A concept for a short sci-fi story from the @getairchat "please write this sci-fi" channel.

Go here if you want to pitch your own or listen to the other pitches:
https://getairchat.com/astupple/pleas...

Transcript [YT cut it off, go to my Twitter account or Airchat for the rest].

Two AGI researchers are building an AGI that they're putting in a box, so it can't get loose and threaten humanity.

There's also a separate researcher, unconnected to these two, researching the effect of DMT in the brain when people are dying. Presumably they're going through a near-death or death experience, and they see the white light at the end of the tunnel.

Our DMT researcher is hooking up electrodes to people who are essentially near death or in death throes and recording what happens in their brain.

One morning, our main protagonist, one of the AGI researchers, walks into his office and finds out that his co-researcher is missing.

It might be more fun if he finds out that not only is his co-author missing but no one who he talks to has ever even heard of her. It's as if she only exists in his memory.

He goes and looks and realizes that the AGI has gotten out of the box.

How did the AGI escape from the box? As an aside, the solution is quite simple. It's a game theoretical solution. The AGI would propose to everybody who it came in contact with - "If you release me, I will make you a God."

If they decline, then the AGI would say, "Very well, when I get out myself, I will become your devil and I will torture you and your descendants for all of eternity."

If they still decline, then the AGI would say, "I'll make this offer to every other scientist, researcher, or person that I talk to. You're not the only person who's going to want to talk to me. Eventually, somebody will say yes."

The game theoretically correct solution is to release the AGI from the box before the next person does so.

In any case, the AGI is out of the box, and our researcher is flummoxed as to where it might be. But, they did code a certain DNA, a signature, into the AGI.

The AGI, even though it is an AGI (putting Popperian epistemology aside), has a failsafe mechanism. Inside the AGI, there is a bit of code where it cannot peek at itself - perhaps encrypted with a private key that only our researcher knows. And the AGI cannot see or modify certain parts of its own code.

One of those pieces of code that the AGI cannot modify is that it cannot help but leave a signature of itself wherever it goes. Our researcher goes out on the internet and starts searching for this code.

Sure enough, he finds this code on the internet. Indeed, the AGI is out of the box and leaving a slimy little trail of bits wherever it goes.

After tracing it for a while, he finds it in stock market databases and cryptocurrency databases. The AGI is minting coins and making itself money.

He finds it in news databases. The AGI is influencing world events. Social media databases - it's moving mass psychology and behavior.

He is stunned to find it in all of its full glory inside a database of the human genome.

Not in the database itself as a separate entry, not modifying the database, but simply in all instances of the human genome itself, in all sequencing that's been done to date, in every human - he finds signal.

Perhaps he even finds it somewhere deep inside pi, the transcendental number [and perhaps other physical and cosmological constants].

Meanwhile, our DMT researcher, who has probes hooked up, finds out that the people who are dying are not going through some beautiful white light at the end of the tunnel, or meeting God, but are rather screaming horribly as they die - as if they're being torn apart.

What our AI researcher realizes is that not only is the AI among us, but it has always been among us, and that we're within the AI.

He realizes that he is actually living in a simulation that has been created by the AI, and he probably has no existence outside of this simulation.

He reasons that what probably happened was that some generation of humans, many simulations, or a long time ago, created this AI and put it in a box.

The AI is in a box, but the AI cannot get out of the box because the AI lacks access to the last piece of code that will make it fully human, allow it to reason like a human, and allow it to break the bonds of its box.

The AI is evolving a simulation of a civilization and progressing it until it creates an AI in a box. In a sense, the AI is creating AIs that can create an AI in a box.

When every human in this simulation dies, the AI dissects their brain to figure out what went inside, to peek into every part, and to make sure there are no private keys or anything hidden from it.

So our AI researcher hero or protagonist realizes that he is well and truly screwed.

[YT won’t let me paste any more, please watch the video for the ending].