How to Ask the ‘Right’ Questions about Artificial Intelligence in Social Sciences? Human-Centered AI as a Problem and as a Solution

Abstract

In 1985, an STS scholar, Steve Woolgar, published a paper, “Why not a Sociology of Machines? The Case of Sociology and Artificial Intelligence”. Yet, there is no solid sociological (social sciences in general) tradition of studying AI phenomena up to now. Why? Part of the answer: Woolgar’s question – are artificially intelligent machines sufficiently like humans to be treated as the subject of sociological inquiry? – directed scholars in the wrong way. In this paper, we demonstrate why this direction is wrong and which one should be taken instead. We begin with a rather short discussion of the “artificial intelligence’ (AI) definition; then, after a brief sketch of the human-centered system strategy, we delineate five types and three levels of social sciences research on AI. After that, we characterize our understanding of the problem of AI and Ethics. Further, we claim that HCAI is a united idea for the AI community, including social scientists who study machines and algorithms in society. In conclusion, we summarize our argument on why Woolgar’s question is ‘wrong’ and what the ‘right’ questions are for AI research.