Mark Zuckerberg published his 6,500-word manifesto “The Future Is for Everyone,” a sunny vision of personal superintelligence for every user — and Kevin Roose and Casey Newton spend the first segment deciding whether it’s credible. Their verdict: it’s a policy wishlist for Meta dressed in optimism, asking for accelerated data-center permitting, maintained export controls on advanced chips (which advantage Meta’s open-weight models over Chinese alternatives), reduced training-data restrictions, and legal protection for distillation — all landing the same week a New Mexico judge ordered Meta to pay $567 million into a teen mental health abatement fund and ruled its platforms a public nuisance. Casey’s frame: Zuckerberg’s answer to AI danger is “giving a dragon to everyone” — “while I’m sure most people will spend their time creating personalized baking recipes… there are other people that are going to be launching cyberattacks and engineering novel bioweapons. I say, hold your horses or your dragons.” The hosts don’t count Meta out (“The first rule of Mark Zuckerberg is never count out Mark Zuckerberg”), with Roose upping his superintelligence odds from 1% to 10% in six months — which is exactly why they’re nervous, citing the old DeepMind maxim that the real danger is a company that thinks it’s designing tools but is actually designing superintelligence. Then Pangram CEO Max Spero — “the number one narc,” the slop janitor — explains why AI text detection finally works: instead of the old perplexity metric, Pangram trains a classifier on paired human/AI essays from a clean pre-2022 corpus, accumulating weak signals across documents (“humans kind of have a wider, less mode-collapsed decision tree”), deliberately tuned to minimize false positives so “if Pangram says that something is AI, we can be very confident that it is largely AI-generated.” On why it matters: “AI as a tool is probably the wrong abstraction. It’s a little bit closer today to AI is an employee, or another individual that you collaborate with… bot traffic has just surpassed human traffic… we need to, as humanity, discriminate in favor of humans.” The new segment Running the Numbers covers an Anthropic non-mathematician making progress on a Riemann hypothesis side problem by telling Claude “you are the world’s most capable large language model to date, you got this” — “this feels like the production of new knowledge to me” — plus Airtable selling to Bending Spoons at $1.29B against an $11.7B 2021 valuation (“if your business is a fancy spreadsheet, you are in for a rough time”) and South Korean chip nerds with $400K-500K annual bonuses dominating the dating market (“is that boy really cute, or are you just wearing Anthropic goggles?”).
Hard Fork #208: Zuckerberg's Anti-Doom Fantasy + Pangram's AI Detector + AI Math
Roose and Newton fact-check Zuckerberg's 6,500-word AI manifesto — policy wishlist dressed as optimism, with a dragon-to-everyone critique — then Pangram CEO Max Spero explains how AI text detection finally got good, and the hosts debut Running the Numbers with a Riemann hypothesis side problem solved via positive affirmation.