Embracing Nouvelle Game AI
At the time of writing, most professional game AI developers are the first to admit that their job involves little AI technology at all. Indeed, techniques such as pathfinding or scripting are generally not considered AI (especially not in the academic sense). Although this classical game AI approach serves its purpose, trends in recent games reveal a need for improved NPCs to increase the entertainment levels. Therefore, advancements in AI technology are necessary to assist the development of realistic and intelligent NPCs.
The solution is partly based on academic AI techniques discussed throughout this book (such as neural networks and decision trees). Although their theory has been researched extensively over the past decades, this in itself is not enough to drive the necessary revolution in AI technology. Future progress in game AI development will be driven by changes in methodologies and practices that are capable of coping with the following:
This book takes a modern approach to these open problems, inspired by principles from the nouvelle AI trend in robotics and concepts reminiscent of agile development of software. Both these approaches are popular in their respective domains and are proven to work in professional environments. As demonstrated throughout this book, their applicability to game AI development is unquestionable—notably feature-driven development using evolutionary (incremental) design, with rapid iterations backed by extensive testing.
In fact, this approach is arguably necessary to develop adaptive AI, because of the level of complexity and the unpredictability of the task. Much research still needs to be done on the subject, but this book is written in the hope that the information herein provides a significant step in the right direction.