A note on certain implications of clinical artificial intelligences for the field of medico-legal semiotics

Abstract

Artificial intelligence has profound implications for the filed of clinical practices, and also for semiotics and law. In this article, we articulate and explain the different types of Clinical artificial intelligence (CAIs) as their normativity often stems from their type (symbolic or connectionist) (Harnad, 1990), and relative autonomy/agency. Older, symbolic AI, while more explainable, did not offer the potential that offer the current, second generation CAIs. The intelligibility of the reasoning used by CAIs remains largely opaque and generally unintelligible and unexplainable for human interpreters, even sometimes counter-factual (Lee & Topol, 2024). This is also true of the most recent so-called “explainable” AIs, that remains imperfect and only very partially explainable (Reddy, 2022). The most recent literature reveals that the very question of AI explainability continues to be one of the most heavily debated concerning CAIs (Hildt, 2025). In this article, we will reveal that the solution to the black-box problem of CAIs resides in an investigation in the (bio)semiotic nature of both CAIs themselves, but also the problem that surround their explainability. We conclude with solutions to promote transparency in the use of CAIs.

Ähnliche Arbeiten