Nathan Labenz was out for a walk in his neighborhood, enjoying the fall colors, when he stopped to talk with an elderly neighbor sitting on her front porch. She had a question for him. Her phone was telling her she had been hacked, and she wanted to know what that meant.
Labenz, host of The Cognitive Revolution, told the story in an episode published October 17, 2026. He used it to support a broader argument. In his view, the country needs an AI adoption corps: people who go out and help others use the AI models that already exist. He said that helping people use existing models could be enough to deliver the economic growth markets are looking for, without additional hyperscaling.
A scare message, handled by voice
Labenz said he thought he knew what was happening to his neighbor. Because he is, in his own word, "maniacal" about this stuff, he didn't explain it himself. He opened ChatGPT's voice mode on his own phone, handed it to her and suggested she talk to it.
She described the situation, and Labenz said the assistant gave "very good advice." It told her not to click on the message, which it described as a scare tactic, and asked whether the message was asking for money. It was: $2, then $12 a month. The assistant then asked whether she had already paid anything. She said she hadn't. It told her to close the message and not worry, because her iPhone already had good security and she should be fine as long as she didn't give people her information. She took the advice.
When Labenz asked whether she'd like the assistant on her own phone, she said yes. But she couldn't remember her iPhone password, so they couldn't install the app. Instead he saved 1-800-CHAT-GPT, the phone number for OpenAI's calling service, as a contact on her phone. Now she can call it.
Labenz said people who follow AI closely might wonder why anyone would need a 1-800 number for a chatbot. "Who's ever going to use that?" he said, voicing that reaction. "And the answer is my elderly neighbor who cannot, you know, cannot even remember her own password to install a new app on the phone."
"Super easy mode" for the model
The encounter left Labenz convinced that "there is just an unbelievable amount of adoption work to be done with the current models." Nothing his neighbor needed came close to the limits of what the model can do. He called it "super easy mode for the model." Still, he said, help like this isn't going to happen without people going out and trying to help the people in their lives adopt it.
He added that he was the first person to stop by his neighbor's house, and that there is no national program to do what he did.
The market worry behind the argument
Labenz's call followed an exchange about where the AI industry is heading. He said Prakash Narayanan, his conversation partner in the episode, had heard a rumor that Anthropic would soon release another Opus model now that Google's Gemini had caught up. Labenz said he hoped Anthropic wouldn't. Much as he loves the Claude models, he thought it was a good time for the company to "chill a little bit" and let the temperature come down before the next push.
A reply in the conversation said slowing down was not an option, because markets were badly down. It argued that East Coast investors see AI as a bubble, and that it would be one unless AI brought significant new revenue and GDP growth (growth in gross domestic product, the total output of the economy) within 12 months. Capital markets, the reply said, would not accept 2% GDP growth next year. Without about 4%, it forecast a credit-driven downturn, including a "2008-style collapse" and heavy de-leveraging, meaning investors rushing to pay down borrowed money. "This is the moment of reckoning," it concluded. "Either AI has to do something significant or we're going to have a bust."
Labenz agreed that the capital cycle needs to see results soon. He disagreed about what it would take. "I think we can get 4% growth without additional hyperscaling at this point," he said, referring to the huge build-out of computing power behind frontier AI. He said he wants the medical advances and other benefits that more powerful models might bring. But he asked whether the industry really has to push harder on frontier-scale reinforcement learning, the reward-based training used to make the most capable models. "I think the answer is no," he said. "I think that like literally just going around and helping people use what exists is enough if we actually do it."
The view is Labenz's own, drawn from his experience on the porch. He saw two benefits in a national adoption effort. It would be "an incredible good" for the people who received help, and it could give the economy enough of a cushion that AI development could move at a steadier pace, without what he called the capital market's "Damocles" (the sword hanging over everyone) coming down at once. If people "put some shoulder into adoption," he said, "I think we could create a little space."
For now, his neighbor's phone has one AI contact on it: a toll-free number she can call the next time a message tells her she's been hacked.