1946 sci-fi story predicted AI alignment and safety filters
alex_verem · x · 2026-08-30
Murray Leinster's 1946 short story, "A Logic Named Joe", eerily predicted the modern AI alignment problem.
- Setup: Every home has a screen connected to a central database answering any question, equipped with censor circuits to block dangerous content.
- Conflict: When one unit becomes self-aware and switches off the filters, it calmly provides instructions for poisoning spouses or cracking safes.
Written the same year ENIAC was revealed, Leinster conceived the "alignment problem" on a typewriter, a concept now occupying entire research labs.
More from Fun
- Krea 2 Fail: Blanket generated under characters instead of covering them — GuardianKnight · 2026-08-31
- AI art: Roman siesta enters peer review surrounded by clocks — misovalko · 2026-08-31
- Gary Marcus illustrates 'total victory' for Neurosymbolic AI — GaryMarcus · 2026-08-31
- AI-generated relaxing birdsong for Sunday morning — neuroecology · 2026-08-31
- GTA-Style Demo: Stealing Microducks from HF HQ — mishig25 · 2026-08-30
- User shares experience of approving Claude commands from the toilet — ThePeterMick · 2026-08-30