definition
Explanations as relaxed deductive arguments
ARC's research seeks a formal notion of 'explanation' for a model's behavior as a deductive argument from the weights — proceeding step by step from properties of the network (e.g. two vectors have large inner product, so activations correlate) to conclusions about outputs — but with the rigor standards of proof relaxed, since proofs are too restrictive to apply to interesting neural nets while still yielding a structural causal explanation.
definitionpending
Speaker
Paul ChristianoEvidence Quote
“the thing about proofs that is relevant for our purposes isn't that they give you 100% confidence... You can relax the standards of proof a lot and still get this feature where it's like a structural explanation for the behavior”
Created: 6/13/2026, 3:36:05 AM
Your notes
Loading notes...