One trend I've been predicting and encouraging for a few years now is the increasing use of AI technology in other aspects of game development — particularly for player control. Game AI has already solved some particularly difficult challenges for making context-sensitive decisions then making sure they happen correctly during execution, and it's obvious why such concepts can benefit many places in a game engine. I talked about this in my keynote at the AI Games Network Workshop on Players & Avatars... Let's just say the subject of applying AI to player control was debated passionately then, so I figured this should make for another heated developer discussion!
At the time I gave my talk, the only examples of AI assistance were things like local navigational aids via smart objects. However, with in the upcoming iteration of the stealth shooter franchise, Ubisoft is set to provide another perfect example of how AI techniques can be used to assist the player locally, without it feeling frustrating. In this the rest of this post, you'll see a video of this “Mark & Execute” gameplay that's new to Splinter Cell Conviction. Similarly to the previous discussion, I'll also provide some arguments on either side why it can be a risky choice, but what the rewards can be.
(You should also read the many interesting and thoughtful comments posted last time around about the Last Known Position visualization.)