Reaction
The alignment-versus-sandboxing-versus-monitoring debate in AI safety
A user argues that AI safety needs all three, comparing the choice to asking whether preventing theft requires laws, police or locks.
TLDR
One post argues that alignment, sandboxing and monitoring should not be treated as alternatives. It likens them to laws, a police force and locks: different safeguards against theft.
Combined views
2.3K
2 Sources, first seen 3h ago
54 likes