Sampler/teaser/trailer: A few brief excerpts
Part One: The 2000s
p. 16. “In other words, MolBars may become the first electronic technology capable of not only speeding up the network model simulations, but also of reaching the density of biological neural systems, including the human brain, while far exceeding them in speed.”
p. 34. “OK, the networks do not have too much to brag about so far, but at least they can do some pattern classification; what real-world tasks can these cognitive architectures perform?”
p. 43. “Oh Kira,” Kate cut in. “Please don’t start all this silly Terminator talk again. He talks about this a lot,” she complained to LL semi-seriously. “Like, ‘What would Sarah Connor say?’ – oh my God.”
p. 60. “…to reach the main goal of human society as I understand it, to preserve the torch of reason from destruction and carry it on, we need a decisive help from technology – apparently from AI – what else? If we do not get such help, we will destroy our civilization much faster than it would take us to evolve into more reasonable creatures.”
Part Two: The 2010s
p. 69. “Yes Virginia,” he said. “With two synapses per the two-way connection in each neuron pair, this is 2×(64×64)×(1,024×1,024) = 233 – almost ten billion synapses overall; this is why we need supercomputing.”
p. 71. “Now let me show you our last, most surprising, and still unexplained result: some (relatively minor) correlation between the network spikes and the input signal patterns persists even at the times corresponding to the input signal already having been turned off.”
p. 78. “Do you guys understand what this means in even more plain English? If my hypothesis is correct, the biological brain’s thoughts will be extended into the neuromorphic hardware!…If such a system is two-way interfaced to a living brain, it may (just may!) suck in human thoughts, and process them, rather than generate its own crazy ideas.”
p. 87. “The 3D system, with its 64-layer floating-gate matrix and more advanced design rules, has a big advantage in density: it is almost 1012 cells per square centimeter.”
p. 89. “His ultra-dense needle-array silicon implants, with 96 microelectrodes per needle, are very cool: each needle is just 50 microns thick and still carries enough thin-film microwires to pass the signals to or from all of its electrodes!”
p. 96. “I believe that if your two-way interface and our network are both sufficiently large, the patient’s brain will gradually start delegating a part of its internal activity to the network, subconsciously using its internal plasticity for the network’s proper reconfiguration.”
p. 98. “So, if such a ‘beast’, or perhaps a system a few times larger, is connected to a brain via a sufficiently large interface, I do not see why the mind’s move into the network must stay partial – it may become full, including all short-term and long-term memories, together with the self-awareness core, i.e. the consciousness of the connected person. I call this, so far very hypothetical full move of the mind the ‘Transfer’, with a capital T.”
p. 102. “If the mind survives that event (for myself, I call it the ‘Separation’), the ‘transferred person’ might feel sad about the biological part’s death, but would effectively continue to live – in the remaining, artificial component.”
“Do you know what a devout Christian, such as my mother, would say about your idea?” Steve asked. […] “She would say that you want to steal man’s immortal soul, and prevent it from going to Heaven by locking it in a box.
“Yes, but that might be good news for the very many whose souls would immediately go to Hell otherwise,” LL laughed.
Part Three: The 2020s
p. 113. “Hi Steve,” he said, “I have to share some really bad news with you. LL’s cancer returned, and doctors are giving him from six to twelve months to live. They are saying that at this stage, chemo would not help. [He] wants to come to Fort Collins, to be connected to our network and test it himself – before he dies.”
p. 116. LL was mostly confined to his medical chair-bed, wearing an ugly helmet, which not only hid and protected the implanted electrodes, but also incorporated the (still unpowered) electronics for biosignal processing: impedance matching, amplification, and time-division multiplexing.
p. 117. At 8:00 the next morning, Saturday March 22, Steve connected the first four of sixteen implants, with almost a million microelectrodes overall, to the large white box containing the neuromorphic network.
p. 118. “Ha,” LL said, “it is as if something blinked, but now I don’t feel anything. Could you do this again, please?”
“We sure can – and will – and many more times than you may possibly like.”
p. 124. The decisive step of their experiment began on June 14. […] After connecting all remaining implants to the network, they started to activate them gradually, cranking up the gain of interface amplifiers. […] First LL felt like he was falling asleep, but what was supposed to be a usual vague dream rapidly became very unusual and real – as if a huge, strange (“multi-dimensional” as he described it later) wave was trying to sway and swallow him.
Any comments, however candid, would be appreciated;
they may be sent to verahtinych@gmail. com.