i read the article you mentioned and found this tit bit quite interesting. " In the 2.0 stack, the programming is done by accumulating, massaging and cleaning

i read the article you mentioned and found this tit bit quite interesting. " In the 2.0 stack, the programming is done by accumulating, massaging and cleaning datasets. " there is large scope for error/issues/politics in each of the mentioned steps. but i think it helped me frame my argument more succinctly AI/ML is not software 2.0 because it is not software. all software needs a ietf RFC :)