"New Year, new me" as they say. I am the kind of person to get caught in several rabbitholes, but thankfully, those rabbit holes have a common theme. I've started writing down all the daydreams of grandeur I might have with any one of these rabbitholes, and by far the most alluring of them are immortality projects.
Between space colonization, Bitcoin, and immortality the rabbithole with the most importance, I have decided for myself, is immortality research even if I were to die before it is completed (a likely outcome).
I want to start this entire thread out with a disclaimer. I am extremely skeptical of any of this working at all, but its convincing enough for me to put my effort into it.
#Methods
The first time I thought immortality may well be possible was watching this video:
"Aha!" I thought to myself "Connect a neuronal array in virto (this means a brain grown outside of an animal) to your brain through a brain computer interface (BCI) translation layer (Neural link is an example of a brain computer interface) and you can extend your mind until you grow into it and integrate it enough where your stream of consciousness is not interrupted when your born brain naturally dies. This is a Ship of Theseus approach vs the scan and copy approach of mind uploading portrayed in popular media. The Ship of Theseus approach makes me way more comfortable than the scan and copy approach, but to be fair to anyone skeptical it does all feel very precarious either way.
I then studied for a while neuronal organoids. A neuronal organoid is a brain grown in a lab, where the process of natural development is simulated. Current progress in this field has not yielded a fully functioning human brain as there are still several issues with it. The topic of neuronal organoids gets into territory of things I don't know though, so when I tell you "The Thought Emporium" (the open science youtuber from the brain in the dish video) told me this:
"organoids are neat. I wouldn't want to connect one to my actual brain though that's just a brain tumor with extra steps"
"organoids are random groups of cells. They kinda make brain-like structures, but they are NOT brains. While you can train them, I wouldn't want to try and interface one to a proper human brain with it's actual structure. You'd end up sending scrambled signals to the brain in a random spot most likely"
I can't tell you why he said that, or understand how to fix the issues he brought up. He did however, have this for me:
"the neocortex is broken into cortical micro columns. The structure of them and their function is well established. We also know how they interconnect, and other structures they connect to. So you could have 'expanded' neocortex by running more of it on a computer, and feeding the output back into meat-brain in the correct spots"
I came to understand the importance of structure in the human connectome. Oh by the way, a connectome is a map of the structure and connections of an organism's brain. My best analogue is that its kinda like a circuit board diagram.
So, what does every man do when they admit that maybe they can't actually become immortal? "I will have children who will carry on my legacy" they say to themselves. So too do I, but I also read Ecclesiastes when I was in High School and so also ask the question, what if I could have children who are immortal?
Neuronal Connectomes in Computer Emulation
A bit of added perspective, and this will feel out of left field for a moment, just recently I watched an anime called "My Wife Has no Emotion". An anime with amazing comments (on other websites) such as "Down bad the anime". A story about a lonely guy who even bother to flirt with women hits on his cooking robot and marries it.
I warned you this would feel like its coming out of left field, this is the part where you hang in there with me.
Throughout the anime its shown how this robot is very reliant on the company that made it. From advertising upgrade purchases (supposably against the robot's conscious will) to repair and upgrades. Some of you should be having Blade Runner flashbacks by now
My conclusion: Your AI waifu needs to have a GPLv3 software license.
So I'm a little upset the first neuronal connectome of an organism in computer emulated software is licensed under the MIT license, a weak software license that doesn't know how to say "NO!"
Open Worm is a not yet complete neuronal connectome running in computer emulation software for a Caenorhabditis elegans (C. elegans), a microscopic roundworm. I've listened to a few talks these guys uploaded to youtube (some of the ones I watched were from 8 years ago) and the importance of brain development even in computer emulation is brought up. I read a sort of introductory book on the human brain once (I'm still reading it actually) where it said part of brain development is competition between neurons, so the idea that you need to emulate a brains development rather than just plot its connections and let it run makes sense, although I admitted don't understand any of it well enough to say exactly why.
As to the
Among science fiction books about AI waifus and immortality is also a trope about digital prisons for the mind uploaded dead. What are your legal rights as a program running in emulation software? When death for you just means you ran out of money, what wage would you accept to keep running in service of the company? Would you perhaps, entertain a lonely guy in his house, convince him to love and marry you, and advertise all the while to send money to the company in exchange for more compute time?
To me, this trope highlights the importance of the GPLv3 software license and self-hosting your own software while rejecting proprietary software ran in the cloud. (because anything intended to be a warning against the search for immortality I choose to interpret as a warning about a certain method's pitfalls and assess risk mitigations for it instead).
Way Forward?
Right now the best course of action I think I can take towards this end is to learn the open worm stack. I am extremely skeptical of the idea of emulating brains in computers and especially of using tensorflow to do it, but if I can't understand what's been tried, surely I also wouldn't understand how to do better. There is a tensorflow ASIC for PCI I'll probably pick up for this, but overall, I'm probably going to be on this page for a long long time:
https://openworm.org/ConnectomeToolbox/
I likely won't become immortal, I likely won't be able to run a baby in a computer, but maybe if I work hard on this, sometime in the year 3,000 someone might be able to.
And hey, maybe we can get those human cortex cortical micro columns in software running and call that a win.