In the 1980s, Judea Pearl developed and popularized Bayesian networks: directed graphs that factorize probability distributions and allow efficient inference under certain structures.
The approach gave AI a clean language for uncertainty—medical diagnosis, vision, sensor fusion—more manageable than many ad hoc heuristics.
Pearl later led the study of causality (do-calculus). In 2011 he received the Turing Award. His agenda still tensions deep learning: correlation is not intervention.
