LLM Writing Tropes.md
tropes.fyi
March 7, 2026
14 min read
🔥🔥🔥🔥🔥
64/100
Summary
tropes.md is a comprehensive file that catalogs common AI writing tropes to help AI systems avoid repetitive patterns. Adding this file to an AI's system prompt enhances its writing quality by steering clear of identified tropes.
Key Takeaways
- The file tropes.md catalogs common AI writing tropes to help AI systems avoid overused patterns in their output.
- It identifies specific word choices and sentence structures that are frequently associated with AI-generated text, such as the overuse of adverbs like "quietly" and phrases like "serves as."
- The document provides examples of writing patterns to avoid, including negative parallelism and self-posed rhetorical questions.
- The creation of the file was assisted by AI, emphasizing the collaboration between human and AI in improving writing quality.
Community Sentiment
Positives
- The identification of common LLM writing tropes provides valuable insights for researchers, enabling better understanding and improvement of AI-generated content.
- The mention of specific overused words like 'tapestry' and 'camaraderie' highlights the potential for quantifying LLM writing styles, which can enhance model training.
Concerns
- LLMs often ignore explicit instructions, suggesting that their pretraining and reinforcement learning weights are too dominant to be effectively overridden by prompts.
- The overuse of certain writing patterns, like 'It's not X -- it's Y', leads to a perception of AI-generated content as lacking depth and originality, frustrating readers.
