bluemoonx
You can make permissions decide which model is used, and only train that on the data that permission allows (someone should build this - permissioned RAG), that solves leaks, but decision-making within a model, or behavior exfiltration comparable to viewing backend source code still seems possible.
It’s basically like client/server security: You can’t “trust the model” in the same way you can’t “trust the client” in a backend/frontend setup.
When used as effectively a point-of-sale, AI seems more hackable than a vending machine - as a boss or assistant even more so.
The closest thing now is amphetamine but it turns people into maniacs - presumably the action here is not actually the brain cleaning function of sleep, rather just preventing reuptake of neurotransmitting chemicals that would cause it, if allowed.
Would be nice to truly provide the mechanism in a synthetic form - I wonder what it would feel like to have the brain “slept” while awake.
There's a class of people that has to have "more" than the rest. And they do that by controlling access to everything. They spend tons of money on distractions and concepts like moneylending which makes everyone have to go through the "loan" and "approval" process - of course only some people get approved lol So nobody else gets a house!
Instead of trying to become better themselves and make new discoveries, and build genuine wealth they can give back, the gatekeeping/inheritance mindset is to stop others who are trying to better themselves. To knock them back. It doesn't stop at preventing others from attaining wealth and status, they don't even want you to have a roof over your head.
If they had it their way, you wouldn't have a bank, house, car, and barely any money. You'd be wearing some goofy uniform working for them living in poverty.