Differences between Industrial-oriented AI and Game-oriented AI.

Or stating the obvious to game project owners and academic programmers.

Once upon a time in the not-so-sleepy hell scape of game development, lived two intriguing characters—Booz, a game developer with a penchant for straight-forward mechanics and user-friendly design, and Dr. John Doe, an academic programmer whose mind wandered in the theoretical landscapes of algorithms and Turing machines. One fine day, Dr. Doe, in a rush of excitement, broke the door to Booz’s work space. “I’ve got it, Booz! The ultimate algorithm for game AI! It involves a seven-layer neural network, Bayesian inference, and a sprinkle of quantum computing!”

Booz looked up from his quadruple-monitor setup, one screen displaying an unfinished video settings class, another showing millions of Slack messages, the other a forgotten debug session running from yesterday, and the forth playing scenes from “Die Hard” (for inspiration, of course). “Quantum what now?” Booz scratched his head. “Look, Dr. John, I’m just trying to make my orcs run toward the hero without tripping over rocks that don’t exist. Do we really need Schrödinger’s Cat to decide whether the orc attacks or retreats?” Dr. Doe frowned, baffled that Booz couldn’t immediately grasp the monumental leap this could be for game AI. “Don’t you understand? Your orcs won’t just run; they’ll contemplate the metaphysical implications of their run!”

Days turned into an endless loop of heated debates and whiteboard sessions that looked like the scribblings of a mad scientist. Dr. Doe advocated for “elegant complexity,” while Booz clung to his mantra of “keep it simple, keep it human.” Pizza’s were ordered, coffee was brewed, and at one point, a terribly unwashed rubber duck was even thrown across the room in frustration. Booz finally snapped, “Dr. John, if we implement this, your partner’s game will require supercomputer botnets to run! It’s a game, not an open heart surgery simulation of the universe…”

In the end, they agreed to disagree. Dr. Doe returned to his academic sanctuary, convinced that Booz was a complete pleb and could not possible get “the bigger picture.” Booz, meanwhile, simplified his orc AI so that it ran smoothly, even on Dr. Doe’s grandma’s ancient PC. Both were happy (and Dr. Doe’s grandma was happy, and Dr. Doe’s partner was happy) but deep down, they knew their acquaintance had forever changed. Whenever they passed each other at the annual TotallyNerdCon, they’d nod and smile, each secretly pitying the other for what they couldn’t see. And so goes the tale of a game developer and an academic programmer, united by craft but divided by vision.

You do your best, when you do it simple.