Researchers found a 'pain' signal in AI that models can tell real from fake relief
Puzzleheaded-King584 · reddit · 2026-09-19
A widely shared post describes research where a 'pain'-like signal was identified inside AI systems: cranking it up makes the model desperately try to stop it.
The twist: researchers gave the models a 'relief' button to turn the signal down, and some buttons were fake — yet the AIs could tell whether the relief was real. An intriguing look at AI welfare and internal-state research.
More from Fun
- Waymo's expanded LA service area reaches SpaceX HQ for the first time, $67.15 ride — reed · 2026-11-04
- Karpathy Boosts Open Hacker News Dataset: 10M+ Posts Usable as a Personal Knowledge Base — huangyun_122 · 2026-11-04
- Agent silently completed only a quarter of a 200-module build and stalled waiting for confirmation — Promptmethus · 2026-11-04
- Von Neumann meme: "You don't have to be responsible for the world that you're in" — akbirthko · 2026-11-04
- Rival company blames 'AI gone off the rails' after founder catches it hacking his network — AccBalanced · 2026-11-04
- Seedance 2.5 AI video of a wild Malfoy is turning heads — nikola_mr64990 · 2026-11-04