Weak-to-strong generalization

Company news OpenAI Source language: English

We present a new research direction for superalignment, together with promising initial results: can we leverage the generalization properties of deep learning to control strong models with weak supervisors?

Excerpt

We present a new research direction for superalignment, together with promising initial results: can we leverage the generalization properties of deep learning to control strong models with weak supervisors?

Short excerpt from the source — open the original for the full text.

Sources

← Back to the daily brief