In the title by bad scenario I mean a case where the AI starts doing something evil.
You can have a number of employees very smart and pay them, but can you have enough scenarios for possible situations?
The answer is that you can’t, if the number of possibilities is too big or infinite.
So probably companies will start asking people to send scenarios to them and will start paying people for them, based on quality.
It’s better to have a budget for that and lower the risk, than to see how your autonomous factory starts building Terminators.
This companies factories probably will be located in special places and around them will be army to terminate them fast if the AI tries to prepare for a war in any way.
Companies that produce AI software and not only them, will pay people to send bad and good scenarios
