Logan Riggs comments on Using GPT-N to Solve Interpretability of Neural Networks: A Research Agenda

[Logan Riggs]

I also expect it to be harder as well, and 10-30% chance that it will require some new insight seems reasonable.