دانلود کتاب هوش مصنوعی و بازی‌ها، ویرایش دوم

It’s my pleasure to introduce the second edition of the Artificial Intelligence and Games book, a publication that seeks to consolidate the variety of ways artificial intelligence (AI) can be deployed in or for games. The intersection of AI and games is a rather loaded one. The field of AI has long held games in high regard, often seen as a mechanism to highlight the capabilities of intelligent algorithms. From the early days of chess and checkers in the 1950s and 1960s through to more contemporary works in Go, StarCraft, and Dota 2 that you’ve heard about in the media. The desire to build intelligent systems that “play” has also sat alongside the growing suite of AI tools and systems deployed within commercial game development. As such, when one says they have an interest in “AI for games”, this can mean many things. You could be building opponents for board games or strategy video games, non-player characters in video games that provide a challenge or support the player in their experience. It could be dynamic difficulty systems that observe players and change the game to suit, or procedural generation algorithms designed to create everything from characters to quest lines, levels, weapons, or even entire games. But equally it can also mean building systems and tools that the player does not observe. Data analytics pipelines for modeling and quantifying player behavior are used in many popular titles to observe behavioral trends. Ranging from analyzing how you win a clutch moment, to understanding whether you figured out how to use a specific menu or interface, or detecting whether other players around you are cheating or toxic. Meanwhile, quality assurance frameworks powered by AI allow for games to be tested during—and after—release to not just reduce bugs in the product, but help developers keep the game from breaking as they’re building it—a common occurrence, I can assure you. If you’re joining this space for the first time, this can all be rather intimidating. The landscape of AI algorithms, methodologies, and approaches, specifically for games, has exploded in the past 20 years. Not only as academic research began to see the value of games as a mechanism to deploy AI in a variety of scientific contexts but also as the games industry has seen the proliferation of these technologies, and recognized the opportunities that exist to integrate it into their production pipelines. Trying to get a grasp on this technological wave and communicate it to a broader audience has long been my mission, largely via my work on the AI and Games YouTube channel. For 10 years, I have sought to detail the workings of dozens of commercial video games and summarize academic research for an audience that quite often does not have the same grasp of the subject as I do. In fact, I’ve been asked on several occasions whether I would be interested in writing a book that consolidates this for a more scholarly audience. It’s a notion I’ve rejected, because not only does that book already exist—spoiler alert, you’re reading it—but it’s written by the two people arguably best equipped to do so. Georgios N. Yannakakis and Julian Togelius have ridden the wave of game AI research since its inception in the mid-2000s, both earning Ph.D.s in AI for games at a time where the world cared little for it. Our authors have been not just at the forefront of many of the methodologies discussed in this book—delivering many papers that are often cited either for their novelty or comprehensive surveys of the subject—but continue to guide the next generation of game/AI scholars and developers in understanding how to make these ideas tractable and practical. Both as active research supervisors and as the organizers of the AI and Games Summer School which welcomes hundreds of participants each year from around the world. Togelius and Yannakakis were the researchers that I was citing in my Ph.D. thesis 15 years ago, and I can’t think of many individuals who equal their contributions to the field of AI research for games. So rest assured, you’re in safe hands. This second edition is a reflection of how much the field has evolved since the original release in 2018. Not only as the technology has seen significant gains but also in how these technologies can be applied in both commercial projects or scholarly pursuits that utilize games for broader research goals. We have entered a period of change and upheaval as AI begins to intersect various corners of society. At present how new AI methods and tooling will influence the development of commercial games is not readily apparent. Progress can often be as fast as it is slow. But the foundational knowledge required to get a grasp on this space can be found in these pages. And so while the field will continue to see a significant change in the years to come, this book will continue to serve as stepping stone for members of the game AI community both young and old.