Moonshots EP 230: AI CEOs Come Online — Sam Altman’s Replacement Plan, Job Loss & Solve Everything Launches

Summary

Two-part episode. Part one covers Sam Altman’s Forbes interview where he says he wouldn’t stand in the way of AI becoming OpenAI’s CEO. Dave shares board-level observations from three companies (including a $2T asset manager) where all plans are being restructured into AI-digestible document formats. Alex claims billion-dollar-revenue AI-run companies likely already exist with human CEOs serving as “meat puppets” for legal purposes. The panel discusses OpenAI’s 70% reduction in model release cadence (97 days to 29 days) and the transition from pre-training scaling to recursive self-improvement era. Sem warns that the window to access frontier-quality models openly is closing. Part two previews “Solve Everything,” a 9-chapter paper by Peter and Alex on reaching abundance by 2035, positioned as their answer to the “Situational Awareness” paper. Also covered: Vision Claw (lobsters gaining vision through Meta Ray-Ban glasses), US job losses hitting Great Recession rates, and the Moravec’s Paradox observation that CEO-level work is being automated before manual labor.

Key Segments

• [00:04-00:12] AI CEO debate: Sam Altman’s Forbes interview, Dave’s board meeting insights, Alex claims $1B AI-run companies already exist, Moravec’s Paradox inversion
• [00:13-00:20] Model release cadence: 70% time reduction, recursive self-improvement era, Sem warns the open access window is closing
• [00:20-00:28] Vision Claw demo, lobsters as embodied agents through smart glasses, creator-vs-consumer imperative
• [00:28+] “Solve Everything” paper preview

Notable Claims

• Alex believes billion-dollar run-rate companies with AI effectively making all decisions already exist, with human CEOs as legal figureheads
• Sem warns the window to access frontier-quality AI openly will close within ~2 years as models go dark for competitive advantage
• OpenAI’s internal models are roughly 3 months ahead of public releases, but 3 months of recursive self-improvement now equals massive capability gaps

Guests / Panelists

Peter Diamandis (host), Alex Weiszner-Gross (AWG), Dave (DB2), Salem Ismail (Sem)

RDCO Mapping

• AI COO parallel: The AI CEO discussion maps directly to our operating model. Dave’s point about restructuring all company information into AI-digestible documents mirrors what we’re doing with the vault.
• Access window closing: Sem’s warning about frontier model access going dark is a strategic consideration for RDCO’s tool dependencies.
• Content opportunity: The Moravec’s Paradox inversion (CEOs automated before electricians) is a strong contrarian Sanity Check angle. The “Solve Everything” paper is worth tracking as a reference source.