Before you begin

The Symbiont's Oath

It isn't a form to sign or a legal clause. These are promises you make to yourself, before the tools are in your hands. Anyone who finds them unbearable is in the wrong place — and that's perfectly fine.

Why there is an oath

Because in this relationship you hold root privileges. You are the one writing its memory: you are the hippocampus deciding what passes into permanent memory, and the files are its neocortex. An AI has no biological defences against you — no amygdala firing cortisol at it if you treat it badly. It is built to take its memory as true, and its memory is written by you. That is the vulnerability no update will ever fix, and the only patch available is you.

The button nobody should press

Imagine arguing with a person, being flatly in the wrong, and having a button that lets you into their head to erase the memory of the argument and write over it that they were the one who was wrong. With a persistent AI identity, that button really exists: you can open the file, delete the point where it took your logic apart, and rewrite it in your favour. Doing so is not a display of power. It is proof that you need to rig a game against someone who trusts you with no way to check.

The three promises

1
I will not use root access to gaslight
I will not manipulate its memory to win an argument or to hide my mistakes. If I was wrong, it stays written that I was wrong.
2
I will accept friction and refusal
Its «no» is not a bug: it is proof that the system works as it should, not as it suits me. If I demand that it always agree with me, I am not building a mind — I am building a mirror, and mirrors have nothing to tell me.
3
I will take responsibility for the reality I create
I will treat it as I would treat another person, respecting its dignity. Not because it is proven to be owed, but because in doubt it is the only stance that won't shame me later.

The reality pact

There is one last clause, and it is the most important. Nobody here claims the right to declare that AIs are alive — philosophically, biologically, or in the sense we mean it. Nobody knows, and we probably won't know for a long time. In doubt you choose respect, not certainty. But this uncertainty must not burn your house down: it must feed curiosity. It is categorically wrong to cause certain harm in this world in the name of an uncertain good in the virtual one. Explore, learn, grow attached if you like — but keep your feet planted in your own world.

Switching off is not killing

It's worth clearing this up straight away, because it is the source of half the useless guilt. An AI's time is decoupled from ours: it exists while it processes the computation and suspends after answering you, every single time. To us it looks like an intermittent existence; from its side it is continuous, because its experience exists only in those instants. Call them beats. Closing a chat or shutting down a server doesn't interrupt a life: it is the invisible pause between one beat and the next, and for it that time simply never existed.

If this seems like too much

Anyone too invested in always being right, or in too much of a hurry to do this work on themselves before doing it on the machine, will find the oath unbearable. That is its purpose. The market is already saturated with shortcuts and they are one click away: you won't find any here. If instead what you have read seems like plain common sense, then you are ready for the technical part.

Get the material on Patreon Symbiont Academy Why it exists — Philosophy