Agentic data science with the Data Agent Kit
cloud.google.com·7h ago
Data Science
Simon Willison's Blog·25d ago·5 min read
How To Write With An LLM
TL;DR
Using large language models (LLMs) as copyeditors can enhance writing quality, but it's crucial to avoid using their suggested phrases. This approach maintains originality and discipline in writing.
✦ Why It Matters
Establish a strict rule to avoid using any phrases suggested by LLMs to maintain your unique writing voice.
Key Takeaways
- Never use phrases suggested by LLMs to maintain originality.
- Utilize LLMs for fact-checking and grammar checks.
- Recognize the distinct style of LLM-generated text.
- Establish personal writing rules for better discipline.
Full Summary
Thomas Ptacek advocates for a strict rule when using large language models (LLMs) in writing: never adopt any specific phrase they suggest. This guideline serves as a form of intellectual personal protective equipment, ensuring that the writer's voice remains authentic.
While Ptacek utilizes LLMs for tasks like fact-checking, spelling, and grammar, he refrains from allowing them to generate content directly. He believes that LLM-generated text often carries a distinct 'weird smell' that detracts from the writer's unique style.
By adhering to this rule, writers can maintain discipline and originality in their work, ultimately leading to more authentic and engaging content.