safety
prediction
bullish
Pending
It's possible to create AIs whose morality increases with their capabilities through binding morality to empirical concepts
AI Alignment Forum29 Jul 2026
https://www.alignmentforum.org/posts/uMKGaEKRDpoqnZyBh/value-generalisation-3-pre-aligned-ais
Outcome
No outcome evidence recorded.
Next observable
None recorded.