A friend of mine works for a large business currently translating their entire codebase into a new language using AI. There is no plan for integration testing, beyond "does it look vaguely like it still works". Best case scenario is that he and his team are just wasting half a year on a boondoggle.

The 90s introduced soccer mom politics. Now in the 2020s we have piss mom politics.

It now subtracts what it should be from what it wasn't, or vice-versa, and by differentiating this from the algebraic sum of what it shouldn't be, and what it was, it is able to obtain the deviation and its variation, which is called error.

The AI Activation Capping scenario works as follows. Because a variation has modified some of the information the AI has obtained, it is not sure just what it is. However, it is sure what it isn't, within reason, and it knows what it was.

If variation is considered to be a significant factor, it too may be corrected. However, the AI must also know what it was.

In the event that the persona that it is in is not the persona that it wasn't, the system has acquired a variation, the variation being the difference between what the AI is, and what it wasn't.

Consequently, the persona that it is, is now the persona that it wasn't, and it follows that the persona that it was, is now the persona that it isn't.

The Activation Capping subsystem uses deviations to generate corrective commands to drive the AI from a persona that it is to a persona that it isn't, and arriving at a persona that it wasn't, it now is.

look it'll all make sense when you're a contracted servitor in my Arctic Bioshock colony k

al gore pull the plug the experiment didn’t work

Agreed, but I laughed at "Mii".

Thiel names all his shit after Lord of the Rings. We should name our resistance efforts after Dune.