So I was surfing the web today, not really looking for anything in particular when I came across an article about "Machine Learning in Games development." And since I had recently talked about things gamers might like to see in games I thought , "Self, this Nick Palmer guy might have a point." After reading the article I thought , "Wow this cat knows his stuff." So I decided it needed some added exposure.

In the article, Mr. Palmer talks about the fake learning machine versus the real learning machine. Many game developers, I'm sure, use the first because it's quick and easy. But the second would truly make the game better albeit more difficult for both creators and players.

Yes! That's what I want. I want a game that learns to suck as much as I do at some things, like sniper rifles. But I also want a game that learns I can't use them and so behaves in different ways. I want characters in games to react differently depending on how I respond to them and not just because I chose option 1 (of 3) but because I show a specific pattern in my responses.

Mr. Palmer has put together a well written and insightful look at how to make these things possible.

You can read more about it here. Now if only he would take the next step and develop the learning algorithm for us to use we could all have a more challenging game experience. (And he too would then be a rich man).

Did you enjoy this article? Please bookmark it onto: