A.I. Safety Goes Mainstream + a ‘Hard Fork’ Exit AMA

Kevin Roose and Casey Newton discuss the sudden mainstream shift toward AI safety concerns, triggered by researcher Jacob Coxon’s resignation and the Hugging Face incident. They analyze the unprecedented call for a coordinated global slowdown by rival leaders like Dario Amodei and Sam Altman, contrasting this with the Trump administration’s skepticism. The hosts also address the show’s final episode, announcing their new NPR series "Machine Gods," and answer listener questions about their personal history and production process.

Topics discussed

Ripple advertisement Anecdote about writing a history of science Introduction of hosts and show finale Discussion on the sudden shift in AI safety discourse Jacob Coxon's viral resignation post The SF bubble and median dinner table conversations P-doom probabilities and researcher optimism Early signs of agent collaboration and risks Internal conflicts between safety and capability teams Call for a global coordinated AI slowdown Agreement among AI leaders and Elon Musk's stance Lack of trust in tech companies and government reliance Congressional efforts for bipartisan AI safeguards Public fear and media requests for commentary Kevin's earlier predictions and current optimism Concerns about recursive self-improving agents Cost of slowing down model releases Critique of marketing hype and self-interest Mark Zuckerberg as a messenger for AI safety Humanity banding together and unlikely alliances Existential risk moving inside the Overton window Risks of rogue agent swarms by non-experts Distribution of models to the public Regulations for hobbyists and tinkerers Closing thoughts on the show's themes Ripple advertisement Stay Calm Investing book promotion Announcement of the new show Machine Gods Listener question about how the hosts met Kevin and Casey's first meeting at a book party Pitching the show to the New York Times Developing the signature cold open laugh Producers selecting the best cold open jokes Journalism-first approach to podcasting Editing out boring parts of conversations Kevin's favorite podcasts: history and tennis Admiration for the Search Engine podcast Pretending to hate a world-class podcaster Use of prep documents and structure Joke writing and the writers room Advice for young podcasters: improv comedy Giving men a podcast to contain their bits Editing out jokes and standards department Low conflict and friendship off-mic Therapy and navigating conflict as adults Kevin's Jimmy Buffett moment Displaying values and pragmatism in media Learning new things and changing minds Connecting with listeners at live events Ripple advertisement Stay Calm Investing book promotion Why Kevin says 'frontier' that way Nostalgia for obvious AI misalignment Incorporating Web3 into new ventures Tech movies and TV shows in the fall Comparison to the Alien franchise Preference for period pieces over current events Wear and tear on AI chips Network effects of being in San Francisco Recruiting and density in the Bay Area Value of a college diploma in the AI era Uncertainty about the future of college Protecting against rogue agent swarms The end of an experiment Dodge the issue of current events in the book Episode without AI: quantum computers Internet culture and TikTok trends Dislike of current internet culture Journalism as a team exercise Resonance with a larger group of people Processing the rapid changes in AI Finding out what the show is by making it Evolution of tech and AI in 2.5 years Awakening that the world is changing Wish for the podcast to be important Gratitude to the crew and early supporters Skepticism from Kara Swisher's audience IBM advertisement Credits for music and video production Staplers in the new office

Episode summaries and topics are generated from the audio by Castria — a free podcast app that skips ads automatically. IP Geolocation by DB-IP

A.I. Safety Goes Mainstream + a ‘Hard Fork’ Exit AMA

Summary

Topics discussed