Narrative Review of Game AI 2000 Onwards and Future Research Directions

Introduction

Artificial intelligence is used to process necessary information to execute intelligent reasoning comparable to that of humans. Artificial intelligence is a broad term that encompasses a variety of technologies. AI is referred to as “cognitive technologies'' since it is a group of technologies that have the ability to perform tasks which normally need human intelligence. What differs with AI is its ability to learn from data sets and adjust according to its behaviors to optimize outcomes rather than following and implementing a set of processes and rules. (Deloitte, 2017; Quantum, 2017).

“Game AI” includes Programming and design techniques like neural networks, finite state machines, path finding, models of emotion and social situations, and decision-tree learning, “Game AI'' refers to a “intelligent” behavior that makes the with its desires which appears to respond to a player's actions which appears connected to the meaning of the player's actions (Matas, 2003).

In the area of Game AI, there is a lot of research going on. It may be used in various domains, including adversarial planning, real-time reactive behavior and planning, adversarial planning, and decision-making under situations to help address challenges. (Cabrera et al., 2015). It involves choosing or mapping complex strategies, refining and defining components, inventing move sequences which are contextual, and responding to human actions at the strategic level are all issues that gaming environments encounter. Various aspects of a game's virtual environment which includes characters, objects, sound, lighting, camera and sound, are situated at the interface levels. This altogether hampers the facilitation of interaction. The current state of machine learning, especially in computer games, is formulated chiefly and produces intelligent characters that are human-like. This focus is carried out irrespective of complex opponent behaviors that emerge through various learning techniques. However, research on whether these behaviors add to the player's enjoyment is yet unclear. Researchers hypothesize, for example, based on the various amounts of multi-player online games played regularly, that developing human-like opponents enables the player to obtain more satisfaction.

Within the game AI field, Yannakakis and Togelius identified ten essential pieces of research. The researchers were more concerned with the relationships between them and the influences and interconnections they had on one another. The Game AI textbook, published in 2018, provided a better understanding of game AI. The research had three essential components of game AI. In procedural content generation (PCG), non-player characters (NPCs), procedural level generation (PLG), and player experience modeling. Real-time strategy games, which make up most video games, are also utilized as test-beds and frameworks in this process. This is done to test new AI algorithms for real-time strategy and AI.

The study's research aims are as follows:

• 1.

  What are the aspects and ideas of progress in the Game AI literature?

• 2.

  What are the recommendations for enabling Game AI implementation?

Methodology

Articles were extracted from reputed publishers, indexed databases and web resources (Google Scholar, Wiley Online Library (Wiley-Blackwell), Taylor and Francis online Science Direct/Elsevier, Springer Journals Database, Sage Journals) and screened and classified using the process of exclusion and inclusion (Figure 1)

The following criteria were used to choose articles for this review:

“Artificial Intelligence”/”Machine Learning” in “Gaming/Simulation” and “Game AI.”

Publications from 2000 onwards and up to April 2021 were selected for further review as articles previous to 2000 could not be found in the literature by the authors. Research Papers published in group and high-quality journals, review articles, and conference proceedings with a high citation index were evaluated.