Ask AI: Why do LLMs like to use em dashes so frequently, and more importantly, why do they not put spaces before and after the dash when using it?

Great observation! Here’s what’s going on with language models (like me) and em dashes: **1. Why LLMs Use Em Dashes Frequently** - **Text diversity and structure:** Em dashes are a stylish, flexible punctuation mark in English. They're widely used in many types of writing (from journalism to fiction) to set off parenthetical information, indicate breaks, or add emphasis. LLMs are trained on such varied writing and may "learn" that em dashes are a natural way to break up complex or nuanced sentences. - **Pattern matching:** LLMs mimic human writing patterns. Since many human authors favor the em dash for effect or clarity, LLMs adopt similar habits when generating their own text. **2. Why No Spaces Before/After the Em Dash?** - **Standard American Style:** In US English style guides (like the *Chicago Manual of Style*), em dashes are typically used *without* spaces (e.g., “He was sure—it was obvious—that she knew.”). - **Training Data Bias:** Since a lot of the LLM’s training text comes from such conventions, it learns to produce em dashes with no spaces around them. - **Other Style Guides:** In some British or journalistic styles, an en dash with spaces may be used — like this — but that's less common in American professional writing, so LLMs follow the dominant pattern. **Summary/TL;DR:** LLMs use lots of em dashes because they’re common and versatile in English writing. They don't put spaces around them because that’s the prevailing style in American English and the majority of the data they're trained on. If you prefer a different dash style (e.g., spaced en dashes), you can always prompt the model to format things your way!