AI Topics on Educational Serious Game Development

Videogames have gained importance over the last few years, not only as entertainment products but also as a learning tools, advertising platforms and even as instruments for different types of scientific research. These are only a few of all the possible uses that videogames can have, but all of them can be grouped into a same category, which is called 'Serious Games'.

Focusing on the education genres of Serious Games I stumbled across several topics involving artificial intelligence that I found of great interest:

Natural Language Processing

Natural language applications that have been developed are far from perfect, but when correctly implemented, they can create experiences that can't be achieved with regular applications. One clear example is the game Façade, a game in which the player can play by saying what comes to his or her mind and the AI will make the characters react accordingly. Playing this game will reveal the advantages and disadvantages of such a broad gameplay mechanic: when the user plays the game as it was intended, the experience is enhanced to the point of being memorable, but when the player tries to 'game the game' (try to break it by writing nonsense or complex phrases), the AI tries as best as it can to give the player a reaction, but the game won't be able to react accordingly to each and every sentence thrown at it, which breaks the suspension of disbelief.

Another, more limited example is Scribblenauts series. In these games, the player has to solve different puzzle-like scenarios by using what he can find on said scenario. The catch is that the player can type anything up and that thing will materialize in front of his or her eyes. For example, if the objective is to reach an object that is too high, the player could type 'table' and 'chair', then place the chair above the table and jump on top of it to reach the object, or maybe another solution would be to type 'stairs' and see if the type of stairs that appear can be used. Crazier options include trucks, airplanes, atomic bombs, black holes, Santa Clause, Kraken, God, Chutlhu, and even meme characters like Longcat. Like Façade, this game won't be able to react to each and every word you throw at it, but because you use single words instead of complete sentences, the effect will last longer.

It would be interesting to see if natural language processing can be implemented in an education serious game to enhance the experience and improve the retention rate of the player.

Automatic Difficulty Adjustment

Being able to adjust the game's difficulty on-the-fly can be a great way to keep the player interested without the game becoming stale or too hard. In the case of a serious game focusing on education, this means keeping the player in the best possible learning state. There have been several attempts to do this, for example Watcharasatharpornpong proposed the use of genetic algorithms to automatically adjust the difficulty of a platformer game by generating levels that match the player's skill. Another example could be the game 'Warning Forever', a shooter in which the objective is to destroy the boss of the game in each level untill you die. The catch is that the boss learns from your previous actions and adapts accordingly in the following level, for example, if your strategy was to stay in the middle of the screen most of the time, it's probable that the next boss will turn around to force you to move from the center and change tactics.

It can also be interesting to see if this would be effective in an education-based serious game.

Expressive AI

Expressive AI is a term created by Michael Mateas which involves the AI making a game feel 'alive'. A simple example he provides is the famous game Pac-man. In this game, the ghosts' behavior make the player think that they are actually setting up traps but in fact, they are only following simple but really good thought out patterns. This effectively makes the game feel 'alive'.

The game Façade is also an example of a game Expressive AI because the characters really behave as expected creating the illusion that they indeed have feelings. This also makes the game fun, because it keeps the player interested and wanting to know how things will turn out, mostly because it's the player that's causing those reactions.

This is a broad topic that could find several implementations in an education-based serious game as a means to maintain the player engaged or as a means to improve the learning and retention rate of the player.

Educational games are perceived as 'boring', so creating an effective educational serious game that is perceived as a 'cool' game is quite a challenge, but with the correct implementation of several different techniques, it is possible to develop a fun game that engages players and at the same time, teaches them something without them noticing.

References:

• Watcharasatharpornpong, Nirach. Automatic Level Difficulty Adjustment in Platform Games Using Genetic Algorithm Based Methodology. (2009)
• Mateas, M. "Expressive AI: Games and Artificial Intelligence", in Proceedings of International DiGRA Conference. (2003).
• Wilcox, Bruce, Beyond Faade: Pattern Matching for Natural Language Applications. (2011)
• Bellotti, F., Berta, R., & De Gloria, A. (2010). Designing Effective Serious Games: Opportunities and Challenges for Research. International Journal of Emerging Technologies in Learning, 22-35.
• Iwatani, Toru. PacMan (1980)
• ProceduralArts. Façade (2005)
• 5th Cell. Scribblenauts (2009)
• Ohkubo, Hikoza. Warning Forever (2005)