AI in video-game industry

This report is on how the video-game industry has provided a way for practical research in the field of AI. AI plays a key role in the video-game industry, which is comparable to the film business in size. Unlike in the movies, it's often up to a computer or game console to create a sense of reality for the player, and standards of realism are going up all the time.

One of the problems researchers have in the field of AI is that corporate funding of research can be unpredictable when a immediate commercial application can not be guaranteed from that research. I believe that one of the main reasons that the video-game industry can help provide a viable research platform in the field of AI is that this industry is one of the few that their customers demand better use of AI in each successive commercial application. Although historically, computer games have allocated less than 10 percent of their processing cycles to AI routines. Recently, however, more substantial AI techniques have been making their way into commercial systems as low-cost, high-end graphics cards and increased CPU processing power free system resources.

As the video-game industry becomes more competitive and "great" graphics become common in most new games. The importance of having the best AI in their games can be the deciding factor in whether consumers buy their games.

Work on games has had several traditional justifications. Given uncertain rules, playing a game to win is a well-defined problem. A game's rules create artificial world states whose granularity is explicit. There is an initial state, state space with clear transitions, and a set of readily describable goal states. Without intervening instrumentation, games are also noise-free. For these reasons, as well as for their ability to amuse, games have often been referred to as "toy domains". To play the most difficult games well, however, a program must contend with fundamental issues in AI: knowledge representation, search, learning, and planning.

AI techniques offer the promise of creating engaging and dynamic interactive entertainment with strong narrative components. For AI researchers working in the context of video games, research challenges are as complex and compelling as many real-world problem areas; gaming environments offer unique interfaces and modes of use and an extensive existing base of potential users.

Gaming environments pose a range of problems, at both the strategic and interface levels. Strategic-level challenges in computer games can involve mapping or choosing between complex strategies, refining components of a strategy by formulating context-specific move sequences, and detecting and responding to human users’ actions. At the interface level, intelligent components inside a game must control how the game world is presented to the users. Areas of application include the following topics. Path finding, at a low level, determines how to move multiple characters across unfamiliar and possibly dynamic terrain can be a problem central to many of the most popular game titles.

Natural-language comprehension and generation could be considered one of the most important factors in fostering the perception of a characters intelligence and its ability to communicate using more than repetitive, preprogrammed scripts or templates. Learning is being tried in a number of new games,such as Black and White (http://www.bwgame.com ).They are beginning to include characters that adapt to a player's behavior as a prominent game feature. Intriguing challenges arise mainly because of the limited interfaces typically provided between a learning agent and a computer game user.

Task planning is when a game controls individual characters according to scripts that are written at design time. As a result, a characters behavior is limited in its ability to respond to unanticipated run-time information. Combinations of generative and reactive planning algorithms may provide the means for creating customized, unique behavior that changes each time a game is played.

Several of the more popular games, including titles such as Quake 2, Unreal Tournament, Half-Life, Homeworld, and Descent 3, provide application program interfaces (API) that allow users sophisticated control over the dynamics of the game world. These APIs are either written in C/C++ or in a Java/C++-like scripting languages. Andy Gavin at Naughty Dog Software

( http://www.naughtydog.com ) developed a Lisp extension called GOOL (game object-oriented language) and used it to build the character behaviors for Crash Bandicoot, a top-selling game on the Sony PlayStation.

In addition to the technical strengths of the computer game as a test environment, the economics of the gaming industry makes gaming environments intriguing targets for the development of AI techniques. For several years, annual computer game sales have exceeded Hollywood's domestic box office receipts. Statistics released by Forrester Research show that total computer and console game revenues for 2000 exceeded $5 billion; Forrester also projects that gaming consoles will exist in 70 million U.S. house-holds by 2005. For their potential social impact, intelligent techniques integrated with entertainment titles may match any other area of AI research by market penetration alone.

In this paper I chose to focus on the "newest" computer and console games instead of the research being conducted in "classical" games such as chess, checkers, and go. While research in "classical" games might lead to a major break through in AI, I do not believe it will. The focus on specialization of hardware and software designed to play one game very well,i.e. IBM's Deep Blue chess player, does not seem to break new ground in the field of AI. As one of my classmates pointed out, are they actually adding anything new to the field of AI or are they just creating powerful calculators.

Epstein,Susan.,Game Playing: The Next Moves

Hirsh,Haym.,Playing with AI,Trends and Controversies

Young,Robert & Robert Amant,2001.,Artificial Intelligence and Interactive

Entertainment,Intelligence