Revisiting the Photoshop of AI Debate
| « Previous Event | Next Event » |
Michael Mateas, Director, Associate Professor, The Center for Games and Playable Media
Wednesday, February 1, 2012, 11:00 AM to 12:00 PM
Location: Engineering 2, Room 180 (The Simularium)
Hosted By the Center for Games and Playable Media
Abstract:
Several years ago noted game developer Chris Hecker argued that graphics has made huge strides because it has found a style vs. structure decomposition: the underlying triangles of polygon models are the structure, while the textures put on the triangles is the style. This allows a clean separation between all the math necessary to manipulate and render 3D models, and the artistry of creating textures for these models in programs such as Photoshop. He argued that the reason artificial intelligence in games has not made as much progress as graphics is because a structure/style decomposition has not been found for AI, and that what we need is a Photoshop of AI. This spurred a debate in the game AI community about what a Photoshop of AI might look like, and whether it's even possible. In this talk I respond to the debate by arguing that a Photoshop of AI is impossible, and that future progress in game AI will be made by a new breed of artist/programmers who deeply understand both the aesthetics of behavior and the technical details of AI architectures and algorithms.
Bio:
