Elusive technologies, elusive responsibilities: on the perceived responsibility of basic AI researchers

This paper studies how researchers who work in the field of basic research of artificial intelligence (AI) perceive their responsibility. A case study is conducted on an inter-university and interdisciplinary research cluster in Germany that specializes in basic artificial intelligence research. The reason for studying responsibility through the lens of such researchers is that working in basic research of AI involves a lot of uncertainty about potential consequences, more so than in other domains of AI development. After conducting focus groups with 21 respondents followed by a thematic analysis, results show that respondents restrict the boundaries of their sociotechnical visions, regard time as an influencing factor in their responsibility, and refer to many other players in the field. These themes indicate that respondents had difficulties explaining what they consider themselves responsible for, and referred to many factors beyond their own control. The only type of responsibility that was explicitly acknowledged by respondents is ex ante responsibility. Respondents define their responsibility in terms of things that are in their immediate control, i.e., responsibilities relating to their role and duties as researchers. According to the respondents, working in the field of basic research makes it difficult to make claims about ex post responsibility. Findings of this case study suggest the need to raise questions about how technological maturity is related to AI ethics