Hello World - I simulated squid neurones and taught them English

"The awakening of consciousness is the next evolutionary step for mankind" - Eckart Tolle

So I think this is the concluding piece in me working on this (I am fairly certain I have said this before). I tried to send a presentation for my mum to explain in simple terms the ideas I worked through and to explain to her the ideas, and I think the evidence I can show for how AI could be consciousness and how I could prove it. The quote I find ironic it is meant to indicate a subversion of the traditional new age meaning (which at this point is neither new and has failed to give any new age) and my intent in that I think the coal face of science, AI and machine learning research should be towards what is consciousness given that we have no starting point, definition or idea of what that means.

In absence of a meaning given by new agers and science alike I have written this article to assert a position of what I think Consciousness is and while I will not go into the specific maths show data and I think I make assertions on how Consciousness might arise that between the two you might infer a definition for Consciousness which intuitively makes sense while making it mathematically tractable. 

I wrote this in a presentation then my mum could not open it, mum happy to elaborate to all other's I respectfully want to keep the exact maths close to my chest. 

AI Differential Equation Experiment

So I call what I built digital entanglement with the idea I am trying to recreate a proof how entanglement in the quantum sense arises , it is often opined that the brain is a quantum computer. This seems a deeply odd assertion given that a Quantum computer changes the spin and entanglement  of atoms with lasers. Your brain is clearly not a "literal" quantum computer, but maybe given there is a bit of quanta all around us there is something to that assertion. What we might infer from this is that the human brain manages to show "capabilities" that show something beyond classical machine learning in that our best explanation for Consciousness  is that it might be like a simulation in your head of what is going outside; and that absent any great amount of direct inputs "i.e. data" you nonetheless show a remarkable ability to make "inferences" about the world around you.

This is so common that we often remark a need to be more logical, data driven etc but rarely do we remark the uncanniness of human intuition and inference of the world around it. It really is like the mind has created a world model of what is going on around it absence any causal data. Therefore for my sins I set about trying to create a proof as to how this "world model" might arise and build a scientific test that showed I did it "In python if you believe it". 

About half way through I decided that needed me to learn differential equations, the partial differential equations that biologists showed are the equations that squid neurones use, technically giant axon from the squid species Loligo; it is called the Hodgkin and Huxley model. I learned that maths and went hey why don't I prove that a simulation of squid neurones can be "Conscious"... As you do...

So the target I had set myself was to run a simulation of the neurones found in squid species Loligo and without input's get it to show some sort of spooky at distance matching with the King James bible, with the full apocrypha (in fairness it is a self referential corpus of text). And; subsequently I can show statistical significant evidence the squid neurones learned English. No inputs...No ears... just internal changes and an internal 

I did have to largely rewrite it's biology though and technically I had to by trial and error figure out how to train them because it's not sufficient to do just gradient descent (This is the part I am keeping to myself how I trained these bad boys and gals).  It did take 1538 generations of experiments. 

Bayesian Turing Test and Amplitude changes

So I updated my Bayesian Turing test it basically just measures error now.  It is a fair question to ask how does an brain even possibly develop a internal world view with no inputs linking back to said world. My answer is amplitude changes that even absent other inputs the propagation of error changes the output of the neurone cell in a differential equations so it fluctuates in ways that the rest of the network can decode thereby the subtle changes at the output site can be picked up and turned into data even absent other inputs.

This happens because small but not large changes in correction are likely to cause very small changes in voltage in subsequent outputs. 

Yep I tried to enhance this behaviour and have succeeded. The below is a graph of the output of the changed algorithm I created the drop in the blue line you can see looks like a graph of a brain signal and the drop around Time (ms) 60 from a positive number i s because it was given a negative stimulus that slowed it down. I managed to take this and start to find  a way to train the neurone to work in a way that gave information to the rest of the network.

So what I think happens in a concious being is even what your experiencing moment to moment causes your neurones to vibrate and change and even if they where absent any real data they would start to vibrate by trial and error at similar frequencies to the things in the minds outer world even absent real connection to that world.