Quite so. In my humble reading of the current publications which declare ‘A.I.’ the large majority are (just) iterative statistics. And since most Bayesian implementations are iterative to start with, they get to declare themselves a “A.I.”. (12,000 years ago, when all computer terminals were green ascii, ‘A.I.’ were all language parsing databases)
Or as one of my colleagues bemoaned recently: “pfff… all A.I. is is just was we used to call curve fitting”
(whutza “fax machine”…?)
