The bot currently scraping AO3 for content to train an AI is going to have the most awful spelling because its not professionally edited, so American English and British English are often sprinkled through the same fic.
Color and colour may appear in the same text.
It's going to have such confusing garbage input, it's hilarious.
plus so much adult material. so much. why does it keep predicting "tentacle' is the next word?