Positronic Brain Project

Warning: I'm don't know if someone already had this idea, i couldn't find it on google, maybe i fail to find the right keywords for this.
Studying the anatomy of a neuron you can realize that, it's just a cell that receives signals throught its many dendrites, receiving each from another neuron, and the neuron outputs signals throught its axon, that may lead to many others neurons... But it may be reversed in some cases, but overall it's that...
Then, we have some rules, a neuron only retransmit a signal received when it reaches some pre-assigned level of power, and the output is always the same... And it has to wait a few time to transmit again...
So it's relatively simple the operation of one... Not really, but let's believe that it is.... :P

The positronic brain will not have any positron as it could result in a dangerous thing, as positron + eletron aniquilate each other...
The idea is to build something like an transistor, but shaped like a neuron, with a small processing center, with a few single rules programmed, and a lot of input and output cables covered with carbon nanotubes, or something smaller....
Each cable would be randomly connected to another, let's call it neuronit (neuron + bit) or neuronites...
The size of a neuronite really matters, as the smaller it is, the more fits in a small space, like a head...
The size of cables may vary a lot, but each neuronite needs at least a thousand of those, as some may just fail to find a connection.... The end of the cable need to be magnetic, to connect automaticly to the axon that is a cable that it's not fully shielded with carbon nanotubes...

After building a bunch of neuronites, you just drop them over an recipient, and drop some kind of eletrical shielded fluid, and them start it...
Starting teaching....
This kind of assembly process assures that your positronic brain is unique, as you can't tell which connections had success, and which hadn't....
The only failure in this project is that not every brain will work properly as it can have lesser connections resulting in lower IQ...
My calculations says that it will take at least 25 years to build an operational model, as well to develop the technology that will allow the build of neuronites small enough to build self-aware robots... I'm just not sure about energy consumption, as it would require a small nuclear reactor, but it will not care in the future as it we find out how to generate energy like plants do, carbonic dioxid is free in the air....