Oh hai Spectral Slug: AI makes new D&D monsters
Research scientist Janelle Shane trains a neural network to create D&D monsters. Behold Hatfright, Giant Fraithwarp, Wendless Woll and Marraganralleraith.
Artificial intelligence has advanced in interesting ways, and often peculiar ones.
AI can transform an episode of the retro Bob Ross painting show into a bizarre, kaleidoscopic acid trip and reimagine Nic Cage as Indiana Jones. Now it seems AI is pretty good at playing the role-playing fantasy tabletop game Dungeons & Dragons.
Janelle Shane, a research engineer for an optics company who also likes to experiment with neural-network programming, trained a machine-learning system to create new monsters for D&D.
Shane gathered the names of 2,205 creatures from the second-edition Dungeons & Dragons monster manual. Her neural network then transformed those names into new imaginary creatures like a Wolfworm, Spectral Slug, Jabberwont, Burglestar and Marraganralleraith, to name just a few.
Of course, you can't play D&D without coming across some dragons, and the neural network came up with those too, including Curple Lard Dragon, Will O'Dragon and the dreaded Death Seep Dragon.
If an AI making bizarre D&D creature creations wasn't cool enough, Shane also created an algorithm called a recurrent neural network to generate D&D spells like Barking Sphere, Hold Mouse, Gland Growth, Summon Ass and Shield of Farts.
Up next, Shane is hoping to train her neural network to generate D&D character names and backstories. She's asking D&D players to submit their characters' names, races and classes.
In addition to her D&D AI adventures, Shane also trained a machine-learning system to generate new phrases for Valentine's Day candy hearts.