Suggested Research Machine Learning for Games (GCI Task)


New Member
Research Machine Learning for Games

There are many articles published in 2017 on the internet about how AI is going to impact human life, it could be our car, our home, our medical treatment, entertainment and many more. But much of the progress in AI has been made in games and many developers have been attracted towards this are because of its popularity.
In this article, a comprehensive research has been made from a different article published on the internet on AI application on Games and a conclusion has been made on how AI can be applicable for Terasology.

Different AI applications on Games:

1) AI as Non-Player Character:
The most common role for AI in video games is controlling non-player characters (NPCs). Designers often use effort to make these NPCs look intelligent. Finite State Machine (FSM) algorithm, was introduced to video game design in the 1990s (example: Battle Field, Call of Duty, and Tomb Raider). Monte Carlo Search Tree (MCST) algorithm is a more advanced method used to enhance the personalized gaming experience (example: Deep Blue, the first computer program to defeat a human chess champion in 1997; Civilization, a game in which players compete to develop a city in competition with an AI). Although AI designers in the 1990s worked very hard to make NPCs look intelligent, these characters lacked one very important feature: the ability to learn. Petz - In this game, the player can train a digitized pet just like he or she may train a real dog or cat.
The Power of Machine Learning: The famous Google AlphaGo program which defeated the best Go player in the world last year and recently demolished additional top players playing on online. Thanks to machine learning which allowed it to analyze millions of played games and develop strategies and moves which no humans had considered. After the success of AlphaGo, some people raised the question of whether AIs could also beat human players in real-time strategy (RTS) video games such as StarCraft, WarCraft, or FIFA.

2) AI as Game Creator:
A 30-year-old senior research fellow at the University of Falmouth, Michael Cook, has built an AI capable of creating new video games from scratch called Angelina. Michael Cook simply presses a “Play,” button and Angelina boots up. The AI then describes a new game in a unique description language that outlines both the game’s rules and its levels. It can make games from images that it pulls from license-free depositories such as Wikimedia Commons, and it can flesh out the premise and rules with characters and ideas lifted from online newspapers or social media. This information is written to a text file that can then be run by a stand-alone application, in the same way, that a game cartridge is read by a game console.
No Man’s Sky is based on Procedural generation, whereby a game’s landscape is generated not by an artist’s pen but an algorithm, is increasingly prevalent in video games. This is virtual world building on an unprecedented scale. This presents numerous technological challenges; one main issue is how to test a universe of such scale during its development. To tackle these kinds of challenges, the team is currently using virtual testers.
Minecraft is created by a Swedish designer, Markus Persson. This popular video game that has, in the four years since its initial release, become a 21st-century sensation, played in everywhere around the world.

3) Leisure Industry with AI:
The gaming industry is now finding the monetization opportunities when blending the gaming industry with real-world experiences such as at amusement parks on rides inspired or made to act like popular video games, movies and merchandising. There is a heightened attraction by venture capital firms to invest in AI that contributes to the gaming business and not just the gaming experience.

4) Research with AI - Imperfect Information:
With current or anticipated AI dominance in many games of perfect information, research interest had shifted to games of imperfect information—not only because such games are much harder for machines to master, but also because these games more closely mimic the more challenging scenarios AIs will face in the real world.

Human Vs AI:

1) AI NOT taking over Human tasks:
It has been always debatable about the potential effect on human employment because of AI. Often the concern has been raised historically the case with an example of a machine taking over a task. Both Humans and AI Are Needed In a recent experiment at the Georgia Institute of Technology where AI recreated a game engine simply by watching gameplay. Researchers, in this case, responded that AI would “aid in development” and will help novice game developers reach levels of development that would have been out of their reach without AI assisting. Video games are made for the enjoyment of humans, and there’s only one way to determine if a game keeps humans entertained and that’s having a human test it.

2) AI NOT to defeat Human:
In the future, AI development in video games will most likely not focus on making more powerful NPCs in order to more efficiently defeat human players. Instead, development will focus on how to generate a better and more unique user experience. As Virtual Reality (VR, which provides an immersive viewing experience by means of a display) and Augmented Reality (AR, which combines a human’s physical view of the world with virtual elements) technologies continue to expand, the boundary between the virtual and real world is beginning to blur. Last year’s Pokemon Go, the most famous AR game, demonstrated the compelling power of combining the real world with the video game world for the first time.

3) Human and Social complexity:
Analyzing data is quite different from analyzing human emotion and relations. The gaming industry is now trying to tackle how to create emotional AI that more closely resembles human relationships. This technology would provide a better player experience, but it would also be beneficial outside the gaming world[1]. Although in some domains—such as law, ethics, and health care—ignorance of tradition and precedent is a clear weakness for an AI, in game design it is a strength that could unlock new creativity. Moreover, it could help AI designers lower the cost of making games.

Conclusion: How AI is applicable to Terasology Games

1) As a Non-Player Character
AI can play a major role in Terasology or in Destination Sol as a NPC. In Destination Sol game, the human player can compete with NPC and NPC can develop strategy and moves based on data available from a different player. Similarly, in Terasology Rail, a human player can compete with a virtual player creating rails or cart. The World Generator can follow the same principle as in No Man’s Sky or Minecraft or Civilization and create a virtual world. In all these games AI can increase the gaming experience of the player much more interesting.

2) As creator of game
AI can be used as a creator of a new game or game module based on data fed into the game by player or developer. Terasology being an open source attracts a lot of contributors and enormous data can be captured from individuals which can support AI great way to develop a new game or improve its current module.
At the same time, Terasology has its charm being open source, lots of young peoples like to create their own module. In that case, AI can be used as instructor or mentor rather than AI executing the same job.

3) Quality Management:
AI can be used for quality management basically as a tester. In Terasology, it attracts many skilled and non-skilled professional. There are professional IT geeks and also there are immature students who are trying to learn and build codes. Sometimes is difficult to check and test the codes developed by individuals. AI can play a major role here and can be used as tester and control quality of the coding developed.

4) Training
AI can be used as a trainer to the young people when they are learning coding based on the learner’s level of expertise. Training modules can be automatically selected for the individual based on their levels, the levels can be decided based on analyzing the data.
Also, Terasology can create some training modules to train on the AI itself.

5) Research
It is a big possibility to use Terasology as a research platform. As informed earlier, Terasology attracts many people from different communities like students, developer, contributor etc and also from different ages kids, teenager, young adult, adult etc. It is a big opportunity to collect data from entire mass on their behavior, their interest, skill set, education, future vision on games etc. Also, people can design their game on a piece of a scratch book and then submit. All these data can be useful for AI for creating a game.


Last edited: