Navigating many voices: lessons for ELSA/RRI from the pragmatist perspective of ELSA-by-design

The societal uptake of AI has led to a proliferation of AI ethics guidelines that prescribe universal principles for ethical AI. However, these guidelines face limitations as the abstract nature of principles makes it difficult to translate them into practice and risks interpretations that allow for ethics-washing. Context sensitive frameworks, such as ELSA, RRI and VSD, are more attuned to the context-specificity required in AI development. However, these frameworks have faced limited uptake. Building on the tradition of these frameworks, we are developing a pragmatist approach named ELSA-by-design. We reflect on the implementation of this approach in the ELSA Lab Poverty & Debt, discussing the challenges we have faced and highlighting the opportunities of an iterative, experimental approach to ethics.