For the most part, approaches to AI ethics have considered abstract ethical notions, such as whether AI systems are trustworthy, sentient or have agency.