Zuckerberg’s Anti-Doom Fantasy + Finally an A.I. Detector That Works + A.I. Math
Summary
Hard Fork hosts Casey Newton and Kevin Roose discuss Mark Zuckerberg’s AI manifesto, analyzing its policy goals and safety risks. They interview Pangram CEO Max Spiro about the rising effectiveness of AI text detectors in identifying "slop." The episode also announces the show’s upcoming finale and an Ask Me Anything segment.
Topics discussed
Intro: Dream Beans AI app and cartoon slop
Show intro and listener feedback request
Mark Zuckerberg's 6,500-word AI manifesto
Meta's lawsuits, teen safety, and AGI ambitions
The need for tangible AI benefits over essays
The three eras of AI slop and detection
How Pangram detects AI text using perplexity
Humanizers, model updates, and detection challenges
Why detecting AI matters and Substack integration
Anthropic watermarks and future detection tech
Pangram's proudest catches and interview wrap-up
Running the Numbers: AI and the Riemann hypothesis
Airtable acquisition and SaaS valuation drops
Dating math: Chip engineers and AI wealth
Listen ad-free on Castria