Here at Video Game Intelligence (VGI), we are interested in exploring various different types of game Artificial Intelligence (AI) developed, be it from AAA game studios or Google’s research labs. Whether you are a casual gamer curious to learn what goes behind the scene, or a serious computer science student considering a career in artificial intelligence or video game development, our website intends to offer useful information and insights on game AI to sate your thirst for knowledge. For any unanswered questions on the topic or requests for investigation into the AI used in any specific game, please contact us at caiyiming2015@gmail.com.
What is Artificial Intelligence in Video Games?
Traditionally, AI is used to define any intelligent behaviors exhibited by man-made, non-living things, specifically machines and computers. In computer science terms, AI is defined as the study of ‘intelligent agents’, or an intelligent system capable of adapting to changing environments and goals, learning from experience, and making appropriate decisions given the computational limitations.
On the other hand, AI in video games typically refer to any piece of the game program that is used to control and thus define the behavior of any uncontrollable elements such as NPCs and procedurally generated level designs. It can refer to the behavior of those random opponents that you face in an action game, or the algorithm used to generate those seemingly random and yet always-beatable maps in those tactical role playing games (RPG) such as the Disgaea titles (check out the video below for an example, note that all the maps are randomly generated in real time and some hardcore players would beat literally thousands of these maps before moving on to the next game).
As we can see from the above example, AI is frequently used in video games to exponentially increase the number of variations that the player can interact with, without needing the game developers to manually design every single level since that would take a huge amount of labor (and thus money). While ‘real’ AI techniques such as machine learning are often not featured due to their complexity, Random Number Generator (RNG) is a heavily ‘abused’ mechanics that can be found in almost any game AI, especially since it is a very convenient and easy way to increase variations. There are many game AIs that are built solely using patterns determined by some combinations of RNGs, even though the field of computational randomness is not all that related to the study of artificial intelligence.
Venn diagram to demonstrate more examples of Video Game AI and Computer Science AI
While researchers often consider the term ‘Artificial Intelligence’ used in video games a separate term from the broadly defined ‘Artificial Intelligence’ as an area of study of Computer Science, modern game AI design is gradually blurring the boundary between the two terms. Various researchers have taken interest in implementing highly advanced AI techniques in video games to further add complexities and variations, and perhaps even a sense of ‘human’ in the behavior of the NPCs. Conversely, various AI research have utilized games as a testing medium to test the intelligence level of ‘pseudo’ general purpose AIs and new AI techniques. In fact, game playing is now considered an independent branch of AI research, with various interesting projects. such as IBM’s Deep Blue (1998) and Google DeepMind’s AlphaGo (2015) and AlphaZero (2017), shedding light on the seemingly limitless problem-solving potential of AI through traditional game mediums like Chess and Go.
Video Game Intelligence 