[Nevermark]

I wonder how good the best language models are at detecting bs.

So many experiments to run...