We’ve talked a lot about the many strides made in electronic computing during WWII, and we’ve talked a fair amount about the new kinds of computers invented during the War such as Turing’s “The Bombe” and the early stages of ENIAC. However, we’ve sometimes presented artificial intelligence applications of these computers as being way off in the distance of inventors’ minds at the time, and this is not the case at all. The Homeostat System, which began development during WWII, was in many ways the first fully-functioning computerized neural network. Dr. Ross Ashby, a professional psychiatrist, actually led the push to develop the Homeostat prior to the War, but he didn’t have much support or an engineering team willing to work on a project that sounded like a novelty. After all, given how limited computers were at the time, what possible use would an AI system have? During the War, the need to get bomb control units to adapt to changing conditions created a need for an AI that was actually feasible with the limited computing power at the time.
In my opinion, the design of Ashby’s Homeostat Machine was not a precursor to the design of most modern day based neural networks because the invention of the perceptron neural network in the late-1950’s radically changed how AI would be designed. However, in terms of showing that the computers of day were capable of serving as practical artificial intelligence for important tasks, the Homeostat stands out as one of the most important inspirations.
https://blogs.bl.uk/science/2016/04/the-thinking-machine.html