Artificial Intelligence in Games

Almost all games have bots or game mode where you play against the computer – Even the ones that are popular for their multiplayer gameplay have bots (Fortnite, Warzone). When discussing bots I don’t only mean AAA games or demanding games. The way those bots or computers are designed to play against humans in different conditions can be referred to as Artificial Intelligence in Games. It is fairly common as almost all of the games have a computer mode. Sure application and uses can differ from game to game; some games may use more advanced AI algorithms compared to others.

Artificial Intelligence in Games: Bots in Fortnite

There are different ways in which AI and game development are growing through each other. Even though AI keeps on being accustomed to bring life into video games, computer games are currently being designed with the purpose to study their patterns to improve their algorithms. Some of the leading AI labs like DeepMind and OpenAI apply and improve their AI through its application in games to advance their research in Self Learning AI. Self-Learning AIs such as these haven’t been used on commercial game development yet because developing these advanced AI could take a lot of time if not more than required for making the video game.

Is It Truly Artificial Intelligence Though?

One could argue that Game AI isn’t true AI at all since for most of the part the AI works with a predefined algorithm. Not learning but working in existing measures (more like if this, then this else this). Here for those of you who want to analyse game AI more, check out this problem (here’s my solution to it. Now though this is not a self-learning AI since it calculates the output for different input. You can think how Chess AI operates now).

The absence of recognizable jumps is on the grounds that the fundamental AI administering how those virtual elements carry on — and the AI controlling procedural age apparatuses — has not gone through a revolutionary change throughout the long term. “Two of the centre segments of business game AI are pathfinding and limited state machines,” clarifies Julian Togelius, a partner educator at New York University’s branch of software engineering and designing who has practical experience in the convergence of AI and computer games. 

Artificial Intelligence in Games: Pathfinding Machines

“Pathfinding is the means by which to get from direct A toward point B in a straightforward manner, and it’s utilized in all the games constantly.” A limited state machine is where a [non-playable character] can be in various states and move between them. So we can likewise contend that AI in games is, even more, a popular expression utilized as an advertising instrument as opposed to a self-learning PC device. So we can also argue that AI in games is more of a buzzword used as a marketing tool rather than a self-learning computer tool. 

Research purposes

Artificial Intelligence in Games: AlphaGo

Apart from video games, more sophisticated use of AI is done by researchers through games. Often researchers make smart AIs that are capable of beating almost all players. AlphaGo can arguably be called the best go player in the world, whereas Deepmind’s Starcraft 2 AI can beat 99.8% of all players. This is truly a feat because AI such as the one in Starcraft 2 often improves after playing a lot of games with players of varying capabilities. Research AIs improve continuously without human intervention from playing games.

Artificial Intelligence in Games: Starcraft 2

Researchers are continuously working on pitting AI against humans in more advanced games (Starcraft 2 and DOTA 2 are big achievements). One reason why researchers often use video games as a means to improve their AIs is the ease of collecting data. AI improves with the increasing quantity of meaningful data hence it’s crucial for the development. AI researchers are largely using games as a way to benchmark the intelligence level of a piece of software and because virtual worlds, with strict rule and reward systems, are a particularly useful environment to train software in. 

Why is Game AI so different from DeepMind’s AI that plays games?

Both are very different because one can be tweaked just enough to have a controllable output whereas the other is uncharted territory. I have been randomly drawing stuff on Google Quick Draw for more than two years now. It’s an online game which was created by AI researchers at Google that asks you to draw stuff and if your drawn stuff can be guessed correctly by AI then you score a point. This game has drastically changed since it was launched as it feeds on data to grow its database, hence, becoming more and more accurate.

A game such as QuickDraw could use such sophisticated AIs since the AI is not directly hindering human performance. But is it truly a video game? No, it isn’t a commercial game but an ongoing research project to make their AI more intelligent. Every time you play Quick Draw you are contributing to make it better than before.

In any case, there’s a valid justification why most games, even the latest huge spending titles utilizing the most sophisticated design tools and innovations, don’t utilize this sort of cutting-edge AI. That’s because self-learning programming would probably make most games unplayable, either on the grounds that the demonstration of playing the game would be excessively uncontrollably capricious or because AI would carry on in a manner that could make telling a story.

“Game developers tend to prioritize the kinds of actions that we can predict. Even though it’s very interesting when AI does unpredictable things, it’s not necessarily super fun for players,” explains Tanya Short, a game designer

Final Thoughts!

AI is an emerging field of computer science so its commercial applications are limited to spambots, chatbots, virtual assistants etc. We are very optimistic that we may see much more use of AI in game development and games in near future. As soon as the leading game engine developers adopt and blend some of the Artificial Intelligence algorithms, we will see floods of games using it to make games better and quicker than before.

Though we expect True Artificial Intelligence to be used a lot in games and other apps too, we don’t expect it to be used for the creation of bots or game AI. Some of the expected uses can be for visual recognitions in future when you log in for your Activision or steam account or maybe to create your look-alike character in a video game – Games like Cyberpunk definitely hint developers want playable characters to be as customizable as they could. Anyways We are very excited to see the future of AI, games and game AI.

Liked this piece? Also Read: Cloud Gaming Explained: Could It Be the Next Big Change?

For Exclusive Social Media Content Follow Us on:

Leave a Reply