Posted by: crudbasher | September 13, 2011

Using AI For Learning Games

Probably the biggest problem in creating an education system where every student has their own personalized curriculum is one of scale. It’s not really practical to have a teacher create personalized lessons for each student so how do you do it? How do you mass produce customization?

Well, on one hand we are dealing with computers so at least you don’t have to do something physical. Information is pliable. I found this article below that might give a clue how to do this.

(cc) gnta

I have speculated that the best way to do a custom learning environment would be to use Artificial Intelligence. Imagine a computer that would be with the child from birth. It would be able to watch everything the child does and identify areas of strength and weaknesses. It would then be able to create lessons on the fly as the child goes through life.

I know a lot of teachers would object that an AI is impersonal and you always will need a flesh and blood teacher. That may be true but I don’t think so. When the telephone companies switched over to direct dial, there were people who said you would always need a human operator to connect calls. I would think that an AI system which is with a child 24 hours a day would have a much better insight than a teacher who gets a child for 6 months, sometimes for only one hour a day.

(cc) Tinkerbots

I still think there will be needs for teachers, I just think it’s likely that 80-90% of the learning will come from AI at some point.

Please check out the article. The sort of technology they are talking about can be applied to automated, personalized lessons I think.





  • AI based game characters

    tags: technology AI innovative

    • the creation of a game that turns gamers into game developers.
    • Called Storybricks, the game, developed by London-based Namaste Entertainment, uses artificial intelligence software to allow people to program their own characters and storylines – some of which may continue forever.
    • Storybricks splits behavioural software commonly used for games into user-friendly “bricks” that can be connected together. Instead of defining the details of how to walk into a house, for example, a player can define character traits in advance so that a character “wants” to be “inside the house” and the software takes care of the rest.
    • Such bricks include a number of drives that can be programmed into any character, such as “want” or “fear”.
    • The game’s designers hope Storybricks, due to launch early next year, will inspire players to devise characters, scenarios and emergent stories that Namaste’s staff could never have envisioned. If the concept is successful, it could spawn a vast landscape of bespoke, user-generated games – any of which could be endlessly tweaked by a group of friends.

Posted from Diigo. The rest of my favorite links are here.



  1. I think, first pass, the AI could be evaluating the child. The AI would determine the child’s proficiencies and best learning methods. Then a human teacher could creatively connect the best learning methods of the child to the child’s deficiencies. That way the human is doing what she does best ( creative solutions ) and the computer is doing what it does best ( unblinking measurements ). It also allows the teacher to build on what the child already knows without rehashing what the child does know, which can lead to boredom. Once you have a solid evaluation of a child’s proficiencies another approach could be creative matching of groups of children. With a proper mix of talents and an interesting problem to solve a group of children can progress each other.

    • Quite so Tom. The more I think about it having an AI act as a constant evaluator and guide is a capability missing from the current school system. Teachers only get a glimpse of a child’s capabilities but don’t get to see the big picture. Imagine if all the teachers in the world could submit their lessons into a global market that the AI could then pull from. Thousands of variations of lessons for all different types of learners would be amazing! Your idea about matching children in group lessons is pretty much possible today with the Internet, but nobody has really written that app yet.

      Thanks so much for your excellent comment! 🙂

  2. Matching children in group lessons is possible today, but I think the quality of those matches would be random. We don’t have that list indicating a break down of concepts acquired per child. I think that is the cornerstone, because it is the most difficult problem. The best thing our society has come up with so far is testing. That is the tool we use to expose whether a person has mastered a skill or concept. If we could build an AI evaluator with the features you and I are talking about we wouldn’t even need tests anymore.

    • Yeah the more i look at the whole of education the more I realize effective assessment would make more of a difference than anything else. AI evaluation could do that eventually.

      • Do you have any pointers to people currently doing that AI evaluation research?

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s


%d bloggers like this: