Nicholas Renotte: I Trained AI To "DOMINATE" Brackey's Game

I dove into training a Reinforcement Learning agent in GoDot to absolutely schmooze Brackey’s prototypical game, inspired by a mining-optimization demo I saw at a think tank. It was a Sunday-night whirlwind of fun, laying the groundwork for more custom RL environments and riffs on Brackey’s tutorial—huge thanks to Ed Beeching (GoDotRL) and the legend @Brackeys for making it possible.

All the code (14 versions and counting) is up on GitHub—just shout if you need the hefty 167 MB environment files—and I’m flying to Florida to present tomorrow, so hit me up if you’re around! While you’re at it, check out my socials, join the Discord, and let’s keep geeking out over AI and game dev.

Watch on YouTube