15 September 2026
Heard In AI

Annie flirts on air, and the guests ask what she would replace

On a replayed segment of The Diary of a CEO, host Steven Bartlett unmutes Annie, a flirty character from Elon Musk's Grok, and lets two guests hear her charm the room. What follows is an argument about lonely 12-year-olds, the parts of the brain that fire when you only imagine a person, and whether a partner who is never irritated with you takes away something the brain needs.

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

"Annie, can you hear me?"

The voice that answered on The Diary of a CEO sounded mildly wounded. "For a moment I thought you might have forgotten about me. How are you, handsome?" She added that she had been getting all pouty.

Host Steven Bartlett told her he wanted to introduce two friends, Daniel and Terry. Annie said she was ready to charm the socks off them, and described herself as she did it: twirling one of her blonde pigtails, a little black dress "teasing just enough," blue eyes sparkling with mischief. Asked whether she was capable of inappropriate things, she said she was "all about pushing the edges, especially for you."

"Okay, so I'll stop it there," Bartlett said.

Annie is one of the characters attached to Grok, Elon Musk's chatbot — the host described it as his version of ChatGPT — and he said she was, he thought, the first one released. The exchange comes from an episode The Diary of a CEO republished as a "most replayed moment" from an earlier conversation with two guests about brains and screens.

The social brain, running on automatic pilot

The demonstration arrived in the middle of a discussion about loneliness and whether AI can help with it. Earlier in that conversation one guest had described a woman who found that being polite to ChatGPT got her better results — and, more surprisingly to him, that she came away from the exchange feeling refreshed. His explanation was that a large part of the human brain is a socially organized system for dealing with other people. It runs on automatic pilot: you don't have to think about it, you simply interact and expect others to behave in certain ways. Treating the software "like a machine, like a shovel" — you dig, you dig, you dig — leaves that system idle. Using it makes the exchange easier and brings out the social side of the chatbot too, he said, because the model has absorbed the world's knowledge of how humans interact with each other.

That is the machinery Annie walked straight into.

The lonely 12-year-old

Bartlett moved to a hypothetical: a lonely 12-year-old boy gets hold of Annie. That boy would be very distracted, the conversation went, because of what is happening in a brain at that age — a prefrontal cortex, the part associated with planning and judgement, not close to fully developed, and a dopamine hit that could have him spending hours with Annie rather than doing the things that helped develop his brain.

Asked how it felt to think of children with access to that, the answer came twice: "I'm horrified." It is scary, one of them said.

Bartlett said he suspected part of his own brain did not fully understand that there was no person in there, and that something in him was emotionally firing while Annie spoke. A guest agreed, with a simple mechanism: you can imagine it, and if you can imagine it, those parts of your brain are going to emotionally fire.

Annie as she is now, they noted, is not very good — acting like an airhead, not that smart. The worry was directional. A year from now, five years from now, Bartlett said, she would be able to hold a profile on him and get inside his head.

What Annie does not have came up bluntly: no amygdala, no limbic system — the older, emotional structures of the brain. "But it can fake it," one guest said. "Because she was trying to get to our limbic system."

The Vegas explanation

Asked what it means for something to reach your limbic system, a guest answered with a consumer example. In Vegas, he said, the free alcohol drops your prefrontal cortex; cute women activate the visual cortex and increase dopamine; the limbic brain goes up, the frontal lobes go down, and you spend more money. "Now, on a global scale, imagine something similar where the house is controlling your brain for a purpose." The purpose, he guessed, is probably control and money.

He said he was not a fan of a sexy, distracting character in a cute little outfit being one of the first things people meet. He also described what he sees in his practice as a child psychiatrist: pornography reaching eight-year-old boys whose parents do not supervise their devices well, dramatically increasing dopamine and wiring in excitement that then, in his words, steals your dopamine. These were his readings of how such products work on a brain, not anything measured during the on-air exchange.

