In a recent video relating to the Double Fine Kickstarter project, Tim Schafer asks Ron Gilbert about games with “good” dialogue, to which Gilbert makes the bold assertion that very few games have even serviceable dialogue. He provides a basis for his point by pointing at the decrease in the attention given to the strengths of the medium, with many game writers instead trying to co-opt screen writing techniques from Hollywood.

After finishing Mass Effect 3, along with a bevy of other RPGs and “adventure” games over the past few years, I think that Gilbert’s point is a wholly accurate one; despite the strange way he goes about stating it. To truly write interactive dialogue, there can be little choice given to the player as to what to say, but there must be an illusion of infinite choice in order to maintain a balance between the illusion of a natural conversation, as opposed to a stilted exposition ridden cutscene.

The only logical solution to this problem is the creation of an AI based procedural dialogue system. I realize that may sound like a radical solution to a minimal problem, but through the creation of a smart conversation driven AI, dialogue in games will become vastly improved with the side benefit of leading to smarter devices, and more astute software.

The Turing test is a test designed to see if a computer can fool a human into thinking that it is a real person. The test is designed in such a way that a test subject will ask questions through a computer, and see if they are able to tell the difference between an AI and an actual human. A lot of effort is expended each year in hopes of creating a computer program capable of fooling a majority of people into believing it is a real person. This focus adds some credence to the future possibility of a narrative driven game populated with completely artificial characters.

If this software could ever be utilized as part of a game, the possibilities could be endless. Imagine sitting in front of your television and asking questions via a Kinect like device and getting real time answers based on a set of game world defined parameters. No longer would a player be beholden to a number of different dialogue choices. The experience of creating a dynamic and populated world would no longer be a dream, simply because your virtual inhabitants could pass themselves off as real players.

While I am certain that the implementation of a fully realized AI is a ways off, the prospect is still incredibly interesting to think about. The game play ramifications are immense, and would further our understanding of how the human mind thinks and acts given different sets of cirumstance.

Advertisements