The contemporary discourse on artificial intelligence (AI) in scientific research is oftentimes driven by the assumption that larger datasets yield better models, and that such models inherently improve scientific understanding or technological capacity. In practice, however, the pursuit of data – especially sensitive human data in the field of healthcare – reveals fundamental ethical and governance tensions. Below we outline three core problems that require rigorous analysis: data scarcity, legitimacy, and the semantic inflation of “AI” as a label that obscures distinct technologies and ethical concerns.
Khutsishvili et al. (Fri,) studied this question.