20 October 2023

Game AI Pro 360 - Guide to Tactics and Strategy

There are many books about artificial intelligence (AI), but most are either very academic, take a purely educational approach, or focus on more applied topics such as machine learning and data analysis. The bibliography on AI focused on video games is very scarce, especially if you are looking for it in digital format (in my case Kindle).

However, one of the great authors in this field is Steve Ravin, not so much for his own contributions as for his work compiling those of others in his works. His is the series of books "AI Game Programming Wisdom" of which this is one of his latest exponents. As in the previous ones, each chapter of this book is an article by a different author, developing a concept related to AI in video games. Unlike the previous books, which had articles covering a wide range of areas, in this one all the articles focus on the facet of analysis and the making of tactical and strategic decisions.

The main value of the book is that the authors of the different chapters are professionals specialized in this field who work in reputable video game studios. Thus, the problems, nuances, and approaches they propose are backed by their extensive experience. The authors' competence is also evident in the academic air that emanates from each of their pages.

However, the main handicap of the book is precisely this: the work background of its authors. Coming from large studios, they must be subject to confidentiality clauses and I suppose that is the reason why they do not go into detail about specific implementations, since they could compromise the intellectual property of their companies. These authors can explain the general concepts adopted in their games, but they cannot reveal the details of how they have put them into practice without compromising the core of those games.

For this reason, the book moves at a very conceptual level in which they propose really interesting things, but it falls short when it comes to explaining how they have implemented them. Code snippets are conspicuous by their absence and when they appear they are mere pseudocode that in reality does not fully explain the concepts raised. There have been complete chapters in which I lost the thread after a few pages and I did not know how to fill in the gaps that the authors were leaving. In others I have understood everything they were proposing, but I was left wanting an example of a real implementation.

Probably I lack the level that the authors have, but my conclusion is that it is certainly not a book to get started in the field of AI for video games. For this, I recommend the book "AI for Games" by Ian Millington before this book. This last book, does go from an introductory level to a medium-high level and with all the details. I thought that this book would already leave me prepared for others, already at an intermediate level, but it turns out that not for this one by Steve Ravin.

Do I regret having read it? No. When I was able to get the hang of the chapters I enjoyed them and I came up with a couple of ideas that even though I don't know how to implement them yet, they have been suggestive enough for me to try them at the first opportunity.

Verdict of the book: not suitable for beginners, but an interesting read to get you familiar with the concepts that drive the AI of major video games.