I appeciate how Schmidhuber puts focus on non-US-centric history of Deep Learning and modern AI.
He sometimes overdoes these. Even when something is not related to Deep Learning and modern AI. He tries to push very narrow and one-sided views without much room for nuance.
For example- "Turing used his (quite inefficient) model only to rephrase the results of Gödel and Church on the limits of computability."
He is trying to say that Turing did nothing new, his work is a mere extension of Godel's (!), and he only accounted for practical computational reality. That is not true.
He goes on further by saying, "Later, however, the simplicity of these machines made them a convenient tool for theoretical studies of complexity." Like, Turing and Church's work was only good for that. I digress. I believe that Turing's work is more seminal and paved the way for modern general computers.
Another bit here says "The formal models of Gödel (1931-34), Church (1935), Turing (1936), and Post (1936) were theoretical pen & paper constructs that cannot directly serve as a foundation for practical computers."
It is putting the works of Turing, Post, Church, and Godel all together. Like they are the same. I see the logic as they are all impractical, but they are different levels of impractical. Putting them all under one bracket like that makes no sense.
I respect Schmidhuber a lot, but some of his behaviour is borderline childish which are more pronounced in his claims about AI history. You should be skeptical about what he says (and what anyone says).
Despite all this, I did not know many things written in this article. I learned a lot and had fun reading it.
I first read about Zuse in Isaacson's Innovators book. Did not know that he used a high-level language for a chess program. Fascinating. He is very underrated.