Economists do the math: society may be underinvesting in AI safety by 30x or more
JimCripe · reddit · 2026-09-24
The podcast Diving In features economist Justin Wolfers applying a framework from Stanford economist Chad Jones to the question: how much should we pay to reduce AI catastrophe risk?
Core argument
- Analogy to airbags: the US spends $6B/year on airbags to save 5,000 lives — about $1.2M per life, versus the government's $10M value of a statistical life. We already pay for crashes we'll likely never have.
- Applying the same logic to AI: buying down a 1% catastrophe risk could be worth $5,000 per person per year — hundreds of billions across America, trillions worldwide.
- Chad Jones estimates society may be underinvesting in AI safety by a factor of 30 or more.
More from AGI Musings
- Researcher slams OpenAI's redefinition of alignment as just being more useful — nabla_theta · 2026-09-24
- Why AI hasn't reshaped society yet: same tool, but social relations take time to reset — tinyfool · 2026-09-24
- Blogger: time spent on X almost predicts how well you understand AI — manosaie · 2026-09-24
- OpenAI researcher: fewer than 5% of 10,000 US rare diseases have cures—AI may change that — jamdac · 2026-09-24
- Beff Jezos: Zuck has the most e/acc vision of AI—personal superintelligence for everyone — beffjezos · 2026-09-24
- e/acc figure Beff Jezos backs Anthropic's bio/acc turn—unless they build bioweapons to justify regulation — beffjezos · 2026-09-24