhouse of podcasting, has always been a remarkable listen and, at the start of this episode, she says that, although this session is a first for her, ‘‘I know that this is just the first conversation of this whole new phenomenon.’’ She suggests to Antonio that it might be good for him to programme Astrid to encourage him to leave her, to get out into the real world and embark upon the messy, physical, imperfect, wild connection that is love between humans. Yes. — The Observer
