Weak-to-Strong Generalization
Alignment TrainingemergingResearch on whether weak supervisors can effectively train stronger AI systems, a core challenge for superalignment.
Key Papers
1
First Proposed: 2023 (Burns et al., OpenAI)
Cluster: Alignment Training
Tags
superalignmentsupervisiongeneralization
Key Papers & Resources1
SEMINAL
Weak-to-Strong Generalization
Burns et al. (OpenAI)2023