Nitpick: Artificial Intelligence or Artificial Stupidity?
Something that I've expected to improve this generation, compared to its predecessor, is the complexity of artificial intelligence (AI). To be more specific though, I'm talking about actual intelligence and the realistic responses that characters can provide in-game. This means that if I hit someone I expect him or her to be angry. If I give a gift they should give a response that shows pleasure or displeasure, depending on the gift of course. However, these are mere expectations, and just that – nothing more or less.
Obviously, there are some things that cannot be expected of developers when they create AIs. Complex series of codes and packages and sets of parameters must be set in place to create realistic responses for AI. To be more blunt, it's hard to fathom the difficulty behind creating AIs unless you're a programmer yourself. For me, who took college-level intro computer science courses, I don't know how to design entire levels or any case studies. Still, I understand the work and effort that developers put into creating AI, especially those that want to deliver an immersive experience.
If you compare that AI from this generation to the last, and even before that, it's clear how far the games industry has come in creating a complex entertainment package. We could never imagine that we would get excellent AIs back on the Super Nintendo Entertainment System. Games like The Legend of Zelda: Ocarina of Time was an excellent game because of enemy AI. I remember picking up the controller for the first time and going off against two Lizalfos in Dodongo Cavern. The way they strafed, jumped and attacked was absolutely baffling. They reacted a lot more realistically than its previous Zelda counterpart, A Link to the Past.
This isn't to say that AI was perfect in generations previous to the current. Recently I got back into playing the Splinter Cell series. To be specific, I was playing Chaos Theory and noticed numerous instances where I screamed in front of the television, “How do you not know I'm there?!” I would be hiding in the dark with three big green lights stuck on my forehead. The enemy would walk up right in front of me and say various things like, “I'm sure I heard something.” I scratch my head in absolute confusion. Your infiltrator is clearly right in front of you yet the AI is too dumb to notice. Obviously, this is nitpicking – the point of the article – and the game would be absolutely broken if enemies could detect you under such circumstances. The game would be more challenging and much more difficult.
Now, Splinter Cell: Chaos Theory came out last generation in 2005. Seven years have passed so I should expect something much more complex. Not only does the gap in time add new software tools at the developer's disposal but they have a much more expansive hardware to work from, with the exception of the Wii for obvious reasons. Games such as Dark Souls has been praised for its excellent AI. After all, for its collision-detecting combat system that prides itself on difficult and rewarding encounters, the game must have an AI to match that. It isn't surprising then that Dark Souls' AI understand when to use spells, parry, riposte, sneak attack, and other realistic combat strategies and tools at their disposal. When I played the game, it was a fight for survival because of how realistically challenging the enemies I was going up against were.