Kevin Roose Didn’t Use AI to Write His Book About AI
Kevin Roose discusses his new book "The AGI Chronicles" and his departure from the New York Times to launch Machine Gods Media with Casey Newton. He explains his vision for AI coverage that avoids both hype and dismissiveness, describes the competitive dynamics between OpenAI, Anthropic, and Google, and details how he used AI tools to research and edit the book while maintaining transparency about the process.
Summary
In this interview with Wired's Katie Drummond, tech journalist Kevin Roose discusses his transition from the New York Times to launching a new media company called Machine Gods Media with fellow writer Casey Newton. The company will produce a podcast called Machine Gods in partnership with NPR, reportedly receiving an offer exceeding $5 million—significantly more than the initial Bloomberg report of $5 million.
Roose explains that he and Casey identified a gap in AI coverage between two extremes: those dismissing AI as hype with no real impact, and those promoting it without discussing risks. They aim to practice "AI realism"—taking the technology seriously while acknowledging both its power and dangers, without slipping into doom-mongering. They chose NPR as a partner for its independent mandate, broad reach through radio distribution, and willingness to let them maintain creative control.
Roose's book, "The AGI Chronicles," documents the race to create artificial general intelligence through reporting on three major companies: Google (which developed the Transformer architecture), OpenAI (founded by Elon Musk and Sam Altman to challenge Google), and Anthropic (founded by Dario Amodei and others who left OpenAI). He describes these companies as engaged in a "blood feud" rather than friendly competition, with all key decisions made by approximately 50 people in San Francisco.
When asked whom he trusts with AI's future, Roose declines to trust any single person, noting that AI leaders are flawed but generally more thoughtful about risks than social media pioneers were. He argues they genuinely believe AI development is inevitable and that safety-conscious companies should lead rather than bad actors. He emphasizes the need for broader perspectives—philosophers, artists, lawyers—in AI decisions, highlighting Amanda Askell at Anthropic as an example of how hiring virtue ethicists has improved AI safety considerations.
Roose expresses qualified optimism about AI's potential in medicine and disease research, citing DeepMind's AlphaFold breakthrough. However, he acknowledges concerns about the speed of development and notes that the people building AI have unusually high tolerance for radical social change compared to the general population.
Regarding his book's creation, Roose was transparent about using AI tools extensively: he filled NotebookLM with research materials and used it for querying, organizing notes, fact-checking, and editorial feedback through what he calls his "Council of Claubes"—Claude instances assigned different personalities and viewpoints. He ran the completed book through PanGram to verify it wasn't AI-generated. He argues that while people intellectually accept AI-assisted work, they emotionally prefer knowing humans created content they consume, though he notes people cannot reliably distinguish AI-written from human-written text in blind tests.
About this episode
The author of The AGI Chronicles says his book “was written by a very tired, overworked, under-slept human being.” Join WIRED’s best and brightest on Uncanny Valley as they dissect the collision of tech, politics, finance, and business, from Alexis Ohanian’s newest tech venture to the effects of inaccurate information from artificial intelligence (AI) chatbots on social protests. Learn more about your ad choices. Visit podcastchoices.com/adchoices
Key Insights
- Roose argues that the AI industry is characterized by a 'blood feud' rather than healthy competition, with companies like OpenAI and Anthropic that spawned from the same people now functioning as mortal enemies in pursuit of AGI.
- Roose claims that approximately 50 people in San Francisco made all relevant decisions about AI development, and they are not representative of normal society—they have unusually high tolerance for radical change and enjoy upheaval in ways most people do not.
- Roose contends that AI leaders genuinely believe AI development is inevitable regardless of their efforts, so the best outcome for humanity is for safety-conscious companies to reach AGI first rather than bad actors, terrorist groups, or less thoughtful competitors.
- Roose observes that people in blind tests prefer AI-generated writing to human-written text but experience plummeting approval once they learn AI created the content, suggesting the value is psychological rather than based on actual quality differences.
- Roose argues that bringing philosophers, artists, and creatives into AI decision-making at companies like Anthropic—exemplified by virtue ethicist Amanda Askell—has tangibly improved how AI systems are designed and trained for moral behavior.
- Roose claims he feels more hopeful about AI risks now than months ago because the conversation about AI safety has entered mainstream discourse, whereas previously only a small community of researchers took these concerns seriously.
- Roose contends that if something goes catastrophically wrong with AI, it will likely result from human error and boring mistakes rather than the dramatic rogue AI scenarios often depicted in science fiction.
- Roose used AI extensively as a research and editing tool while writing his book, including using it for fact-checking, note organization, and receiving editorial feedback, while remaining transparent about the process to maintain credibility.
Topics
Transcript
I recently chatted with McKinsey's Dan Swan about how businesses successfully transitioned to AI, the coming agentic wave, and even the sad loss of tennis line judges. Here's a sneak peek. Some people are buying software and calling it a transformation. The people that are really getting the most value from it are fundamentally doing something very different, which is they're taking that technology and they're incorporating it with changing the capabilities of their people, fundamentally redesigning their processes and tracking the impact through the bottom line. There's much more in the full interview with Dan Swan, which can be found in its entirety on Wired.com. Also, check out McKinsey's playbook, Rewired, available at your favorite bookseller. From Wired, this…
Full transcript available for MurmurCast members
Sign Up to Access