Artificial Intelligence in Games News – Week 12
Screenshot from Street Fighter II
Welcome to the weekly roundup, where we draw your attention to everything newsworthy in artificial intelligence in games! If you’ve found or shared something that you think belongs here, contact us on Twitter! Last week, we took a trip to China and heard about how AI is crushing even more professional human gamers. You can find that here.
Street Fighter II Artificial Intelligence
Here’s one for all of you bytecode enthusiasts. Gamasutra has dug up an article from back in January that deconstructs the AI in Street Fighter II. As the game is now over 25 years old, the article provides us with a fascinating look into the level of sophistication AI had back then. Briefly, the AI engine is split into three operations; waiting for an attack, actively attacking and reacting to an attack. Scripts are chosen based on variables such as time left, and of course, some of these scripts do let the AI cheat! It is possible for example, for characters to attack by simply following instructions meaning they cannot fail and are invincible!
Fighting The Artificial Intelligence Blitzkrieg 3!
We spoke about Blitzkrieg 3 earlier on in the year and commended it for its impressive neural network powered AI system, Boris. Although not technically news, here we have the perspective of a regular player, and it’s great to see how the game is developing. The standard AI has all of the typical RTS behaviours fans of the genre have come to expect, including units that engage effectively whilst avoiding rough terrain and obstacles. However it’s multiplayer AI Boris is most impressive. The author notes that during every playthrough the AI feels not like the same AI employing different strategies, but as if he is fighting a different human player every time. He also points out that Boris is tough, but not unbeatable and the trick is to play unpredictably to outwit the AI.
Screenshot from Blitzkrieg 3
Artificial Intelligence Engine Investment
Absentia, the company behind Norah AI in India has recently received a large investment fund. The company aims to be the first to automatically generate content for different media. It has already had success monetising its platform for virtual reality and augmented reality. Furthermore, the system could be used to generate content for a variety of games, animations and graphics. The company uses neural networks in its product and are said to have a major advantage in India. A really promising and interesting new AI contender!
Screenshot from ‘Quick, Draw!’
Google Drawing Stuff
Remember ‘Quick, Draw!’, the game that Google released that provided you with a word for you to draw and for the computer opponent to guess? Well, another research team has used those millions of sketches to build a system called Sketch-RNN. The system, similar to a child, mimics the humans drawing. For example to draw a cat it first looks at over 70,000 images of human-drawn cats. What’s most interesting about this AI is its ability to understand the components of the image using shapes. Most AI simply checks pixel for pixel and can’t separate an eye from a nose. Sketch-RNN is able to finish sketches humans started and could even help suggest edits to finished art.
Need More News?
BrainyBeard keeps you up to date with all of the latest artificial intelligence news in games. Sign up to our newsletter below to ensure that you don’t miss our first newsletter going live later this week!