Pablo Moreno-Ger’s blog

This weekend I’ve been playing the videogame America’s Army. It is a competitive online game developed and freely distributed by the U.S. Army.

First of all, a brief disclaimer: Although it is marketed as a tool to learn about the life in the army, strictly speaking it can’t be considered as an educational game. Actually, the game is designed as a recruitment tool, giving a very partial view of army life mixed with highly patriotic messages about honour, duty and the pride of being an American soldier.

However, there are very interesting things inside it that anyone interested in the field of “serious gaming” should take notice of:

First and most important: The game is FUN. This is the part that many educational games often forget. A boring game is not better than a boring lecture. In AA we find cooperative factors, teamplay with other humans, ranking systems… all these factors make the game attractive and urge the player to progress in the game and replay the missions again and again.

Also very important: It is a high quality product with a high level of realism. It features state-of-the-art graphics, realistic 3D sound, elaborated smoke and fog effects, etc. It truly feels professional and could compete in shelves with A+ games (remember, the game is distributed free of charge).

Its quality, its widespread success and the big entity behind the game (Department of Defence) make AA a reference in the field of serious games.

As of learning value, the game itself includes some training lessons to be played offline before deploying into online matches. These lessons are practical (like the typical videogame tutorial) and some others are theoretical, with the objective of transmitting that a military career includes a good deal of interesting studying. The theoretical lessons are studied inside the 3D game world, with the player entering a classroom where there is a teacher and some slides are presented on the wall. After a few minutes of lecture there is a short assessment.

This theoretical part is as boring as a regular classroom, although very condensed. And it is always possible to find amusement in watching the body language of the NPCs in the most realistic representation of the classroom ambient that I have ever seen.

It is also interesting to point out that the game imbues some additional knowledge in a subtle way. The stereotypical videogame portrays the player as the absolute hero, which obviously doesn’t happen in modern armies. The realism of the game makes it impossible to win any kind of mission on your own and it is imperative to cooperate deeply in order to achieve good results. In addition, once the results are achieved, the entire team is rewarded. Here you have a rich and successful tool to teach the importance of team play.

And for the end, a small rant: Playing this game helps in my research and is as relevant as evaluating a tool or reading an article. Why do I see myself forced to play the game at home during the weekend? We must not forget one of the greatest barriers of game-based learning: The social perception of videogames as simple entertainment.

I wrote a short paper some moths ago about videogames and Artificial Intelligence, reflecting on how the videogame industry conceives (and values) the AI from a point of view that is opposed to that adopted by academics in the field of computer science. I wrote it for a course in my PhD, and I have now published it here.

The main idea behind it is how videogames don’t actually use any modern AI techniques. With the exception of Black & White, no videogame uses bold technologies like machine learning, intelligent and adaptative behaviour (as opposed to scripted AI). Any decision in that industry must be accompanied by a report on its potential economic benefit in the short term. The market is so demanding and there is so much money being risked, that no company will allow the developers to boldly go where no man has gone before. That is, of course, unless those developers are Peter Molyneaux and have the capacity to guarantee success only by signing the product.

On the other hand, there is a thriving activity in AI in computer science at an academic level. Machine Learning, Computational Creativity, Case-Based Reasoning, Fuzzy Logics… All of them big words with dedicated conferences. However, videogame AI has barely evolved in the last few years. They use the old techniques better and better, but its still the same technology.

There are, however, powerful reasons for this. Will a game that includes an advanced AI system that rises its budget in 500,000US$ generate that additional revenue only because of that? The answer is, not probably. So, why bother?

Moreover, the audience must be prepared. Imagine a game with a truly complex AI with the capacity to truly surprise the player. If the AI is to innovate and dynamically change its behaviour, it won’t be possible to test that AI in depth. The QA department can’t guarantee 100% a correct behaviour because if they did so, then the AI would not be truly innovative. Will the players (the buyers) accept a scenario where the AI learns the wrong lessons and starts to degrade its behaviour to the point of needing a full reset? Will they accept having to start a game from the beginning again due to an AI glitch?

I would be willing to accept it for the sake of progress and innovation, would you?

You can read the whole paper here, although it is in Spanish.