Could compute end up split 98:1:1 — most of it defending against malicious AI?
seanmcdonaldxyz · x · 2026-10-25
The author speculates that over time compute could settle into a 98:1:1 split: 98% spent defensively stabilizing against a tiny (0.0001%) malicious minority, 1% on recursive self-improvement, and only 1% left for human purposes.
An open-ended conjecture pointing at a counterintuitive future where defensive spending consumes nearly all compute.
More from AGI Musings
- "Humans are reality's sensory layer": devs concede coding supremacy to Opus — robleclerc · 2026-10-25
- 'We thought an internal model winning IMO gold was peak insane — lol' — flowersslop · 2026-10-25
- Hobbyist trains a 118k-param JEPA to rediscover Mendeleev's missing elements — it fails — WarningSure1670 · 2026-10-25
- GMO CEO warns AI misuse may collapse the internet; X users push back — basedjensen · 2026-10-25
- Redditor tallies top-11 AI debt: $650B on-book, $3.5T off-book AI commitments — mynameiswearingme · 2026-10-25
- Longevity researcher's question sparks debate: what happens to inheritance when nobody dies? — SydSteyerhart · 2026-10-25