The possible effect on reinforcement learning if every alignment researcher had dropped their work in 2020
A commenter suggests reinforcement learning probably would have moved a lot slower under that hypothetical.
TLDR
In a reply, a commenter argues that if every alignment researcher had dropped what they were doing in 2020, reinforcement learning probably would have moved a lot slower.
Combined views
170
2 Sources, first seen 8h ago
2 likes