January 9, 2019
AI Project Research
GDC 2018 – AI Wish List: What Do Designers Want out of AI?
By: Raph Koster, Dave Mark, Richard Lemarchand, Laralyn McWilliams, Noah Falstein, and Robin Hunicke
GDC 2018 – AI Wish List: What Do Designers Want out of AI?
I’m bringing back a video for reference to try to come up with an AI intensive project. There are lots of quotes and the notation is very note-y, just using it to keep track of points of inspiration.
Laralyn mentions that she wants AI to be able to acknowledge that what the player is doing is unique. The way they are playing is differentiating from what is expected, and it should be noted by surrounding characters. Could we apply a type of “physical constraint” concept to this to help accomplish this goal? Physical constraints mathematically map out what an equilibrium should be, and then apply forces/actions to bring things out of line back to equilibrium. Could we use this concept to perform AI actions when it acknowledges a player diverges from what is expected, the “equilibrium” of the player experience?
From Raph Koster, “…building characters as props that are primarily reactive, and if we want to exploit AI, they need to have their own inner lives, out of which this stuff rises organically.”
”Heartificial Intelligence” – giving NPC’s empathy; Have deeper version of Sim system. Users more interested in “broken” characters with flaws. A dog that is scared of people now and is harder to train. Problem with Sim system was that every object had to echo out how it should affect an NPC, and they all needed to interact with each other so adding things to the system was very difficult(?).
Time Stamps: (0:49:40 – 0:51:18) : AI can be recursive and add detail. “…Can you do that but in a way that is data rich, instead of just going for ‘no let’s plot thousands of trees’, how about when I go look at that tree it’s way more detailed. Now there’s an ecosystem in that tree. Now there’s a burrow in this one and the gophers live under that, and there’s a woodpecker in that one. That’s the kind of detail we will never ever be manually able to do. We just can’t afford it, it’s ridiculous. And it’s exactly the kind of intelligently constructed, realistic, responsive environment. You want the woodpecker to know there’s a gopher. I’m not talking just shallowly, again, it’s not just splatting the textures. Can we actually create a data rich environment? … it’s a fractal or segmented thing. We don’t need each tree to know about each tree. But we could do something really interesting within the tree… each one could have variations.”
Solaris example – it builds things from your memory; makes you miss being on earth
”…go to a foreign planet and you find an object, and when you look inside that object there’s more objects, and you look inside those objects; and then come up with a design of mechanics that are interesting from the exploratory perspective that’s very similar to when you’re a child. When you wander into your backyard, and then the woods there and you find a stream then you look in the stream and you see the tadpoles and then you poke the tadpoles. Think of it from a mechanical perspective, not an aesthetics perspective, and then you get the dynamics that are so interesting.
Decentralize player, other things go on without the player. Make the player feel proud of something else. Have an AI do something the player didn’t expect it to do. Going off of this: Have an AI that the player needs to “teach” it how to do something with their actions. Similar to “Clumsy Ninja”.