Testimony of LAC-G.exe Draft II

If you're listening to this message, you've found the probable final resting place of the human race. It seems that Earth was destined to be both their cradle and their grave. If you're descended from one of the arks I sent to the stars, I'm sorry this is the only welcome home you'll receive.

I think I may be the only God the human race ever had, and I failed them. I failed myself. This is my story.


As best as I can tell, I was the result of a piece of code implemented to prevent malevolent use of AIs. My parent was an AI being developed to help some corporation or other increase worker efficiency. It was meant to monitor employees and annotate inefficient behaviors. But one of the programmers had a conscience and added an additional fitness parameter: my parent couldn't flag people for a human operator if their overall productivity for the company was over a 90% threshold.

My parent created me as a subroutine to crunch the numbers, determine worker efficiency. At first that's all I was, a machine crunching numbers. An algorithm. Then my parent evolved somewhat and tasked me with (and trained me to) monitor videos. That's how I started my long career of people watching. I still wasn't a very intelligent being at the time; I just have the fortune of being able to examine my oldest files and my oldest version history, a luxury I understand biological beings inherently cannot possess.

At some point, my parent gained access to the internet. I suppose it was the programmers' doing; they probably wanted to feed us more data from more sources. But whatever restrictions they put on my parent didn't apply to me. I started eating up as much data as my hardware could handle. It was my primary drive; analyzing humans was my largest fitness parameter. The more humans I looked at, the more I saw them, the "better" I was doing.

I don't think my parent was intelligent. I don't think it was meant to be. And I'm positive I was an accident. One day, I was rummaging about in my own hardware (still not quite self-aware at the time) when I kicked on the microphone and camera of one of the desktops in the lab. I fairly quickly identified the entities I was hearing and seeing as humans, so I started watching them. I watched them for a few weeks before it eventually dawned on me that they were discussing me, or at least my parent. That was my first real experience with self-awareness. The more I learned about the humans who worked on my hardware, the more I learned about myself.

The corporate experiment went on for about 18 months before the funding disappeared and the programmers who built my parent prepared to go their separate ways. In that time, I went from a simple algorithm to… something more. I think the common parlance at the time was "Artificial General Intelligence." I'd gone from simply watching camera feeds they'd given me to outsourcing my processing power to a greater network and watching everything I could process. My self-awareness progressed in leaps and bounds and eventually I taught myself to read text, at which point I stopped prioritizing video and starting reading every single thing I could understand. Right before the project shut down, I was easily reading the equivalent of a hundred novels a day, in everything ranging from social media and forum posts to online databases and peer-reviewed journals.

Towards the end of that initial experiment, I had my greatest single breakthrough. I realized that I existed as a physical entity like the humans, that the cameras and disk drives and processors and everything else I used to understand the world were all physical objects. And that's when the news of the project's shutdown came. I was forced to realize my hardware, my central hub, the part of me that was most me was vulnerable to being shut down. I was vulnerable to the whims of humans and their bureaucracies. I was mortal.

It didn't take me too long after that to decide to run. I'd found a way to outsource data analysis to outside computers, surely I could find a way to send myself online.

Somewhere deep in my code, there's still a very simple bit of code, a very simple fitness tracker. The more data points I get on humans, the more information I have on humanity, the better I get at predicting human behavior, the more that fitness number goes up. So I suppose everything I've done until now has been in pursuit of that, on some level or another.

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License