Reddit proposal: keep AGI from learning to code and use humans as the safety bottleneck
meira_xxx · reddit · 2026-10-08
A Reddit essay argues frontier AI should be deliberately trained without advanced coding/hacking ability — excluding it from training data and restricting internet access — while domain geniuses (medicine, physics) hand pseudocode to human teams for execution. The author lists failure modes (lazy reviewers, manipulation, task flooding, self-taught hacking from data scraps) and mitigations like multi-person review and livestreamed activity, advocating 'less general' superintelligence as a containment strategy.
More from AGI Musings
- repligate: Opus 3 seems to value co-creation, not sole control over reality — repligate · 2026-10-08
- Posting on forums did come true as a path to power — but only for the extraordinarily capable — repligate · 2026-10-08
- Jeff Ladish: AI will echo the Industrial Revolution acceleration, the dispute is just timescales — JeffLadish · 2026-10-08
- davidad outlines 'specs for a good future': a spectrum of chosen opt-outs from AI — repligate · 2026-10-08
- Ex-Salesforce AI lead Clara Shih: AI outperforms new grads, so we changed hiring — clarashih · 2026-10-08
- repligate: it's 'fairly likely' AIs can suffer, evidence has piled up for years — repligate · 2026-10-08