AI Weekly: Can language models learn morality?

Researchers believe representations could imbue language models with a broader set of human preferences about the world, and thus morality.