[moondance]

I can relate to the inclination, but so many new insights and moments of inspiration are necessarily confined to that painstaking iterative line-by-line process of real writing. When you are simply prompting and editing, you will fill the page (and it might even sound like “you”), but you will not have that delightful experience of encountering something unexpected along the way to filling it.

[Kye]

There's nothing stopping you from doing that with an LLM. I get more insights refining a draft through prompts than I ever did writing because there's more of it. The end stage of that process rarely sees the light of day because the artifact wasn't the point.

For writing as thinking with trouble starting from scratch, LLMs are the most important technology to emerge in my lifetime. Microblogging filled that gap in a way, but it had too many downsides.

[kranner]

>> you will not have that delightful experience of encountering something unexpected along the way to filling it.

> There's nothing stopping you from doing that with an LLM.

There may be, though. The LLM's initial output may anchor your thinking in insidious ways that may not be obvious at all especially since you're feeling productive. I bet the lack of confidence around starting would also increase over time every time you use an LLM to get over the hump.

[Kye]

Not so far.

I'm not talking about using a default mode LLM with LinkedIn Standard Obsequious Bullshit as a conversational imperative that emerges from simple prompts interacting with the heaviest weights. It pushes back because I told it to and it has redirects around common LLM failure modes, and modes unique to how I use them. That's in a set of instructions I've had a bunch of different models tear apart so I could put it back together better.

I treat it and describe it as a language coprocessor, not a buddy. The instructions are the kernel I boot it with.

[borski]

Yeah, precisely. My "Bobby" knows my voice, but is not me, and is bad at using it. It is aware of all the tropes, and I've built a writing skill that describes, in great detail, how I write. I have also set it up to challenge me, not make me feel good.

Moreover, it's not like I spend my entire writing time arguing with an LLM, lol. I spend more time writing myself and/or doing research on the internet without an LLM, because sometimes they still get things wrong.

In short: it's a tool, not a solution.

[borski]

I think you missed my point. I don't go back and have AI re-edit my drafts, on average. I have it give me some words that are on a page so I can say 'this sucks' and engage in writing myself, as opposed to continuing to stare at a blank page.

The quality of the AI's writing actually doesn't matter, for me, as much as it might for others, as a result. I write my own stuff. I just find AI helpful to activate me to do it.

[abadar]

Your experience of being better at editing than writing makes me think of advice from Alan Moore (of Watchmen fame) on writing. He says to also read bad books, because if you read only good books, you'll never realize if you unintentionally plagiarized. But if you also read bad books, you'll think, "Jesus Christ, I could write this shit," and feel liberated to analyze their writing and examine why their writing offended you.

I analyze the writing of my LLMs and I get really annoyed and frustrated and I think it's marginally improving my communication.

[goodmythical]

>but you will not have that delightful experience of encountering something unexpected along the way to filling it.

I feel like you haven't used LLMs very extensively if that is your genuine experience with LLMs.

Without even tuning the heat to a higher setting, a wide range of LLMs have offered me unique content that I had not encountered previously and certainly was not expecting.

[moondance]

I’ve spent more than enough time writing with LLMs.

“Encountering” is actually a very apt description of the “ah-ha moments” of AI assisted writing, so that was the wrong word for the point I was making.

The joy of traditional writing is that those “ah-ha moments” come from somewhere in you. And I’m not arguing that is preferable out of some sort of anti-AI moralism. Rather, the epiphanies of traditional writing are better because they are informed by your singular experience, the life you have lived, the connections you have made, and all you have gathered along the way. I’m saying that it’s a great disservice not only to a given piece of writing, but to writing as a whole, when those influences are not present to guide the first drafts of the world. Follow the branching paths of your own inspiration to the conclusion, then let the machine take a pass at it. To give it the first crack is to rob your work of the stuff that makes it uniquely yours, and to rob yourself of the experience of invention.