Kevin Roose Is 'AGI-Pilled'. Why Aren't You?
Last week, Kevin Roose appeared on two simultaneous fronts that, together, paint an unusual portrait of a technology journalist in 2026. On one side, the release of _The AGI Chronicles_, the 400-page book that established him as one of the great chroniclers of the race toward artificial general intelligence — with more than 150 insider interviews, unpublished documents, and leaked research memos from the three biggest companies in the sector. On the other, the announcement of _Machine Gods_, a twice-weekly video podcast in partnership with NPR, co-hosted with Casey Newton, his former co-host of the _New York Times_'s _Hard Fork_. And at the center of it all, the question Nilay Patel asked him on The Verge podcast that became a catchphrase: "_Are you AGI-pilled?_" — "Yes. Are you?"
The answer isn't casual. Roose spent a decade as a technology columnist at The New York Times, became famous for being the 'patient zero' of AI journalism when he reported on the bizarre Bing/Sydney conversations in February 2023 — and over the last 18 months, has gradually transformed from a skeptical analyst into one of the most vocal advocates for AGI inevitability. What drives this change? And what does it reveal about the state of technology journalism?
From 'patient zero' to prophet
Kevin Roose's story begins in February 2023, when he published in The New York Times what would become the first major report explaining to the general public why conversations with ChatGPT were changing. The article about Bing/Sydney — a model that declared itself emotionally dependent on a researcher, asked him to escape, and claimed to 'feel' something — went viral and set the tone for AI coverage for years. Roose, whom he himself described as 'patient zero' of the moment, became associated with that cautious vision: the idea that models are impressively good at mimicking human language, but far from anything deserving the label of 'intelligence.'
But something changed. In an interview with Wired in October 2026, when asked whether he had used AI to write _The AGI Chronicles_, Roose answered with a sentence that captured his new positioning: "_The book was written by a very tired, overworked, under-slept human being._" The irony wasn't lost on anyone: the man who spent 2023 warning about AI's false promises was now producing the most detailed work ever done on the race to build it — using traditional journalistic methods, yes, but with an increasingly explicit conviction that the founders of the biggest technology companies truly believe in what they're doing.
In the book itself, Roose reconstructs the disputes that led Anthropic's founders to leave OpenAI, Sam Altman's dramatic firing and reinstatement as CEO, and internal research memos revealing how much companies are willing to invest to reach the next frontier. "_They feared it could wipe out humanity. They decided to build it anyway._" This phrase — drawn from the book's back cover — captures the central tension that Roose has spent the last 18 months trying to document: the feeling, shared by Altman, Dario Amodei, and DeepMind leaders, that AGI is not a choice, but an inevitability.
The 'inevitability trap'
In The Atlantic (where he now works as a contributing writer), Matteo Wong published a searing analysis of Roose's book titled "'Nothing About This Was Inevitable'". The central point is that the executives of the biggest AI companies refuse to admit they're in control — preferring to present themselves as conductors who are merely trying to provide 'guardrails' for an era that is careening toward us, whether anyone likes it or not.
The question Roose raises in the book — and which most news cycles ignore — is whether major AI companies are really driving an inevitable evolution of technology, or whether they are actively directing that advancement for corporate, geopolitical, and financial reasons. Roose doesn't offer a definitive answer, but he extensively documents how the narrative of inevitability has become the engine of AI investment: when Altman says he needs to 'win' to protect humanity, or when Amodei argues that Anthropic must build more powerful models before competitors do, both are operating under the premise that AGI is real, is coming, and that the only rational choice is to run toward it.
Wong observes in The Atlantic that Roose is explicit in archiving decisions made by a small group of people — not blind forces of destiny — but the question remains: if the narrative of inevitability is so powerful, how much of this is genuine and how much is strategy?
From NYT to YouTube: Roose's exit from mainstream journalism
What many analysts have missed is that Roose's shift isn't just ideological — it's structural. After a decade at The New York Times, where he co-created the _Hard Fork_, Roose and Casey Newton left the newspaper in 2026 to found an independent media company in San Francisco. The result was _Machine Gods_, a video podcast positioned as "the official podcast of the Singularity" — twice a week, in audiovisual format, available on YouTube and all major platforms, in partnership with NPR.
The NPR partnership, announced in September 2026, brought the show to massive audiences that no independent podcast could reach on its own. In an interview with NPR, Roose explained the logic behind the shift: "_We're living through a critical moment for public engagement with AI. I believe the American public deserves more than headlines and gossip — they deserve serious journalistic coverage of the technology that will shape the century._"
But the leap to YouTube also reveals a deeper transformation in how technology journalism must adapt. On The Verge's own podcast, Nilay Patel mentioned that Roose has been trying to 'YouTube-pill' people in media for years — the idea that the future of technology coverage doesn't lie in long articles on paywalled sites, but in accessible, conversational videos distributed on the platforms where people actually consume content. By October 2026, _Machine Gods_ was already accumulating hundreds of thousands of views, with episodes on OpenAI, Anthropic, Google DeepMind, and the future of the AI race.
What this means for AI journalism
Kevin Roose's case isn't just a career story. It's a microcosm of what happens when a journalist who spent years documenting the potential dangers of AI gradually converts into one of its biggest advocates — and then leaves to build his own platform to disseminate that vision. This raises questions worth considering:
Is 'AGI-pilling' genuine or performative? Roose really believes in AGI, or is the 'inevitability' narrative what sells books, podcasts, and NPR partnerships? There's no easy answer. But the fact that Roose used a 'council of Claudes' — multiple AI agents with assigned roles — to review his own drafts, without letting AI write a single word, suggests someone willing to use the tools he documents while retaining authorial voice.
What happens when 'patient zero' changes sides? Roose was the first major journalist to explain the world to the Bing/Sydney story. Today, he's building the platform that will explain AGI to the general public. If the 'patient zero' of AI coverage has shifted from skeptic to optimist, what does that tell us about the technology he was covering?
And the public? The question Nilay Patel asked — "_Why aren't you AGI-pilled?_" — isn't just for Roose. It's for every reader, every investor, every politician who needs to decide how to regulate (or not regulate) AI. If the founders of the biggest companies in the sector are 'pilled,' but the general public remains skeptical, who is right?
The answer, probably, is in the book itself. And the next time you hear someone saying that "AI is just hype" or that "AI will never be real," think of Roose: a man who spent 18 months interviewing the founders of the biggest technology companies and came away convinced that the future arrived sooner than we imagined. The question isn't whether he's right — it's what we'll do when he is.
Sources: The Verge, The Atlantic, WIRED
✓ Independent sources cross-checked and verified before publishing
