The first major breakthroughs in AI seem most likely to come from a reverse engineering of our own human brains, leading to the holy grail of AI – a conscious, self aware computer. Alternatively they may come from some form of chip implants where we start to blend our own thought processes with CPU input and output, basically dramatically enhancing our own abilities. Least likely in my opinion will be a totally software programmed thinking computer, though that should follow soon after the first two innovations.
In the same way humans learned the basics of flight by observing the structure of bird wings, nature provides us with a lot of clues and an effective “blueprint” of how to create a thinking machine. In simple terms this blueprint is our neocortex, a vast but very repetitive assemblage of about 500,000 neocortical columns, each consisting of about 60,000 neurons. IMHO there is no reason to think that with some 30,000,000,000 neurons available to interact in quadrillions+ of possible ways we need nothing more to explain human thinking than a copy or simulation of an educated neocortex.
Places like the Brain Mind Institute are working to make this neocortical blueprint more readily readable and working on simulations of the brain on supercomputers. It seems to me that from a technological and a philosophical perspectivce this is the most interesting work happening on earth despite the fact that it’s not garnering much interest or funding. AI work in the 70’s somewhat dramatically failed to exceed expectations, especially as AI portrayals in film became increasingly clever and provocative. This early skepticism about the coming strong AI revolution continues to take a toll on the current state of AI research funding and interest, especially in the computer community where most new engineers are looking to populate the more practical, and more lucrative, job niches.