Comment by aoeuid

We have some good ideas about how mammalian brains get "aligned". How can we train/scaffold LLMs in a way that we can apply the same approach?