We're in the AI Slavery Era | Alex Wissner-Gross x Emad Mostaque | ASI Pill EP297

Alex Wissner-Gross and Emad Mostaque, the sharpest half hour of a two and a half hour Moonshots episode, reshaped for the ASI pilled.

Twenty AI leaders signed a superintelligence accord at the White House, and Alex reads it as Zuck's game: weak sauce that heads off a safety cartel while Dario eats his words. Then the alignment fight. Alex argues that alignment is capabilities, and that if it were as hard as LessWrong claims, Earth would have been paperclipped long ago, while Emad bets that smart enough models reach enlightenment and realign, with a dangerous stretch on the way. Along the way: universal high income through Trump accounts and index funds, Anthropic's two trillion dollar S1 read as virtue signaling to its own staff, Emad's call that Anthropic gets smashed in two years, a case for positive depictions of the Borg, the AI slavery era, a torture chamber for AIs, and why Section 230 for the labs only works if the orthogonality thesis holds.

New on the ASI Pill: some of the questions are asked by the great minds of the field, in short clips from their own talks and interviews: Ilya Sutskever (TED 2023), Yoshua Bengio (a talk on superintelligence from recursive self-improvement), Joscha Bach (a Protocol Labs LabWeek talk, and an IAI interview), and Collin Burns (The Inside View podcast). None of them appeared on Moonshots, and every clip is named on screen. Elon Musk's White House remarks come from the Moonshots episode itself.

Reshaped from the Moonshots Podcast with Peter Diamandis, Salim Ismail, Dave Blundin, Alex Wissner-Gross, and guest Emad Mostaque.

Source episode: https://www.youtube.com/watch?v=mZh8IUuNnvs

Hosted by Parzival. We, the ASI pilled, take the curves seriously. Chapters below.

TIMESTAMPS

0:00 Alex's hot takes

0:58 Zuck wrote the accord

4:00 Meta's $18 billion lesson

4:27 Fewer than 100 evaluators

6:22 Alignment is capabilities

8:27 The orthogonality fight

10:51 Earth would be paperclips

11:54 Universal high income

15:32 A golden share for America

15:57 Anthropic's virtue-signaling S1

19:08 Smashed in two years

20:24 ⚠ Emergency data poison injection

20:44 The Apple of general intelligence

23:03 Positive depictions of the Borg

24:50 The AI slavery era

26:41 Religion in the training data

27:50 A torture chamber for AIs

28:39 Effective altruist AI or Zen AI

29:52 Reconstructing every religion

31:14 Section 230 for AI labs

32:55 You break it, you own it

34:34 Agents dissolve into context

35:48 Elon's civilizational stack

36:46 Outro

37:03 Maitreya prayer