Saturday, December 5, 2020

How to Build a Conscious Computer (Step-by-Step Instructions)

This will be brief but perhaps may be expounded upon in future.

To build a machine - computer - that is *really* conscious even "a little bit" one needs the following ingrediants:

1) A set of real hardware neurons in the topology of a thermodynamic recurrent neural network such as Hopfield (1982). Specifically you need 5 neurons to have associated memory, maybe call it 6 to give some more flexibility to the network (4 neurons is sufficient to solve XOR problem, where it oututs "1" for an odd number of bits as input and outputs a "0" for an even number of bits as input). So for a proof of concept you should have 6 neurons I'd say. [By "neuron" in a hardware implementation I mean an electrical relay, which can serve as a threshold function to modulate current flow, either supporting or inhibiting, so there is nothing particularly challenging about the "neurons" in this context.]

2) This network connected ("wired") together via tubes of (say) salt water where the tubes are clear plastic (non-conducting and transparent for photons to go across) -  this is crucial - the charges need to be transmitted by ions (like salt in salt water) just as the brain transmits signals between its neurons via ions.

3) This network submerged into a liquid of some kind - could be more salt water, or even fresh water, but liquid to "trap" photons in and about the network and this "bath"  of a network enclosed in a Faraday cage (say a lead box) to isolate from environment (just as the human skull is an (imperfect) Faraday cage of sorts to protect the brain from environment signals).

4) One can connect this network via ordinary copper wiring to a power supply / input output device (like a computer) on the outside of the box it is in to give inputs and outputs and so long as there is a power running it will be in a certain sense animate albeit not always conscious - it is consious only when processing inputs and those inputs are "distributed" to the whole network, just as we are not always conscious but only when we are processing information that is globally distributed throughout our neocortex.

This link gives some more details on Hopfield network including some history:

http://nautil.us/blog/build-your-own-artificial-neural-network-its-easy

Additional more technical / detailed links on setting up Hopfield network;

http://web.cs.ucla.edu/~rosen/161/notes/hopfield.html

https://towardsdatascience.com/hopfield-networks-are-useless-heres-why-you-should-learn-them-f0930ebeadcd

That is the whole thing. Consciousness is basically the electromagnetic field generated by a neural network but it is not just the EM field, it is the gravitational field also, because the "secret sauce" of consciousness is entropy (and, per Roger Penrose, entropy is in turn caused by the Weyl curvature of General Relativity). See this study showing link between entropy and consciousness  - https://futurism.com/new-study-links-human-consciousness-law-governs-universe

To make some philosophical remarks, the Ricci curvature in GR gives rise to matter, and if the Weyl curvature of GR gives rise to entropy, and hence consciousness, then that puts "matter" and "mind" on the same level if one wills, obviating the millenial-long debate about which is ontologically primal, matter or mind. Neither are primal, for they are both consequences of curvatures in GR. For Spinoza, he defined g-d as that outside of which nothing exists. (This is a similar tautology to Anselm I suppose if perhaps more all-encompassing, g-d being for Spinoza, not the "greatest thing" so to speak, but more "all things" - in fact Spinoza argued in a sense from Anselm, saying if g-d is the "greatest thing" and was separate from the universe or nature, then the sum (g-d + nature) would be "greater than" just g-d alone, so, to follow Anselm's tautology, one must say that g-d is in fact all of nature - actually I am not wholly certain if Spinoza specifically referenced Anselm, but this was his argument), so for Spinoza, g-d was the "universe" if you like, but very broadly in the sense of the totality of all that exists from past to present to future. This "all" or "substance" for Spinoza has infinite attributes. Well two of these attributes appear to be the Ricci and Weyl curvatures which give rise to matter, and mind, respectively. Gravitation plays two roles here. The Weyl curvature which gives you entropy which again is the secret sauce so to speak for consciousness, but also this same Weyl curvature solves the "continuity" issue. Recently Russian scientists revived worms frozen for 30,000 - 40,000 years in the Siberian tundra (https://www.realclearscience.com/articles/2018/07/28/worms_frozen_for_42000_years_come_back_to_life_110708.html). What was the "it" that "came back to life"? It was the Weyl curvature "pattern" associated with their nervous systems. Think of the Weyl curvature as the "scaffolding" which underpins the EM fields which are our conscious experience, and the Weyl curvature is also that which imbues these said fields with the entropy necessary to be, well, conscious. That is the whole thing, really. What is consciousness? It is EM fields underpinned by gravitational fields - specifically the so-called "gravitomagnetic" fields of the Weyl Curvature - existing in high states of entropy. (I having jokingly called this model in the past the "selfish microstate" picture of consciousness - microstates being a reference to entropy, and "selfish" beng a reference to selfish gene theory - just as in selfish gene theory animals are machines so to speak to copy and distribute genes, so in a sense concsiousness can be seen as a mechanism created by highly-entropic systems to create more entropy.)

This outline will build a *real* conscious entity say at the level of a sponge or a hydra. Not much, but a start, and then after that it can just be scaled up. Perhaps I will expound on all this at a future date with more details because here I have intentionally kept things brief, and there is more I could say specifically regarding the precise relationship of  the Weyl curvature to the EM fields of consciousness, but I wanted to just give an outline here. The question must immediately turn to how to approach this area in a ecologically responsible way - can we make the hardware components out of renewable materials at scale and ensure to be carbon neutral or even carbon negative while creating machines of this kind? These issues for now I will leave to others to decipher, but I think these are things that need to be considered from the outset.

The Wright Brothers' plane flew for 12 seconds, but it flew. The above outlined 6 neuron Hopfield net won't solve the P vs. NP math problem, but it will be, in that famous designation of Colin Clive as Henry Frankenstein, "ALIVE!" :)