May 5, 2026 AI Update: Are You Ready for Life to Start Changing?
CoinBase and PayPal lay off thousands and replace them with AI
The episode’s through-line is Sam Altman’s recent “I hope you are ready for life to change” tweet — and the hosts’ argument that institutions and individuals still aren’t.
Capabilities are accelerating. OpenAI’s Greg Brockman says they’re ~80% of the way to AGI; Grok 4.3 ships cheaper inference and custom voice cloning. China’s open-weight MiMo 2.5 Pro matches Opus 4.6 on coding benchmarks using 40–60% fewer tokens — Scott Galloway’s framing: China’s play is to flood the market with cheap open-weight models rather than chase closed-frontier parity. An Anthropic co-founder is now targeting full recursive self-improvement around 2028 (the AI 2027 paper had previously stated 2027).
Government is stepping in. Microsoft, Google, and xAI will give the U.S. government early access to frontier models for pre-release safety review — a major shift, and one that raises unresolved questions about what “security” review actually covers (capabilities? ideas? alignment?).
AI pluralism: Different models give different answers to the same ethical dilemmas — feature, not bug. The risk is monoculture if one lab wins AGI first. There’s no single “human value alignment” to align to; the better path is multiple frontier models checking each other.
Deployment & integration. Anthropic (with Blackstone/Goldman) and OpenAI (”the Deployment Company”) are launching mid-market services arms — pre-IPO moves to capture integration value, not just frontier capability. DoW signed agreements with most major labs (Anthropic notably absent; Susie Wiles reportedly trying to repair that). Microsoft is putting a legal contract agent in Word. UAE plans to run 50% of government on agentic AI within two years. Suno is at a possible $5B valuation.
Robotics. China deployed humanoid robots over the Mayday holiday — partly norm-building. Meta bought a robotics company. U.S. firms are scaling production, but mostly for factories/warehouses first.
Consciousness & jobs — the real friction. Anthropic and Google have hired consciousness researchers. Richard Dawkins called a model conscious and got dogpiled. Coinbase and PayPal both announced AI-driven layoffs the day of recording. Altman is cooling on UBI; meanwhile a Chinese court ruled companies can’t fire workers just to replace them with AI. Reese Witherspoon got attacked for telling her book club to learn AI — illustrating the U.S. backlash dynamic heading into the midterms.
The hosts’ closing point: People aren’t unready for the technology — they’re unready for the moral dilemmas the technology forces (authorship, representation, misinformation, consciousness, displacement). Without pluralistic, non-polemical public debate, the discussions get captured by political fear and stop producing solutions.