8/8 Full paper: https://t.co/UTE5KPFg23
I'm an independent researcher. No institution, no funding, no agenda. Just the analysis. Would genuinely welcome pushback from economists, finance people, or anyone working in AI deployment.
7/8 This thesis is testable. I identify four specific thresholds: consumer delinquency above 8%, regional bank charge-offs above 2%, 10-year Treasury above 5.5% with GDP below 1.5%, and unemployment above 10%. If two or more hit simultaneously by 2028-2030, the cascade is activating.
6/8 And it can't be stopped. The US-China AI arms race has the same logic as the nuclear arms race. The consequences of letting your adversary develop it first are worse than the consequences of developing it yourself. Every major state actor is accelerating. Deceleration is structurally impossible.
5/8 Here's the specific danger. Lawyers, analysts, software engineers, accountants β 9-11 million workers earning $62k-$182k median β carry $400k-$700k mortgages. When they're displaced, it doesn't just hurt them. It triggers a credit cascade the post-2008 regulatory framework was never designed to handle.
4/8 The financial system has zero slack to absorb this. Credit card delinquency approaching 2008 levels. $1.277 trillion in card balances. 29.3% of auto trade-ins underwater. $18.8 trillion in household debt. The shock is arriving into a system already at maximum fragility.
3/8 AI capability is doubling every 4-7 months. Claude Opus 4.6 can now autonomously complete 14.5 hours of skilled professional work reliably. On SWE-bench, AI went from solving 4.4% of real software engineering problems in 2023 to 71.7% in 2024. This is not incremental.
2/8 Every prior automation wave hit low-wage workers first. Factories. Clerks. Retail. The financial system survived because high-income professionals kept paying their mortgages and driving consumption. AI breaks this pattern entirely. It's attacking from the top.
1/8 I've spent months researching why the AI economic disruption will be different from every previous automation wave. The conclusion is uncomfortable. A thread. π§΅