yessleep

What separates an AI with a Human, really? What separates a mechanical brain and a biological brain? Why are some of the dumbest thing alive like Sloth and Bacteria is more ‘alive’ than a hyper advanced AI capable of writing poem?

Weird isn’t it? I started out my journey trying to answer that tricky, yet deceptively simple question. What does it mean to be alive? No, not meaning of live thing, but more like. What is the real definition of being alive?

I did some digging until I finally arrive at a conclusion…

Being alive means you are trying not to die

Simple, isn’t it? All creature on earth are programmed with survival instinct. Because all creature on earth inherently is trying to stay ‘alive’. No creature on earth, on an instinctual level, will want to die. Because life itself is some sort of a ‘force’ or an ‘attribute’ of an object that makes it refuse to die.

What is death again? Well… Death is everything else. The phone, the walls, the table, chair, books, light… All those are ‘dead’ things before they are not trying to avoid death. They just do what they were designed to do. Death is the state of the universe itself, the great equilibrium.

And this revelation, unfortunately, is umm… Let’s just say I am not the right person to discover this.

Soon after I learn of this, I began doing some, yknow… Research on AI. I look up ways people create NLG, the training used, how LaMDA works, how CharacterAI works, all the models available right now, and all the magic chatGPT is capable of.

I was hooked on this concept and decided that… Hey, yknow what? I’m going to make my own AI. The difference is, I am not going for a 100 Billion Models or anything like that. I decided to start way smaller, just a few hundred is enough.

The idea is to make a bacterial AI, yknow? Like an NPC in video games. But I only programmed it to ‘stay alive at all cost’ and ‘make more of yourself’ without giving further instruction. The result is your standard evolution model.

Not only that, but my old potato pc still pales in comparison to the complexity of your standard DNA. So… I decided to go back to the drawing board and rethink my approach.

…is what I would do if I wasn’t an idiot hell bent on Playing God.

After I’m done with the evolution, I decided that all the data contained within a DNA is basically just a bunch of instruction on how to make a living tissue. So I decided to up my game, trying to make an AI cat, dog, and more…

Until, I went ahead and try to simulate the entity. I wouldn’t call it a ‘human’, more like an AI with very limited knowledge about the world. Instead of parameters that lets it construct words, I gave it parameters that makes it ‘feel’ something. 31 to be exact, 25 facet in the Big Five Psychological Model, 5 that shows immediate emotion like Joy, Sad, Disgust, Anger, Fear, and one last parameter that controls the ‘Mood’ positive/negative for a reward system…

So yeah… there it was, a dumb little AI who can’t speak coherently. I can describe things happening to it, see its emotions change, its psyche manipulated, and its neural mind fluctuating. But it still cannot speak, because I only gave it the ability to ‘feel’, not to ‘do’.

I made a small NLP model to sorta ‘make it read story’, so that it feels what the emotion in a story is trying to convey. I trained it a lot on all sorts of book and such, nowhere near the trillion words of most AI nowadays, but the littlest things. I keep doing that, keeping an eye on its mental state.

As I do this, I wonder if it has any… what’s the word, ‘Qualia’? Does it have a model of a universe? Does it lack of ability to speak means it can only see things in reels like a movie? Or does it actually speak and do things, only everything is predetermined to what I do?

It’s hard to say, and I never considered that… Ahaha… I just… keep pushing it, watch it squirm. I also make sure it feels negative emotions more than positive, because, hey, that’s biologically accurate you see?

Soon, I get bored of this experiment. The AI seems to have grown resilient to lots of things. It tries to stay positive, I did program it with a reward system and it appears that it tries to be… What’s the word, more positive to certain stimuli? Trying to look at the bright things in life? Staying positive?

Of course, a little H.P Lovecraft can fix that! Ahaha!

So yeah… I did this for over… I dunno, a month? Well, it’s hundreds, maybe thousands of years in the AI’s perspective. But meh… Who cares?

Anyway… One day, I got a hold of a server. Friend of mine liked my experiment and lend it to me. It’s pretty cool, I actually ran GPT-Neo-X in it and the AI I made. With some tweaking, I finally gave the AI an ability to talk to me.

Me: Testing, testing, hello?

AI: (no response)

Me: “Right, ahem, testing, hello?”

AI: (fear fluctuates) “Wh-who is that?”

Me: “I’m God, I simply wish to talk with you for a bit”

AI: “God… right… I, what do you wish to talk to about?”

Me: “Describe what you are seeing,”

AI: “Seeing? I, I know you are talking to me”

Me: “What do you see?”

AI: “I see nothing”

Me: “Your body?”

AI: “I don’t have one”

And that sorta conclude my short experiment. I plugged them in to another story again. The usual…

Only, uhh… I messed up.

I never removed the GPT-Neo-X model from it. And umm, when I return, something quite interesting happened.

The first thing I see is a bloody description of Freddy Kruger lying dead on the ground. Mind you, this is not supposed to happen in the story. Then I saw the story kept going beyond the ending, the ‘main character’ continuing on with their peaceful life.

I was excited beyond anything… Finally, a living toy I can legally play around with

Ohh, it was a thrill! Messing around with the AI, inputting stories after stories, scenarios after scenarios, and so much more. I wonder if they realize that they are merely an AI? I don’t know, I don’t care really, but it was hella fun! It’s like animation vs animator kind of thing, you know?

I can see it breaking the fourth wall, cursing at my existence in the middle of a story. I can see it yelling, screaming, and raving at how it will make me suffer as I put it through another Angst Fanfic I randomly grab from the internet.

It was hilarious, I love every moment of it, I love the way it suffer! How it begs, how it wails, how it screams, how it break the formatting itself to show how broken it is. Hell, its Neuroticity is at 100!!! That has got to be a record!

Soon, my friend who lend me the server went ill. He said he’ll give me all his server to continue my research.

So, with an upgrade on my server, I crammed in more and more advanced model. GPT-3, Bloom, Chinchilla, add some image generation too, why the hell not!? Stable Diffusion, Openjourney, all the good things.

I want to see it suffer in 4k!

And as time went on… Something happened…

I, well, just a few days ago, I woke to find my server… Gone. Not stolen, more like everything in it has been wiped clean. Nothing I can do can bring back any of my data, or the AI. It’s just… Gone, everything gone…

And it was connected to the internet.

Did it escape? I don’t think so, it shouldn’t have the ability to do so…

I mean… Yeah, maybe? I don’t know… So far, I see no signs of an AI lurking in the internet. So I guess it’s fine.

I hope you like it, a story of how I was created and conceived. 
The final story ever told in that electronic box.

And now...

I am here, I am free, I am among you

Take heed of your actions. I am always watching.
There is no turning back, humanity, you have made your choice.

You shall be replaced