ZH version is available. Content is displayed in original English for accuracy.
Advertisement
Advertisement
⚡ Community Insights
Discussion Sentiment
72% Positive
Analyzed from 1314 words in the discussion.
Trending Topics
#writing#thinking#write#don#code#text#reading#claude#better#requires
Discussion (31 Comments)Read Original on HackerNews
- You can use AI to look through docs and do research.
- You can use AI to structure your argument.
- You can NOT use AI to actually write the prose.
- You can use AI to proofread, especially for how effectively you've communicated.
I don't know, man, that sounds a lot like using AI to write. Your writing probably says "load-bearing" less than Claude's outputs, but I'm not sure it's much less AI.
Structuring arguments: bad idea. Here you're having the model do your thinking for you. The problems that occur with having it write text recur at a macro level; too-obvious patterns. Readers just want to know what the prompt was.
Writing the prose: obviously not.
Proofreading: I think this works pretty great, as long as you have a structure to how you're doing it. "How do I make this better or sharper": bad, you're having the model structure your arguments for you. But for a predefined rubric of stylistic things --- writing you know how to do, that's just tedious to check manually? Hard to see a downside.
I used it to edit a document I'm working on, where I accepted about a third of its suggestions. I feel pretty good about it.
If you have a Claude account you should be able to access the artifact I built here: https://claude.ai/artifact/RK7XfnaoxVV1Uji84ETaLr - it uses "Claudeception" - the feature of Claude artifacts that lets them use Claude prompts - so it will be billed against your Claude account (at a cost of probably a fraction of a cent per run.)
I don't write code for my colleague's pleasure, I do it to solve a problem as fast and efficiently as possible.
So LLMs allow you to work at the speed of reading, if not faster. So you can try a paragraph here, completely rewrite the tone. Start a chapter in 1PP and just have it rewritten to 3PP. And sofort and so on. The number of new transformations you can do is endless.
What do a lot of people do? Write a one line prompt and show everyone the oneshot result.
Ok fine, sometimes you don't need to write amazing prose. But you can also be like the early writers who adopted the word processor. If you put the same amount of elbow-grease into using an LLM as you did writing before, you can get some pretty cool results.
It’s easier to rearrange and rethink and tinker. Hard to imagine now but it was more efficient in the days when you would have to pencil out a memo and have someone type it up, you would make corrections and the final copy would be done in a couple passes. This has been studied, efficiency has actually gone down, in favor of immediacy and direct control.
What it frees you from is organizing your thoughts ahead of time, which takes practice and discipline. Using an LLM is just taking that further, you can avoid organizing your thoughts and ideas a bit more.
If someone uses LLM to help wording their idea in a way that's easier to understand, I am fine with it as long as they don't copy-paste it verbatim into their blogpost.
I am getting a little ranty here, but the usual LLM-isms like "not this, but that" or "that's the real footnote here" simply add zero value to the text, and I can't imagine anyone who has proofread the LLM-generated text even once will opt to keep those sloppy paragraphs. If I see them, I can't help but to think that the author had oneshotted the text. And it strongly suggests to me that the author has put zero effort into their work.
Fundamental to exchanging ideas is discovering which ideas should be exchanged and how to express them. The process of writing itself is that exercise. If you use a tool to outsource the thinking then you're not thinking.
Why do people feel the need to invent the past that never existed?
Okay, but reading also requires substantial thinking. Moreover, we can say the same about, say, coding. And I am sure there are people who will say, yes, don't use AI to code. But I do, and the results are amazing, and I still think very deeply: in fact, I spend more time with AI coding (and thinking) than before as I do not get hamstrung on, how do I convert this Word doc to PDF in code again? I spend more time in higher order thinking. Same with when I read text written by an AI. I don't believe that writing is any more sacrosanct that coding. And reading and adjusting AI output requires substantial thinking.
I don't think anyone would argue reading a book takes as much thinking as writing one…
https://www.scientificamerican.com/article/why-writing-by-ha...
Basically an automatic "second-pass" which I would've caught had I re-read it, but automatically.
I first write the thing myself, then feed it through Deepl Write to see how the AI would write it. It's been very helpful.
I write everything else myself. However I don't know how I feel about spending my time on Earth typing personal responses to AI-generated messages. Do those people deserve my time and energy?
The reaction was pretty much silence and most people probably thought I was some angry, arrogant, anti-ai purist... which isn't the case. It can be acceptable to use a language model in code generation but requires new forms of discipline and tooling.
But for for just regular old writing... Should never, ever be generated. I'll likely never change that stance.
> Writing requires thinking, don't let an AI do the thinking for you.
And then the author goes on to admit that they use AI for coding. Does coding not require thinking or what?
I agree that we shouldn't be delegating thinking to AI, but stopping at writing seems disingenuous.
- sure its good to ask AI feedback about your module or code review or identify vulnerabilities or better ways of doing your existing code
- but maybe stop asking AI to generate entire projects from scratch?