Discussion about AI/Pattern Recognition/Machine Learning in PvE Games

I'm not talking about the standard AI where it's state machines etc, but more of the pattern reconnition and machine learning.



if you don't know what these terms mean, here are some wiki pages:

http://ift.tt/TK3Fy1

http://ift.tt/KUjnPK

http://ift.tt/KUjoTE



One of things about PvE games is that no matter how hard you make them, some people will figure it out and it'll be easy for them after enough practice. So what can you do to make a PvE game challenging in a good way for people who would have been bored? RGN stuff is 95% of the time a cop out and doesn't make it harder in a good way. After thinking about this it lead me to topic at hand.



This isn't a topic about KF (1 or 2) exactly, but using KF as a basis for examples might be helpful. But how do you think that these tools could make PvE games challenging for all skill levels, particularly the highly skilled players? Always pushing the player(s) without having to add new difficulty levels or something of the like. What things should it focus on for input, what should it change (output), should learning data be volatile or non-volatile across games, etc etc.