iA, the software firm that develops the excellent iA Writer and iA Presenter , shared some tips and ideas on how to use LLMs in a writing process without running the risk of losing your own voice.
Three of my favourites:
Give ChatGPT an outline and let it write for you. Let's turn the tables and have ChatGPT prompt us. Tell AI to ask you questions about what you're writing. [...] Once you have enough to work with, paste your answers into your text editor. You have just written your first draft without cheating.
This is also a suggestion that's easy to generalize for other processes as well, where you want to be nudged in your thinking process.
Worried you missed something obvious? You can use ChatGPT as an editor to list potential flaws, like long words, clichés, or factual errors. [...] Use "list" rather than rewrite. Rewriting can lead to all sorts of unwanted changes. They can be hard to spot because ChatGPT seems so smooth, grammatically and orthographically.
This is gold! I have spent much time comparing the output of an LLM with my own writing to find the differences. This is helps with that, but also makes sure the text still in your style.
What weaknesses have I missed in my writing? Ask ChatGPT for progressively harsher critiques of your writing. That can be unpleasant, but that's how you find and fix your blind spots, without the shame.
This is also good. But here I would like to add a tips of my own: I've recently started to feed both my article and my research material to LLMs, with a prompt to compare the two to suggest if there are any important aspects I've either forgotten or misinterpreted.
Visit the source: ia.net
- Previous: Introducing Raycast Focus
- Next: Building effective agents