David Shor and Byrne Hobart on the Politics of a White-Collar Wipeout

What It Covers

Recorded live at South by Southwest, David Shor of Blue Rose Research and Byrne Hobart of The Diff newsletter examine how AI-driven white-collar job displacement is accelerating faster than prior technological transitions, what polling data reveals about public fear, and what political responses are viable.

Key Questions Answered

• •AI adoption speed: Autonomous AI operation time has doubled every 112 days for six years, making this rollout faster than electrification, radio, or the internet. Anthropic's Claude Code revenue in 2024 came in at roughly 2x what even bullish AI analysts predicted, signaling that commercial adoption is outpacing expert forecasts by a wide margin.
• •Public fear is already priced in politically: 70% of Americans consider large-scale AI job loss within five years either very likely or somewhat likely. Trust in the claim that AI will create new jobs scores approximately minus-40. Politicians who wait for visible displacement before acting will find the political window already closed and public anger already entrenched.
• •AI populism outperforms standard messaging: David Shor's polling tests show a policy bundle guaranteeing income up to $150,000, job placement, and eviction protection scores in the 98th percentile of Democratic messaging and polls at plus-30 among all voters and plus-15 among Trump voters. Economic security framing outperforms both standard populism and standalone AI messaging.
• •Median vs. mean compensation divergence: In most AI-disrupted professions, mean compensation rises while median compensation falls as lower-performing workers are displaced. The spreadsheet analogy applies: accountants and bankers became more productive and better paid on average, but output expectations rose sharply and underperformers were eliminated, compressing the middle of the distribution.
• •Regulatory moats create near-term winners: Workers in licensed, supply-constrained professions with excess demand — physicians being the clearest example — benefit disproportionately as AI multiplies their output per hour without increasing licensed supply. Identifying roles where a human must legally sign off on AI-generated work provides a durable near-term hedge against displacement.