The recent ‘social turn’ in the philosophy of science emphasises that science does not proceed in an ethical and political vacuum and that its value presuppositions can be both a blessing and a curse. As blessings, they can constrain science in the service of ideals of justice, wellbeing, and equality. As curses, they can hide in scientific theories and methods, thus piggy-backing on the authority of science, normalising oppression and marginalisation, and manufacturing controversy where there is none.
This project examines the science of intelligence — artificial and otherwise — within this context. Disentangling blessings from curses is difficult but achievable, and is arguably even more important within AI than elsewhere given what is at stake. This project works toward conclusions about what the science of intelligence should study, by what methods it should study it, and — no less importantly — how this research with its dangers and promise should be communicated to the public. Whenever a project is to be judged on its dangers and promises to humanity, we need clarity about what constitutes harm or benefit, and about how to integrate varied and competing values.