The scientific question addressed here is a question of values. Decent AI comes with quite some requirements, e.g., being fair, accountable, transparent, frugal, certified, privacy-preserving. These requirements are increasingly better formalized, and adopted worldwide with the aim of avoiding and preventing AI-based dystopia.
These directions have been investigated along the past PhDs of D. Kalainathan (causal modelling), R. Bresson (explanability), J. Girard-Satabin (certification); and the on-going PhDs of A. Lacombe (causal modelling), M. Verbockhaven (frugality, submitted); and the post-docs of K. Gasnikova’s, S. Jetley’s and S. Dong (all addressing causal modelling).
… more to come
