LLMs being based on probabilities and these are based on averages, hence whatever the LLM produces must be an average of what its been trained on. So doesn't t

LLMs being based on probabilities and these are based on averages, hence whatever the LLM produces must be an average of what its been trained on. So doesn't the LLM just regurgitating the average writing style of the internet and isn't the author just an average writer and hence the em-dashes.