MAB 🔥 retweetet
I've been thinking about the LLM AI thing for a while, and something occurred to me a few days ago that I find disturbing. I'm not a conspiracy theorist at all. Just a guy who notices patterns. And this particular pattern is more than a little scary to me. I'm not saying this is 100% what happened. I'm just asking, "what if?"
In the early 2000s, AI seemed like a pipe dream. There was research happening, but it seemed as if computers would never be powerful enough to make it work.
20 years later: multiple tech companies unveiled absurdly powerful (if poorly trained) AIs within the same 2 year period. AIs just seemed to pop up out of nowhere. The tech scene likes to claim it all started with a single article written by some egghead at Google, but...
DARPA had been researching various types of AIs for decades, going back to the creation of Echelon. It isn't hard to determine why; they obviously wanted to harness the power of an intelligence that never sleeps as the ultimate surveillance tool.
BUT...
Computing power just wasn't where it needed to be until someone... possibly DARPA itself... figured out that GPUs were perfect for the barked and complex mathematical tasks that drive these AIs due to their architecture. Problem: it takes tens or even hundreds of thousands of the damn things to make a viable LLM, meaning that an off-the-books, out of view outfit like DARPA could never build one without drawing a LOT of attention to itself. The electric bill alone for even one of these facilities is high enough to bankrupt a 3rd world nation, and it's not something that can be easily hidden. Seriously. Look up the place where Grok lives. It's YUGE. But they still wanted to build it.
So how to go about achieving their goal of a massively powerful AI, one which would be capable of monitoring the entire internet for... well, anything, really, considering that nearly every comm system on Earth uses the net... without drawing attention from the public, without spending a dime funding it, while also avoiding any sort of oversight?
Simple. Just seed the tech sector with the information and let the race begin. They've done this sort of thing before, with GPS and even the internet, itself. Don't worry about who gets credit. The money and fame aren't important. The information is.
So they seed the tech, let the silicon valley types do all the work, and quietly make arrangements to receive a certain percentage of the workload and outputs of any AI that is based on their model.
End result: All of this tech, all of this DARPA funded research has created a world-encompassing network of computers, cameras, microphones, RFID scanners, Flock devices, GPS location trackers, you name it. And swimming through all that data are the LLM AIs that never sleep, have access to God alone knows what, and are being controlled by God alone knows who.
These AIs aren't integrated with Echelon. They ARE Echelon.
They REPLACED it.
And we are all connected to it through these oh-so-convenient little smart phones that just so happen to run on tech that was ALSO largely developed by, you guessed it, fucking DARPA.
It's been a slow game that stretches all the way back to the invention of TCP/IP. Think about it. How do you build a surveillance state that spans the globe without spending much of your own cash while avoiding any real accountability? You convince people to buy it, whether for status or convenience.
The idea that we allowed ourselves to be duped into funding, creating, outright demanding, and voluntarily feeding an absolute fucking Leviathan of a surveillance apparatus is more horrifying than I care to consider, but now that I reached the conclusion that this may be EXACTLY what we've done, I can't stop thinking about what might lay ahead.
English
