Hacker News Reader

Comment by nirava | Hacker News Reader

I've found the only good way to write with LLMs is to absolutely not use it for the first few drafts, the using it as a tool to list specific kinds of problems in text (word repetition, unnatural hard to read sentences etc) and approaching those points with caution and skepticism. Even copy pasting a single sentence, I've found, is a slippery slope.

View on HN

Just write, write your own words, then create an effective communication persona within your LLM's non-reasoning (non-Chain-of-Thought) capabilities and have it critique your writing after describing your target audience. Then, write your own revision following the critique. Using this, it is not possible to write "AI slop" unless your own writing is already in the style of LLM output. Which I get accused of when I write using my formal voice, but fuck that sorry for being well educated.

This is close to my approach. I'll use LLMs as a read-only editor for my writing. Sometimes I follow its advice. Sometimes I don't. Other times, it discovers things I want to change, but not in the way it describes.

One thing that sort of worked for me is to write the skeleton myself, i.e. the general ideas and how they fit together, the overall “flow” of what I want to say, then have an LLM fill in the blanks (I either point it at some context.md or have it ask me questions when details are missing).

It’s kind of like designing the high level software architecture yourself and have the LLM write the code for each component.

Not bulletproof, requires some iteration, but miles better than what it would produce on its own.

Try using them to have a long discussion and provide it input data, then have it emit a concise bulleted outline given the points you want to make. Then you can rejigger the outline efficiently as needed and write the text yourself.

Exactly- without constraining the input, you're guaranteed to get slop. First-draft AI copy is usually 95% to-cut, 4% to re-sequence, 1% to keep. Use it to think about iterating, not sweep the solution space blindly.

Reply on news.ycombinator.com