Data quality and privacy concerns in digital trace data

摘要

Machine learning (ML) represents a key link to both data analytics and human–robot interaction. ML in society is a topic of central relevance to computational social science because ML changes both the object and methods of social science. This chapter questions the view that this change implies a shift from error-prone survey data to error-free behavioral data. It draws attention to a possible distorting factor in the latter case, the use of techniques that guard Internet users against being tracked when surfing the web. Data from a local population survey are used to explore supposed sources of variation of such techniques in practice. The analysis reveals three such sources: open-mindedness towards technical innovation, comparative risk perception, and the felt ease with human–robot interaction scenarios and related attitudes towards robots and AI. The chapter uses data from two surveys of the Bremen AI Delphi study on the social and ethical acceptance of robots and artificial intelligence.