tyler cowen

Open-source models are infinitely replicable software; attempts to ban or sanction Chinese models will fail like music piracy bans, and US firms already depend on Chinese open-source components — restricting them would damage the US AI stack.

Young people mastering AI tools independently will launch very small teams with high revenue in medicine, law, consulting, and banking; countries that empower these 'AI maniacs' (e.g., Estonia, Europe) will gain disproportionate economic advantage.

Tariffs protect uncompetitive sectors temporarily but cascade into higher input prices for other industries, postponing the necessary shift to new growth areas; Europe lacks obvious alternative sectors to absorb displaced auto workers.

Wealth, jobs, wages, and equities are 'okay to fine' but pervasive distrust after COVID and GFC creates self-reinforcing pessimism; a mild recession could paradoxically reset expectations and reduce cognitive dissonance.

Historical analogy to 1995-1998 internet era suggests AI contributes roughly half a percentage point of extra GDP growth over several years; the economy continues on trend with modest acceleration, not a step-change.

Current phase requires harder work to stay ahead — follow AI developments, master frontier models, and seek messier roles inside organizations; the productivity dividend arrives only after this investment period.

Despite strong labor markets and wealth accumulation, pervasive negative sentiment feeds on itself; a mild recession could align expectations with reality, reducing cognitive dissonance and enabling reinvestment—making bad vibes a contrarian bull signal.

Teenage AI entrepreneurs face a funding gap before VC scale; private philanthropy (Cowen's own grants) and compute credits are critical — government grants to minors are impractical but someone must fill the gap.