Google DeepMind withholds reward optimization from internal chain-of-thought tokens to keep the reasoning process unfiltered for safety monitoring
Agree
Disagree
Overview
Debates
Related claims
Topics
B
b5a31f81
Topic
AI safety & social impact
1
debate
·
2,589
claims
·
310
news stories
0
B
b5a31f81
Topic
Chain of thought
71
claims
0
B
b5a31f81
Topic
AI interpretability
50
claims
·
5
news stories
0
Opening sign-in…