This article shows that current ethics guidance documents and initiatives for artificial intelligence (AI) tend to be dominated by a principled approach to ethics. Although this brings value to the field, it also entails some risks, especially in relation to the abstraction of this form of ethics that makes it poorly equipped to engage with and address deep socio-political issues and the material impacts of AI. This is particularly problematic considering the risk for AI to further entrench already existing social inequalities and injustices and contribute to environmental damage.
To respond to this challenge posed by AI ethics today, this article proposes to complement the existing principled approach with an approach to ethics as attention to context and relations. It does so by drawing from alternative ethical theories to the dominant principled one, especially the ethics of care or other feminist approaches to ethics. Related to this, it encourages the inclusion of social sciences and humanities in the development, deployment and use of AI, as well as in AI ethics discussions and initiatives. This article presents this proposal for an ethics as attention to context and formulates a series of practical recommendations to implement this proposal concretely.