engineered super-children vs. agi
From the "what in the cyberpunk dystopia am I reading files" comes a new Silicon Valley stroke of lunacy...
When confronted with something profoundly dumb and upsetting, a common refrain from Eastern Europeans translates to "I have no words, only expletives." And dear reader, I have so many fucking expletives on my mind right now as our babbling tech broligarchs decided that the AI Chaos God of their cult must be on the horizon despite having no clue what they're talking about, but I'm going to take a deep breath and explain this latest eruption from our Silicon Valley founts of stupidity as best I can.
You see, the tech oligarchs generally subscribe to the perverse notion that a cyberpunk dystopia is good, actually, and we should all live in one after a guy named Curtis Yarvin plagiarized the works of Neil Stephenson, used them for his onanistic daydreams, then sold them to rich sociopaths who now sing his praises. A pivotal part of their dreams rests on a drastic perversion of a paper by sci-fi author and computer scientist Vernor Vinge.
So far, so weird? Good, because this is where it's going to get way weirder. If you watched the linked short videos, remember that the Singularity scenarios are just possibilities with the least expected one being a super-intelligent computer, or the so-called AGI today's AI salesmen keep hyperventilating about. The fourth and last was genetically engineered super-intelligent humans, also seen as kind of a long shot, but feasible in the early 1990s given our idea of genetics back then.
After an Oxford philosopher decided that AGI was just a matter of time, and a computer scientist who's never been right in his entire life spent a lot of time making wildly incorrect predictions, the tech elite just sort of decided that yeah, AGI is inevitable and with it will come ruin because... it just will, shut up normie. So, to prepare for the inevitable conflict with AGI, they now want to genetically engineer a new sub-species of superhuman geniuses.
I cannot for the love of Cthulhu believe I had to write this as an actual thing I am presenting to you. No, seriously, it's at this point that I had to take a break from this draft and return to it another day because, well, too many expletives and not enough logical thoughts. This is how I imagine NASA scientists feel when they have to read a tract from Flat Earthers.
Plus, here's the thing. They've already tried this and it didn't work. This is because the whole idea of breeding an ubermensch to save the world is based on a guy named Francis Galton not understanding statistics and then turning it into a philosophy for 19th century English racists who thought humanity was going to be wiped out unless they took full control of both society and genes.
Similarly, Silicon Valley is unable to recognize that no, not everything is just yet another machine that can be manipulated and customized at will because they deal with machines that can be manipulated and customized at will all day, so to save humanity from being wiped out, they just need to adjust more machines and we simply need to follow their obvious brilliance, or face extinction as a doomed species that can't rival AGI. Which, again, is surely coming.
Maybe this is why LLMs and AI in general are blowing their minds? They can see the underlying math, they know it's just matrix multiplication, vectors, and some calculus. It should be simple to analyze what they're doing and why, but the scale of the calculations is just mind-boggling. We've never tried these formulas with a quarter trillion parameters before and it's unlocking patterns we didn't know were there or even possible until we tried.
We also can't manipulate every encoding and layer. The model is just too large for that. We'd have to find a way to navigate through billions of vectors, each with up t0 4,096+ dimensions. And so, it seems like magic. Like it's alive. Like it's way too much to control. So, they want to try manipulating something that actually is alive and extremely difficult to control because by its nature, it's so messy, complicated, and in constant flux: human genetics.
It seems like it should be easier. We have 6.3 million base pairs, a mere 3% of ChatGPT's input parameters. There are only 64 codons instead of thousands of numerical dimensions. How had could it possibly be? It's not like genes aren't destiny but a combination of starting points, predispositions, and very general processes, all of them influenced by the environment, genetic detritus of ancient viruses, and subject to mutations and the side-effects of self-correction systems. Oh, no, wait. That's exactly it.
Even if you could custom design a baby for the traits you want, it would be a far worse idea, one that could potentially make humanity more prone to extinction. While nature is trying to pressure the genes of the next generation to survive new pandemics, adapt to new diets, and become more fertile, some VC in the Bay Area is going to be playing whack-a-gene to ensure better scores on some test based on billionaire cyberpunk vibes, and putting their kids' DNA on a blockchain because... reasons, shut up normie.
Look, we're out here trying to figure out how to keep food in the fridge and a roof over our heads while they're debating how to conquer the galaxy with an army of genetically engineered superhuman cyborg children they'll probably abuse and treat like nothing more than cogs in their grand machines – oh, hey, there's a sci-fi novel idea of sorts – and we all just have to pretend this is normal and the future, and if we don't, angry wannabe tech bros will yell at us.
I'm tired folks. I'm so tired. And I think most of us are too. Obviously, this is very dumb and based on a childishly simplistic and incorrect idea of genetics, and a complete misunderstanding of AI under the hood, and it will end in a lot of people who are currently billing themselves as visionary super-geniuses on podcasts made for aspiring hustle hype beasts either pretending they never said anything about it, or eating Costco sized servings of crow.
But the fact that we have to talk about it and that the oligarchs who more or less have our government by the short and curlies are seriously thinking about this as a venture they could and maybe should undertake, just shows how delusional they are, and how little input they get outside of their group chats and echo chambers filled by dependent sycophants who clap like trained seals at every brain fart. And maybe it's time we start calling them out. Forcefully.
