LLM writing smells different. Faint wiffs of “you are a helpful assistant” with strong notes of “car salesman”. I call its patterns LLM-isms. Maybe some of these resonate with you?
LLMs love to tell you it’s not this, it’s that; explanation by way of counterexample.
“Dashi’s job is umami lift and clarity, not dominance.”
ChatGPT teaching me to make soup
“…the problem isn’t the lights, it’s the lack of intentional cable management. Right now the wires read as accidental instead of designed.”
ChatGPT rating my light setup
As a helpful assistant, LLMs want you to want to keep going.
“If you want, I can help you identify a hawk you saw—just tell me the size, colors, where you were (city/woods/open fields), and whether it was flying or perched.”
ChatGPT birding with me
“If you want, tell me size, flatweave vs pile, any stains or smells, or drop a photo and I’ll tailor this exactly (including whether a controlled outdoor wash is safe).”
ChatGPT cleaning my rug
Signs of AI writing on Wikipedia lists more.
As we use LLMs to write and learn to write from LLMs, personal writing is bound to pick up on these scents. It'll be an unfortunate convergence because I find this style a bit tiring.