tl;dr: I trained an AI on names of hacks, and now it can generate new hack names based on it.
Basically, it's a neural network that I fed with a list of hack names I put together from SMWC, then let it train for a while based on it. Without training it produces complete gibberish (as AI does), but over time it fine-tunes itself to make the output fit the training data. You can think of it as learning: the AI picks up on the patterns and learns to reproduce them.
Not going to bore you with any more details because a) I wrote the code a few weeks ago so I forgot the details and b) I'm not a machine learning expert (I barely knew enough to write the code from scratch-ish).
The cool thing is I put the result online, and you can take your very own look at all the wacky new hack names!
(these are not generated on the fly, but chosen at random from a huge list of pre-generated names. If you find a Very Bad Word in there it's not my fault but I can remove it.)
Huge shoutout to Underway who inspired this project: he trained his own AI on VLDC level names
and got hilarious results
. I actually think my AI's results are a lot less funny, but hey, it's a thing that exists.