16 September 2026
Heard In AI

Why an AI-written episode of his own show made this host bet on human company

On The Diary of a CEO, the host describes a test he ran a couple of years ago: an episode of a founder-history show in which AI wrote the script and synthesized his voice, labelled as AI at the top. He says 40 to 50% of the audience reached the end of the hour. His conclusion is not that podcasting ends, but that the purely informational part of it is substitutable — and that coffee shops, workout classes, concerts and dinners gain a premium because people are there.

A briefing reports one development when it happens. We correct or clarify it later; a new development gets a new briefing. How our formats work

A couple of years ago, the host of The Diary of a CEO ran an experiment on another show of his own: a series in which he read out the stories of founders through history. For one episode, about Steve Jobs, he stepped out entirely. AI wrote the script. AI synthesized his voice — that is, generated speech that sounds like him. "I had no involvement," he says. A line at the top of the episode told listeners it was an AI podcast.

He says 40 to 50% of the audience got to the end of the hour.

He had framed the test as a comparison: could a synthesized version of his voice hold listeners for the same average view duration as he did? In the conversation he gives the AI episode's completion figure rather than a side-by-side number for his own episodes. What he takes from it is a split inside his own job. "There is a lot of informational part of what me and you do, where we're sharing just in pure play information that will be taken away," he tells his guest, the investor David Friedberg. The podcast game, he expects, will be "significantly disrupted in some part" — the part where the value is the information itself, delivered in a voice.

The other half of the job

Friedberg's answer to an earlier question about creators and art had pointed at work that cannot be delivered down a wire: his example was yoga teachers. The host picked it up and pushed on what a yoga class actually sells.

"Yoga teachers aren't telling you yoga," he says. "It's also community and gathering in real life and being there amongst other people." Instruction is information; a room full of people is not. On that reading, what an AI voice can copy is the smaller half.

So he expects the in-real-life job world to "explode" and to become "extremely valuable" — and he says this reverses a view he used to hold.

The $40 latte

"I always had this view like, oh, coffee shops will be automated," he says. "But I actually think the opposite is probably true. I think there will be a premium on coffee shops."

His illustration of that premium is deliberately extreme: "I'll pay $40 for a latte because that human is making the latte and I get to go be with other humans." The same logic runs through the rest of his list — a workout class with humans, a concert, a dinner. The figure is his way of saying what he would be willing to pay, not a price he has seen.

From there he sketches a forecast, hedged with "maybe" at each step. The live, experiential part of the economy that is treated as luxury today could become more core: "Maybe we spend 30, 40% of our time, you know, in these sorts of spaces." In that picture, automation generates the income that supports people in whatever work they want to pursue, and the time released from work goes into paying a premium in what he calls the live space economy. The argument is about what people would choose to buy if information becomes cheap — one show's test and one person's spending instinct, not a measurement of the wider economy.

A nurse in a promotional video

Friedberg offers a different route to a similar conclusion, and he is careful about where it comes from. Palantir, the data-analysis company, put out a promotional video he thought was good, in which someone at a hospital was interviewed. He first called her an administrator and then corrected himself: she was a nurse.

As he recalls it, she said she normally spends most of her time filling out paperwork on a computer, and that the software now takes care of that for her. "So now I can spend more time doing rounds face to face with patients." Friedberg's reading: "It's a better job and it's better for her," and a "poignant story about kind of what AI actually does on the ground."

He does not present it as proof. "There's a lot of these sorts of examples," he says. "We could counter argue back and forth about them all day long and probably both be wrong. But this is what I would kind of observe on the ground and what's actually happening." The account reaching listeners has passed through a company's marketing video and then through his memory of it, wrong job title and all.

The host takes the two examples as the point they can share. "I think we agree on the fact that jobs are going to get more human in every sense of the word," he says — the yoga teacher whose class is a gathering, and the nurse who gets her rounds back. It is a tidy pairing for a man whose own voice, in one disclosed hour about Steve Jobs, kept 40 to 50% of an audience listening to the end without him in the room.

Share this article

From the conversation

Podcast episodes

Article history

Updates to this article

Tags

Beyond the copilot: career advice from a panel that disagrees about jobs

On Moonshots, Dave describes a hire in his early twenties who runs his agents entirely by voice, and argues the goal is to manage swarms rather than lean on a single copilot. The panel's optimistic jobs roundup runs straight into Emad Mostaque's warning that today's hiring is "the turkey before Thanksgiving" and Alex's view that every profession, trades included, is only a question of sequencing.

6 min read

Aaron Levie welcomes AI-written code. AI-written board decks "kill" him

On Sequoia's Training Data podcast, Box's Aaron Levie explains why generated code feels acceptable while a generated board deck does not: a presentation is still read as evidence of what its author knows and can execute. He admits the double standard — he uses AI for his own brainstorms and decisions — and describes reading posts twice, once for the substance and once to guess who wrote them.

5 min read

Anthropic's fastest growth scenario grows the pie and shrinks labor's slice

On Moonshots, the panel read out Anthropic's most extreme economic scenario: roughly 15% annual GDP growth by 2030 alongside 17.9% unemployment among cognitive workers and labor's share of income falling from about 60% to 45%. Two investors called the growth number a lowball. Emad Mostaque said the model is missing the thing that breaks it — aggregate demand — and the panel fell into an argument about dividends, ownership and who pays the displaced.

7 min read

The weak point in Friedberg's AI jobs optimism: workers have to choose to move

On The Diary of a CEO, investor David Friedberg argued that AI grows companies rather than shrinking payrolls, using a painter commanding five robots as his image. Pressed on entry-level hiring, Klarna's staffing numbers and a Stanford payroll study, he named the assumption he thinks could be wrong: that displaced people will see an opportunity and take it.

10 min read