SIT740 - Research and Development: Syntopic Reading on AI in Games

This report presents a syntopic reading of three research papers focusing on the application of Artificial Intelligence (AI) in games. The first paper discusses AI-based game design patterns, highlighting the importance of AI in enhancing player interaction, using Spy Party as an example. The second paper explores the use of machine learning, specifically Deep Q-Networks, to improve player experience in real-time strategy games like Brood War and StarCraft, noting limitations in learning growth due to reliance on human input data. The third paper examines the use of AI in strategy games with large search spaces, such as Chess, emphasizing AI's ability to adapt to various situations. The report concludes that AI has a beneficial impact on the gaming experience, while also mentioning the limitations in calculation time. This assignment is available on Desklib, a platform providing study tools for students.

Running head: GAMES WITH ARTIFICIAL INTELLIGENCE
Games with Artificial Intelligence
Name of the Student
Name of the University
Author Note

1GAMES WITH ARTIFICIAL INTELLIGENCE
AI-based Game Design Patterns
AI is mainly used for non-player characters to support the main player of the game.
However, the literature suggests that AI must to handle more player interactions rather than
working in the background.
Various games have been designed such as Spy Party, Galactic Arms Race, Black and
White and many more where the AI interacts with the player to develop a rich gaming
experience for the players. Each of the games have different applications of AI. In Spy Party, the
AI is used as a role-model where the players study the AI movements and one player must
imitate the AI whereas the other player must detect the player from the AI. However, the
limitation to the AI integration is that the AI is still native and not complex.
The outcome achieved was that the games became more interactive.

2GAMES WITH ARTIFICIAL INTELLIGENCE
TorchCraft: a Library for Machine Learning Research on Real-Time Strategy Games
Brood War and StarCraft are different real-time strategy games. The literature elaborates
that the player experience from these games can be improved by incorporating AI. These games
have been used as a benchmark for measuring the progress of AI research.
Popular games such as Mario AI and Minecraft have been interfaced with deep-learning
to develop the bot interactions with the players. Deep Q-Network is one of the solutions provided
in this literature that aided in achieving human-like interactions for the non-player characters.
Various experiments were conducted that collected data from player games and used those
interactions to develop the AI interactions. The limitations would be that the machine learning is
based on human input data collection and thus it limits the learning growth curve.
The outcome is better real-time strategy games with high level game character
interactions.

3GAMES WITH ARTIFICIAL INTELLIGENCE
Hierarchical Portfolio Search: Prismata’s Robust AI Architecture for Games with Large
Search Spaces
Strategy games have large search spaces. This means that the non-player characters of the
game must select one of the strategies among the various combinations that are generally coded
in the database. Chess is one such game where there are huge numbers of strategical and tactical
methodologies that can be applied to progress through the game achieve the objective of ‘check-
mate’.
This literature aids in elaborating the advantages of using AI for such games. An AI has
the capability to adapt to different situations by selecting the best possible method among a
library of methodologies. Various experiments prove that the utilization of AI is beneficial to
develop a rich experience for these games. The limitation being sometimes the calculation time
taken by the AI might not be affordable by the users.
The outcome was challenging bot opponents in these games that can adapt to any
situation.

4GAMES WITH ARTIFICIAL INTELLIGENCE
Reference List
Churchill, D. and Buro, M., 2015, September. Hierarchical portfolio search: Prismata’s robust AI
architecture for games with large search spaces. In Proceedings of the Artificial Intelligence in
Interactive Digital Entertainment Conference(pp. 16-22).
Synnaeve, G., Nardelli, N., Auvolat, A., Chintala, S., Lacroix, T., Lin, Z., Richoux, F. and
Usunier, N., 2016. Torchcraft: a library for machine learning research on real-time strategy
games. arXiv preprint arXiv:1611.00625.
Treanor, M., Zook, A., Eladhari, M.P., Togelius, J., Smith, G., Cook, M., Thompson, T.,
Magerko, B., Levine, J. and Smith, A., 2015. Ai-based game design patterns.