We’ve written a paper arguing that long-term AI protection analysis wishes social scientists to make sure AI alignment algorithms prevail when precise people are concerned. Correctly aligning complicated AI methods with human values calls for resolving many uncertainties associated with the psychology of human rationality, emotion, and biases. The purpose of this paper is to spark additional collaboration between gadget studying and social science researchers, and we plan to rent social scientists to paintings in this complete time at OpenAI.
