How Video Games Laid the Framework for Artificial Intelligence

Do video games deserve credit for artificial intelligence development?

Artificial Intelligence has been a growing field of development for software engineers, computer scientists and just about everyone else in the tech industry. While artificial intelligence seems like a new endeavor for the tech industry, video games have been relying on it for decades.

  • Early AI in video games
  • How it is being implemented today in other technology
  • What’s in store for the future

AI Basics in Video Games

To find the first video game using artificial intelligence you have to go all the way back to 1952 when engineers successfully computerized the game Nim. The game is essentially a mathematically based strategy game where players remove pieces from a different heap, trying to avoid being the person to take the last piece from the board. The artificial intelligence in Nim was the opponent and had a notoriously high winning percentage against even the most intelligent of players. This spurred the computerization of many classic board games like chess and checkers as well. Even the classic video game Space Invaders used artificial intelligence by increasing difficulty, moving in distinct patterns and other actions based on the player’s movements.

Artificial Intelligence was also highly utilized in creating opponents in games like Street Fighter, Pac-Man and Galaxian. This AI was typically designed to give opponent's a higher variety of movement patterns and less predictability.

Madden video game
‍EA Sports’ Madden NFL video games use artificial intelligence to mimic players’ movements, coaching tendencies and even simulate games.

But perhaps the most dynamic and extensive use of artificial intelligence came for sports video games like EA Sports’ Madden NFL. The football video game uses advanced motion capture techniques so their artificial intelligence can perform lifelike movements and gestures. Madden also cycles through different weather conditions that cause players to move and act differently. Weekly updates affect every player’s overall abilities depending on their real life performance. This allows Madden to be used as a very realistic simulation of actual NFL games, making it a unique tool for predicting the outcomes of the entire season.

Learning from Video Games

Scientists use video games as a training ground of sorts, simulating the real world. You don’t need to worry about the physical wear and tear of something in a video game. This allows scientists to make adjustments with just a few lines of code rather than redesigning and entire robot. You also can run thousands of different simulations by doing the same.

There are also plenty of other unique uses for video games in artificial intelligence. For example, a computer scientist at Harvard named Artur Filipowicz was working on teaching cars how to recognize different things on the road. Before even worrying about crossing pedestrians or animals jumping out in front of you last minute, he was hung up on the issue of stop signs. How would a self-driving automobile know to stop at a stop sign? He realized he would need to train his algorithm by showing it thousands of pictures of stop signs. He needed his algorithm to recognize stop signs in the dark, through bright sunlight, during a blizzard, partially obscured by trees, rusted stop signs and much more. Taking each of these pictures along would be an extremely tedious task.

Filipowicz chose to confront his unique problem in an interesting way. He turned to the polarizing video game, Grand Theft Auto V, due to its realistic behavior of its artificial intelligence. He noticed computer generated character driving cars not only stopped at stop signs, but also obeyed many other laws of the road. After experimenting with the code, Filipowicz was able to access thousands of different portrayals of stop signs in the game which he then fed through his algorithm.

What’s in Store for the Future

It doesn’t look like artificial intelligence development is going to slow down anytime soon. Many tech companies utilize video games in their AI development as well. Last year Microsoft began “Project Malmo” in hopes of achieving greater collaboration between people and artificial intelligence. The project is still very early on, but the basis of it is to use the popular video game Minecraft to help create a personal assistant. The idea sounds quite far fetched, but actually makes a lot of sense as Minecraft is literally a virtual playground where players can create just about anything they wish by using simple blocks. The plan is to teach artificial intelligence how to not only cooperate with but help people achieve their various goals. This is definitely a project worth keeping an eye on.

Video games will continue to provide a safe place for scientists to test theories and work out kinks in their algorithms. It is extremely cost effective and all signs point to it being a smart investment for companies.

Video games always have and look as if they will always be integral in understand and testing artificial intelligence. If you look at the progress video games have made in the past 50 years, it gives plenty of optimism as to where AI can go.