Artificial Intelligence and Games

1 Georgios N. Yannakakis and Julian TogeliusArtificial Intelligence and GamesJanuary 26, 2018 SpringerTo our familiesForewordIt is my great pleasure to write the foreword for this excellent and timely have long been seen as the perfect test-bed for artificial Intelligence (AI)methods, and are also becoming an increasingly important application area. GameAI is a broad field, covering everything from the challenge of making super-humanAI for difficult Games such as Go orStarCraft, to creative applications such as theautomated generation of novel AI is as old as AI itself, but over the last decade the field has seen mas-sive expansion and enrichment with the inclusion of video Games , which now com-prise more than 50% of all published work in the area and enable us to address abroader range of challenges that have great commercial, social, economic and scien-tific interest.

2 A great surge in research output occurred in 2005, coinciding with boththe first IEEE Symposium (Conference) on Computational Intelligence and Games (CIG) which I co-chaired with Graham Kendall and the first AAAI AIIDE Con-ference (Artificial Intelligence in Digital Entertainment). Since then this rich area ofresearch has been more explored and better understood. The Game AI communitypioneered much of the research which is now becoming (or about to become) moremainstream AI, such as Monte Carlo Tree Search, procedural content generation,playing Games based on screen capture, and automated game the last decade, progress in deep learning has had a profound and transfor-mational effect on many difficult problems, including speech recognition, machinetranslation, natural language understanding and computer vision.

3 As a result, com-puters can now achieve human-competitive performance in a wide range of percep-tion and recognition tasks. Many of these systems are now available to the program-mer via a range of so-called cognitive services. More recently, deep reinforcementlearning has achieved ground-breaking success in a number of difficult challenges,including Go and the amazing feat of learning to play Games directly from screencapture (playing from pixels). It is fascinating to contemplate what this could meanfor Games as we stumble towards human-level Intelligence in an increasing numberof areas.

4 The impacts will be significant for the Intelligence of in-game characters,the way in which we interact with them and for the way Games are designed book makes an enormous contribution to this captivating, vibrant area ofstudy: an area that is developing rapidly both in breadth and depth as AI is able tocope with a wider range of tasks (and to perform those tasks to increasing levels ofexcellence). The service to the community will be felt for many years to come: thebook provides an easier and more comprehensive entry point for newcomers to thefield than previously available, whilst also providing an indispensable reference forexisting AI and Games researchers wishing to learn about topics outside their directfield of Yannakakis and Julian Togelius have been involved with the field eversince its widespread expansion to video Games , and they both presented papers atthe first 2005 CIG.

5 Over the years they have made an enormous contribution to thefield with a great number of highly cited papers presenting both novel research andcomprehensive surveys. It is my opinion that these authors are best qualified to writethis book, and they do not disappoint. The book will serve the community very wellfor many years to , August 2017 Simon LucasPrefaceOf all the things that wisdom provides for the complete happiness of one s entire life, by farthe greatest is ,Principal Doctrines, 27 Human beings, viewed as behaving systems, are quite simple. The apparent complexity ofour behavior over time is largely a reflection of the complexity of the environment in whichwe find A.

6 SimonIt would be an understatement to say thatArtificial Intelligence (AI) is a populartopic at the moment, and it is unlikely to become any less important in the researchers than ever work on AI in some form, and more non-researchers thanever are interested in the field. It would also be an understatement to say thatgamesare a popular application area for AI research. While board Games have been centralto AI research since the inception of the field, video Games have during the lastdecade increasingly become the domain of choice for testing and showcasing newalgorithms. At the same time, video Games themselves have become more diverseand sophisticated, and some of them incorporate advances in AI for controllingnon-player characters, generating content or adapting to players.

7 Game developershave increasingly realized the power of AI methods to analyze large volumes ofplayer data and optimize game designs. And a small but growing community ofresearchers and designers experiment with ways of using AI to design and createcomplete Games , automatically or in dialog with humans. It is indeed an excitingtime to be working on AI and Games !This is a book aboutAI and Games . As far as we know, it is the firstcompre-hensivetextbook covering the field. With comprehensive, we mean that it featuresall the major application areas of AI methods within Games : game-playing, con-tent generation and player modeling.

8 We also mean that it discusses AI problemsin many different types of Games , including board Games and video Games of manygenres. The book is also comprehensive in that it takes multiple perspectives of AIand Games : how Games can be used to test and develop AI, how AI can be usedixxPrefaceto make Games better and easier to develop, and to understand players and this is an academic book which is primarily aimed at students and researchers,we will frequently address problems and methods relevant for game designers wrote this book based on our long experience doing research on AI for Games ,each on our own and together, and helping lead and shape the research both independently started researching AI methods in Games in 2004, and wehave been working together since 2009.

9 Together, we played a role in introducingresearch topics such as procedural content generation and player modeling to theacademic research community, and created several of the most widely used game-based AI benchmarks. This book is in a sense a natural outgrowth of the classes onAI and Games we have taught at three universities, and the several survey papers ofthe field and of individual research topics within it that we have published over theyears. But the book is also a response to the lack of a good introductory book for theresearch field. Early discussions on writing such a book date back at least a decade,but no-one actually wrote one, until could be useful to point out what this book is not.

10 It is not a hands-on bookwith step-by-step instructions on how to build AI for your game. It does not featurediscussions of any particular game engine or software framework, and it does notdiscuss software engineering aspects or many implementation aspects at all. It is notan introductory book, and it does not give a gentle introduction to basic AI or gamedesign concepts. For all these roles, there are better books , this is a book for readers who already understand AI methods and con-cepts to the level of having taken an introductory AI course, and the introductorycomputer science or engineering courses that led up to that course.