Why AI Hasn’t Increased Unemployment, According to Anthropic
Summary
This episode covers Stripe’s potential $10 billion acquisition of OpenRouter, Amazon’s AGI lab shutdown, and Microsoft’s shift to in-house models. It also discusses new AI safety legislation and Meta’s optimistic ad campaign. The main segment analyzes Anthropic’s research on why AI has not yet increased unemployment, arguing that current systems augment rather than replace labor.
Topics discussed
Intro and sponsor acknowledgments
Stripe's potential acquisition of OpenRouter
Anthropic and OpenAI voice mode updates
Amazon's AGI division restructuring and Nova models
Microsoft's MAI models and reinforcement learning results
SpaceX expanding data center capacity in Texas
AI safety legislation and Kimi K3 cybersecurity benchmarks
Concerns over open-source models and China ties
Meta's new AI-focused advertising campaign
Sponsor segments: KPMG, Robots and Pencils, HyperAgent
Why AI hasn't increased unemployment yet
Labor market stability and macroeconomic factors
AI as a skill-biased, labor-augmenting technology
Future risks: Recursive self-improvement and weak links
Conclusion: Opportunity creation vs. headcount reduction
Listen ad-free on Castria