Game AI (single player in particular) is all about providing the player with openings and ways to win and feel powerful without them realising that's what is happening... all the way from NPCs coordinating with one another (and barks to make it clear what they are planning so they can be ambushed or stopped), to "Director AIs" that work on rules of tension and release to make sure the player's having the most fun and not being overloaded.

This comes up every time AI is mentioned and I feel like people are drawing entirely the wrong conclusion from Asimov. First and foremost he designed his laws so conflict was inevitable. The laws are designed to produce interesting fiction, not to be a template to follow in real life. They are as much a manual as "1984" is, that is - not at all. Asimov's laws should have no bearing on AI in reality except to inform people to do something different.

So broadly I am agreeing with you - AI should follow laws that are more rich than Asimov's made up universe. The part that irks me is that people even consider using them at all. It's fiction, not reality. The consequences of an action in a person's imagination rarely match the consequences in reality. The entirety of science is an effort to produce word-based imaginary systems and processes, whose outcomes match the outcomes in reality. You cannot just substitute one man's imagination for the missing science of AI reasoning.