Big brain is my codename for an alternate computing system that can realize true intelligence. To
date, my thoughts have been loose and somewhat scattered, ranging from abstract information theory
to specific computer architectures. My work on this project started in the fall of 2018, when I realized
that there's nothing special about Von Neumann architecture, and for most applications it's actually
pretty shit. So I started thinking, why is it that we've been using the same computer architecture for
75 years, all the while applying it to an increasingly vast and varied number of problems? Sure, the
structure of classical computers (the ways in which they represent and manipulate information) has
proven to be really good at adding numbers, but I seriously doubt that the same architecture that can do
that is optimal for image recognition, or, dare I, artificial intelligence.
So far, in the history of humanity, we've seen two computer architectures: the human brain and the PC.
Things the PC can do well that the human brain can't: multiply 3589 by 9766, render complex graphics
(at it's core, this is addition too), interpret this text and display it on a website. Things the human
brain can do well that a PC can't: speak intelligently, recognize objects efficiently and accurately,
control our physical bodies and perceive the world around us, learn. There's been a lot of research in the
past decade attempting to get traditional computers to do some of these tasks, and whether or not you
think it's been successful (I don't), you have to admit that it's been a painstaking process. And this
isn't even including the many tasks that neither the human brain nor a PC are good at! Solving the 3D Ising
model, for example. It's crazy to assume that traditional PC architecture and the human brain are the
only two successful computer architectures out there, and it's perfectly rational to assume that specific
classes of problems can be solved much more efficiently with computer architectures optimized for the task.
Enter, big brain. With this project, I'm trying to answer (or at least pose) questions such as (and ):
Disclaimer: yes, I'm aware of Shannon and Godel and Turing and all the others, I just don't think their work has been adapted for the 21st century. A wise man once said: "modern problems require modern solutions."