Caleb Biddulph comments on CoT May Be Highly Informative Despite “Unfaithfulness” [METR]

[Caleb Biddulph]

Many of our findings are similar to what Google DeepMind found in their investigation of faithfulness in CoT reasoning.

This links to an Anthropic paper. Did you mean to link this GDM paper?

[GradientDissenter]

Oh oops! Yes I did. I’ll fix that.