what is the most effective/economical way of strong arming the model out of adjective/adverb/simile abuse and reddit humor without spending half the chat bullying the bot/calling it retarded?

i just want it to trust the users intelligence and its own ability to convey meaning/set a scene using little details and without having to be so on the nose or heggin quirky!