This insane Dungeons and Dragons report will blow your mind

Home » News » This insane Dungeons and Dragons report will blow your mind

A clever researchers used a neural network to create a pretty wild selection of creatures for Dungeons and Dragons.

If you are not familiar with the strange and beautiful world of the popular board game Dungeons and Dragons, it has tons of strange creatures like owlbears or pertytons, which are eagles with the head of dear, as noted by Kotaku. But as it turns out, D&D creatures that were created by a “neural network” have taken weirdness to whole new levels.

Research scientists Janella Shane said in a blog post that she had created an algorithm that used a recurrent neural network to generate more spells and came up with some interesting results, like Summon Ass and Shield of Farts. But she did not stop there, using the neural network to create a weird range of creatures.

Here just a few of the odd creatures, as listed on her blog: vampire bear, kick spirit, purple bird, spectral slug, crystal human, fire brain, dunebat giant, cloud of chaos, goblin dog, plant hound, fire undead lake man, walfablang, dome animal, spectral woof greepy, memeball, and marraganralleraith.

“Neural networks are a type of machine learning program that learns from examples they’re given, rather than relying on a human programmer to invent rules,” Shane explained in another blog post. “In an earlier experiment, I trained a neural network to write new names for Dungeons and Dragons spells based on a list of 365 examples. That’s a really small dataset for a neural network to work with, and I ended up struggling to find training parameters that would strike a balance between word-for-word mimicry of the original list of spells, versus a series of completely made-up words. By filtering extensively through the nonsense, I was able to come up with a short list of interesting new spells. (My favorites were Barking Sphere and Gland Growth).”

She used datasets from the 4th edition list of spells, for example (which included 1,300 spells in all), and then attempted to train a better performing neural network.

Daniel J. Brown

Daniel J. Brown (Editor-in-Chief) is a recently retired data analyst who gets a kick out of reading and writing the news. He enjoys good music, great food, and sports, with a slant towards Southern college football, basketball and professional baseball.

Scroll to Top