beren comments on The case for removing alignment and ML research from the training dataset

[beren]

Yes! This would be valuable. Generally, getting a sense of the ‘self-awareness’ of a model in terms of how much it knows about itself would be a valuable thing to start testing for.