Is there any public information about the back-end? Is it doing Markov chains on some suitable corpus or something?
I think the cleverest part of the project is choosing well-known image macros to riff on - the format is so limiting that it's easier to generate convincing text.
8 comments
[ 3.4 ms ] story [ 30.8 ms ] thread[Racially Insensitive Meme Bot]
"DON'T GET BOUGHT TODAY"
I think the cleverest part of the project is choosing well-known image macros to riff on - the format is so limiting that it's easier to generate convincing text.