Mario AI vs. Infinite, Evolving Super Mario World

Ladies and gentlemen, in this corner, a version of Super Mario World that automatically generates harder levels when the player completes easier ones without failing. A project by Ben Weber, a computer science doctoral candidate at UC Santa Cruz, the game adjusts parameters such as gap size, gap frequency, variation in ground height and, of course, number of enemies.

In this corner, a Mario AI that calculates all possible jump speeds and angles and takes the appropriate course of action based on its surroundings. Created by Robin Baumgarten, the so-called A*-enhanced agent was the winner of Mario AI Competition 2009, whose purpose seems self-explanatory.

The winner is revealed in the video above, but my mind was already blown by the concept alone. [via Geekosystem]