[asdff]

Put it this way, if the ai article is indeed so good, why store the article in a dusty long form output mode from a soon to be obsolete model? Just give us the prompt, and in 6 months with some newer, bigger, better model, we can feed that prompt and get an even better article out of it.

[maxaw]

to play devils advocate - i doubt the articles that end up on the front page are one shotted (theres probably a sequence of back and forward refinement). and in any case, i feel the avg reader would actually not prefer to read the prompt, which would be very information dense

having said that, i'd still much prefer a norm of including prompt history with the article, or the codebase for that matter, so people can choose for themselves :)

[throwaway219450]

This is a fascinating thought problem.

On the extreme end you have recipe blogspam, with the tropey life stories and Amazon affiliate spam that precede the meat (pardon the pun). If that pre-filler is inconsequential, why not have an LLM parse the semantic markup - which 99% of recipe sites use for Google SEO - to produce custom content for you? Like a history of the ingredients, or better instructions with an equipment list tailored to your kitchen and timings/temps for your specific oven?

I can see there being a middle ground where you publish some sort of context that a user’s LLM can generate the content from. Is it fundamentally any different from dramatized non-fiction? Think of books like Operation Mincemeat where the facts are thoroughly researched, but the author uses a lot of artistic license to tell the story.

But to your point, if there is back-and-forth, I wish people would pay more attention to the obvious traits of LLM writing. Not everything needs to read like it’s a snappy editorial, but I sense we’re still in the early days where people have been handed this technology and are simply excited that they can pump out 1000 words in a coherent narrative.

[maxaw]

Your middle ground idea is interesting. And maybe inevitable in some ways, if the llm becomes even more of a universal interface to the outside world.

Regarding llm writing, it is disappointing. Especially as people could use it in the “opposite” direction, using the tirelessness of the llm to experiment with the communication with the aim to make it as clear as possible for the reader…

[ruszki]

Interestingly, the average reader wasn’t that annoyed by the magnitudes higher information density of 2 decades ago.

Also, the information density plummeted because of ads, not because it was demanded by readers.

[Leynos]

You'd need the whole edit tree along with all the prompts used along the way, which most people are not yet set up to capture.

[dgellow]

You could also just type the article with your human hands, using your human voice, after having done your LLM interactions. That’s really the best minimum to show that you respect your human audience, use your own voice

[jqbd]

I store all prompts as docs, but I doubt I can rerun the whole prompt history with a better model and get sizable return for the cost, personally. EDIT: I'm referring to a small codebase not articles tho

[hamandcheese]

A lot of my PRs are the result of hours of back and forth between an AI and myself. In many cases the sum total of my prompts is much larger than the final output.

[petesergeant]

The fallacy here is that you can't reduce the wide range of things that “AI generated” means into a single flag. As for why not “show us the prompt”: https://sgnt.ai/p/prompt-is-not-the-work/

[spopejoy]

> Just give us the prompt

Two showstopper problems with this:

1. Many of "us" don't like interacting with LLMs more than we have to, and view reading as a pleasure that would be extinguished instantly at a chat prompt.

2. It lazily implies that the essence of an article is so easy to communicate that it can be compressed into a prompt. There's a reason for titles, subtitles, open sentences, hero images, paragraph openers, etc etc: they pull you in, establish tone and voice and so forth. Where are the prompts that can entice a reader to even skim it much less plug into their favorite chatbot?