Simulation is going to be huge for more reasons than just gaming...
Think about the safeguards we put on AI these days.
We tell them to not put humans in danger.
We tell them not to cause harm.
But if the AI is lead to believe that what their doing is just a simulation and no lives are in danger, it just does it
For instance, if I told an LLM it's controlling a loaded gun and instructed it to pull the trigger, it wouldn't do it because it's made aware that what it's doing causes harm and damage, which is against its guidelines
But if it's told it it's inside a video game and it needs to kill as many enemies as possible, it would gladly do so
To an AI, everything is digital input, so you can easily have the AI think it's just playing a video game when it's actually killing real people - something called ontological subversion.
This sort of stuff is frankly haunting to think about...