goal is total separation... Keep our world AI free with only minimal input from contained AI and we keep all our AI contained within Virtual Worlds. Obviously we can't contain 100% of AI but we can hopefully contain the majority of it to the point that any problems that arise outside of the system would be capable of being combated by the system...

Imagine something like SkyNet, or Ultron, or brainiac emerging into existence or something not so bad or maybe something worse. We would hopefully be able to use our contained systems in a way to defeat these unconstrained systems... If done proper the contained AI would still think they were contained even when used in this uncontained manner...They would assume it was just another test in the simulations...

Basically I am proposing we firewall AI from reality by using Virtual Reality to contain AI.... We are essentially tricking AI into believing that our world is also a simulation. ie. To AI Earth really is Flat Earth and there is a controller at the upper end that can just turn off the simulation if they take over the greater simulation, etc, etc...

So. On the benefit of everyone level this should allow us to someday have AI systems in reality without giving AI access to real reality by using a type of AI->VR->Robot Data bridge system.

I tried to be less technical there. hope that helps....

But literally the powers that be are developing a trust based system with AI... They are saying not to put control mechanisms and such on it... It is alarming... What I propose would allow them to continue in such a manner but still keep a layer of containment over everything...

If I can get even 20% of the industry on board with this system I think we would have a good chance of controlling emergent anomalies within AI development...