The AI Babysitter Book Learnin University of Chicago
The AI Babysitter
Book Learnin’ ¢ University of Chicago BA in General Studies l Masters work in AI & Information Systems l ¢ Northwestern University l Ph. D work in AI: Robotics & Video Games
Street Smarts Game Design and Tuning Workshop ¢ Indie Game Jam/Experimental Gameplay Workshop ¢ IGDA Education Committee ¢ Currently: Lead Designer: My. Sims ¢
The Designer-Player Relationship Designer Creates Game Consumes Player
The Designer-Player Relationship AI Creates Game Consumes Player
A Formal Model of “Game Consumption” Rules System Behavior “Fun”
The MDA Framework Mechanics Dynamics Aesthetics
Our Task Hide-and-seek game ¢ CMU: “Woggles” ¢ l Tag the T
Our Task Hide-and-seek game ¢ Can be played from both perspectives ¢ the babysitter l and the child l ¢ Simple, 3 room environment
Our Map
Avatar: Babysitter Moves ¢ Speaks ¢ Extra Moves? Why? ¢
Eight Kinds of "Fun" 1. Sensation Game as sensepleasure 2. Fantasy Game as make-believe 3. Narrative Game as drama 4. Challenge Game as obstacle course 5. Fellowship Game as social framework 6. Discovery Game as uncharted territory 7. Expression Game as self-discovery 8. Submission Game as pastime
excitement The Dynamics of Hide and Seek exhaustion
The Babysitter What kinds of fun? ¢ What kinds of AI? ¢
Technical Details Waypoints? ¢ Obstacle detection? ¢ Cover points? ¢
Seek! ¢ Brainstorm: 5 behaviors Babysitter (NPC) l Baby (PC) l Prototype/Play/Discuss ¢ Goal: Clearly understand mechanics, dynamics, aesthetics of this design ¢
Our Map
Results? ¢ What makes playing with the babysitter fun?
New Constraints Player = babysitter ¢ What kinds of fun? ¢ What kinds of baby AI? ¢
AI : Baby ¢ Reconsider: l Hiding behaviors
AI : Baby ¢ Reconsider: Hiding behaviors l “Tells” l
AI : Baby ¢ Reconsider: Hiding behaviors l “Tells” l Group goals? l
AI : Baby ¢ Reconsider: Hiding behaviors l “Tells” l Group goals? l Pacing and Flow l
Seek! ¢ Brainstorm: What changes? Babysitter (PC) l Baby (NPC) l Prototype/Play/Discuss ¢ How has the player experience changed? ¢
Hunter vs. Hunted ¢ What makes playing with the babies fun?
Baby AI ¢ Advanced abilities and moves?
Technical Details Waypoints? ¢ Obstacle detection? ¢ Cover points? ¢
Technical Details Waypoints? ¢ Obstacle detection? ¢ Cover points? ¢ World view? ¢
Technical Details Waypoints? ¢ Obstacle detection? ¢ Cover points? ¢ World view? ¢ Competition? ¢
But Wait – There’s More!! Full-blown simulation of tag ¢ Multiple Babies ¢ “Real Strategic Gameplay” ¢ Showcase character AI. ¢
Baby AI ¢ What does it look like NOW?
Reality is Expensive Can’t do it all by hand! ¢ Consider “style” or “personality” ¢ Easier to recognize/predict l Different targets require different skills l ¢ This means more granularity – more complex representation.
Fidelity ¢ Hiding How much cover is the fern. . . as opposed to the closet or under the table? l Do AIs perceive others and plan around them? l Do they track the babysitter? l What if they can turn on and off lights? l
Scale Did your design scale when player character changed? ¢ Does it scale when the aesthetic itself changes? ¢ Is it still fun ¢ To desgin l To play l
Other Thoughts?
- Slides: 35