The intelligent device will be a piece of software running on a serial computer.
This assumption is entirely pragmatic in nature.
Many have observed that brains are massively parallel in nature. But no-one is going to fund the development of a massively parallel computer at a massive cost on a justification consisting of "if only we had a massively parallel computer". Even if I were able to prove the necessity of a massively parallel computer, I do not have the intellectual standing to have anyone important read the proof, let along build the machine.
Penrose has argued that no computation using classic physics will suffice, and that exotic quantum effects are required to explain intelligence. Penrose may be right, but if he is then I cannot participate and it is unlikely to occur in my lifetime.
Perhaps conveniently, I happen to believe that neither a massively parrallel computer or an exotic quantum effect will be required to construct an intelligent device. I would also like to believe that we have ample processing power at our fingertips already to construct an impressively intelligent device. I have no adequate justification for these beliefs other than the only possibility for my success lies in them.
Page updated