Considering that bias is in the eye of the beholder, a biasless language model is a beholderless language model. The nomenclature is poor, IMO; we should be ta

Considering that bias is in the eye of the beholder, a biasless language model is a beholderless language model. The nomenclature is poor, IMO; we should be talking about bias-aligned models, models that align to our specific sets of biases. That'd be more fair to what's actually happening.