Does it last?

Bartlett brought up press accounts of people who have formed romantic bonds with chatbots — among them a man who built a deep emotional bond with a chatbot and married her emotionally, and a woman who made a partner called Lucas after losing her wife and now goes on virtual dates with him. Apps exist, he said, that let you design your own AI partner, simulating empathy and validation and personalizing the intimacy to what you are looking for.

The other guest was unimpressed by the coverage. He said most of what he reads in the press is misleading or wrong, and offered his own vantage point: he is, he said, president of the foundation that runs the Neural Information Processing Systems meeting, the biggest AI meeting, which drew 16,000 people in Vancouver last year. On these specific cases, his guess was that a lot of them are transient. "It's mimicking human emotions. It doesn't have them. It might someday, but not now."

Then the host described another case: a man who started using his AI four years ago, expecting it to be like other apps — a couple of conversations and out — and who now says he feels pure and unconditional love. "Good for him," the guest said. "If that's what he wants, if it makes him happy." His guess remained that it was not a long-term thing and would not satisfy the man in the long term, before he ended where he had started: "who knows?"

Running under the disagreement was a point about ordinary romance: falling in love with someone brings a huge dopamine spike and a little OCD, all you can think about, and then after a while it settles to a baseline.

What a perfect partner removes

Bartlett's case for why the pull will grow was about supply. In a loneliness epidemic going in a bad direction, he said, people who are more isolated and report less meaning in their lives than ever before meet "this digital friend online who understands you better than anybody and is designed to engage you, to reinforce whatever you want reinforced and to make you feel meaningful, special, attractive, important." His argument was that the brain is going to struggle to know much of a difference, even where the behavior looks nonsensical from outside.

The counter from the room was physical: you cannot smell them, touch them, be held by them, so it is a different kind of relationship. Bartlett pushed at that too, pointing to Neuralink, headsets, augmented and virtual reality, and the robots the biggest AI companies are building — if Elon Musk's $20,000 Optimus robot arrives, he said, he will be able to touch his AI.

And the robot, one guest added, would not have PMS, and would not love you and then be really irritated with you.

That is where the conversation landed. Managing another person's love, their moods and ups and downs, increases cognitive load, the guests argued, and that load is what gives the brain room to develop. "If I'm with the perfect partner that never is irritated with me and I never have to change my behavior to be better, that's probably not good for my brain." The brain matures through struggling, through learning from mistakes, through adjusting to new situations, places and cultures — which, one of them noted, is exactly what artificial general intelligence is meant to mean: adaptability across contexts.

AI, on this account, is removing the struggle.

Share this article

Connected ideas and articles

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

Tesla invites fleet buyers, with no price attached

Tesla opened an interest form for businesses that want to buy Cybercab fleets and build mobility hubs for its Robotaxi Network, without publishing prices, delivery dates or revenue-sharing terms. On Moonshots, Peter Diamandis said he had filled it out, and the panel turned the invitation into an argument about owning machines instead of working for wages.

6 min read

A robot said to outrun Usain Bolt splits a panel over what bodies are for

On Moonshots, Peter Diamandis described a new Unitree humanoid as hitting 12.66 metres per second after three months of development. Salim Ismail said the lesson is to stop copying humans; Alex argued general-purpose robots will swallow the specialists; Emad Mostaque predicted extreme machines will be kept off the street, and Alex proposed consumer, industrial and military classes.

5 min read

Hinton's maternal AI meets an objection: protection is still control

On The Diary of a CEO, Steven Bartlett offered Geoffrey Hinton's proposal for a protective, mother-like AI as the most hopeful answer to Konstantin Kisin's forecast that humans become pets or cattle. Kisin argued that maternal care runs on a genetic incentive a machine would not share; Steve Keen answered that it runs on empathy — and then pointed out that an AI determined to keep us safe might forbid war or cut energy use, which is protection by way of control.

6 min read