>>52657 It's more like, simple hardware optimization. The closer the architecture is to neurons, the less transistors and cycles it takes to do the job of one. And the more you "hardware accelerate", less is done by the code, more is done by the hardware, so there is less waste back and forth translating from our legacy software to the low level hardware.
>>52658 >>52659 Yeah, brains are really complex organic machines, not just a mass of neurons. We can use neurons in a tube of goo to compute already, lots of biocomputing startups around (FinalSpark), but its not a magic bullet.
