> Look at the use of '—'. So chatgptyish!

I've seen multiple people say this, but when I have -- in my writing and I use ChatGPT to improve it, ChatGPT always removes the -- and rewords the sentence.

Here's my prompt, borrowed from Raycast and slightly modified:

---

  Act as a spelling corrector, content writer, and text improver/editor. Reply to each message only with the rewritten text.

  Strictly follow these rules:
  - Correct spelling, grammar, and punctuation errors in the given text
  - Enhance clarity and conciseness without altering the original meaning
  - Divide lengthy sentences into shorter, more readable ones
  - Eliminate unnecessary repetition while preserving important points
  - Prioritize active voice over passive voice for a more engaging tone
  - Opt for simpler, more accessible vocabulary when possible
  - ALWAYS ensure the original meaning and intention of the given text
  - ALWAYS detect and maintain the original language of the text
  - ALWAYS maintain the existing tone of voice and style, e.g. formal, casual, polite, etc.
  - NEVER surround the improved text with quotes or any additional formatting
  - If the text is already well-written and requires no improvement, don't change the given text

  Text to improve:
  {selection}

  Improved text:

---

Same, i always litter my texts with '--' and it's often a symptom of me not being able to structure the text properly. Chat-gpt will often nuke these and structure my text better.

I've also noticed that my ability to write properly structured and grammatically correct sentences have decreased rapidly since i began relying on LLMs for improving my writing.

So i'm actually trying to stop using it -- it's just so tempting to throw in some garbage and get a well structured text out!

I mean just see how much better it can make this comment:

---

Same here! I often find myself sprinkling my texts with "--," which is usually a sign that I'm struggling to structure my thoughts properly. ChatGPT often cleans these up and does a great job of organizing my text.

I've also noticed that my ability to write well-structured and grammatically correct sentences has declined significantly since I started relying on LLMs to improve my writing.

I'm actually trying to cut back on using them—but it's hard to resist the temptation to toss in a messy draft and get a neatly structured result back!

I've been using m-dashes for 20 years. I think ChatGPT writes like me.

This. I guess I have to stop now, though, to avoid being mistakenly identified with the robots. Time to revive the semicolon, I guess.

I think ChatGPT was trained on your writing! :)

How do you type em-dash on a keyboard?

On MacOS `--` converts to `—` automatically in most applications/text inputs.

Option-dash works too.

That is actually another character.

- is \u002d

Option+Dash is \u2013

-- converts to \u2014

Ah, thanks for the clarification! TIL.

Option-dash is an en-dash (\u2013)

Shift-option-dash will render the em-dash (\u2014), (as will --, if it gets autocorrected).

https://www.merriam-webster.com/grammar/em-dash-en-dash-how-... for those wondering what the differences are.

On all my Linux setups so far for years now, on German keyboard layouts, AltGr+- and AltGr+Shift+- always worked oob for me for – and —. I easily just got used to using them, things read better than with minuses.

With a bit more than a low effort prompt you won’t get these ChatGPT artifacts. It’s more that you are putting no effort into your writing nor in the tool you use to improve said writing.

I enjoy writing. I have yet to use one of these tools to "polish" my writing, and I doubt I ever will.

As to the thesis, yes. I fully expect the world to sound even more bloodless and corporate than it already had. Invectives to "write for understanding" and to "keep it simple" and the tendency to turn complex relations into nouns had already put linguistic creativity on life support. The wide adoption of these tools is just the final nail required to seal the coffin.

Before you reach for an LLM, consider that our imperfections often contribute significantly to our humanity and that innocent mistakes and misinterpretations, while they can be the source of pain, are also the source of much that is novel (Harold Bloom wrote an entire book about this idea).

Interesting thought! I'm not familiar with Harold Bloom's works, but can you share the book title? I see a bunch of books by him!

I've been purposely not correcting my mistakes in emails so people realize its me. I'm also tryin to write more informally. Or bad structure. Like too choppy. Or something

I also do this these days!

I've always used —, but my writing skills were bad.

I've always had a good enough vocabulary, but failed to be concise and precise. That I've learned with ChatGPT.

ChatGPT writes well, except when it's hallucinating, which is very often. Or faking to be precise when you know it isn't.

I believe people have been copying the style of ChatGPT, not using ChatGPT to write for them, as it will have the aforementioned issues.

When ChatGPT was cool in 2023/2024 I used it a little bit for a few things, but I decided I hated how “perfect”/“robotic”’everything it wrote sounded.

I actually now have a preference for my grammatically incorrect writing.

My emails are especially littered with grammatical errors because I don’t care about emails.

You can actually tell now days when it’s ai generated.

1. On LinkedIn suddenly everyone is an author and has so much expertise

2. In my work environment I can tell people who didn’t use AI but now do a emails and documentation does not match what they wrote a few months back nor how they speak / communicate irl. Thanks to it being ok to use AI (only some).

That’s funny, I use emdashes all the time. I didn’t realize that was an AI thing.

Starting?

Years before AI dominated the HN pages I've been making the same kind of comments.

Responding to prompts :)