Francesco = good article and you expressed a lot of the issues and concerns regarding the ever present and future of AI. Thanks! However, like so many writers, you seem to be trapped in the what I call the Deep Learning rut. There are other paradigms being explored presently that could advance AGI tremendously given a small portion of the funding being provided to Deep Learning paradigms that, as you expressed so well, have some major road blocks. Artificial Biological Intelligence uses very little computing power relative to Deep Learning, truly emulates biological processes and gives us the capabilities to do unsupervised learning in domain agnostic environments (=AGI). AI in general is not ultimately a mathematical solution but a biological/physics solution and the sooner technologist and scientist jump on this band wagon, we will see leaps and bounds to where we are today.