I don't get it. We use tools to assist in written communication all the time. If someone wants to ask an LLM to check their grammar or edit for clarity or change the tone, it's still a conversation between humans. Everyone now has access to a real time editor or scribe who can craft their message the way they want it to sound before sending it off. Great.

▲

shadowgovt 9 hours ago | parent | next [-]

My personal interpretation of the rule is that if it's human-originated but passed through a layer of cleanup, it's human-originated. For the same reason I'm not refraining from running the spellchecker or using speech-to-text to generate this sentence. "If I could be having my English-speaking nephew type this on my behalf while I told him my thoughts in Japanese, it passes the smell test for human-sourced" feels about the right place to set the bar.

▲

tejohnso 8 hours ago | parent | next [-]

Yes but the guideline states that AI-edited comments should not be posted. It doesn't say it's okay as long as it's "human sourced" or "human-originated".

So if your layer of cleanup is AI assisted, then it's in violation.

Part of the problem I was getting at is that the requirement of "Don't post AI edited ..." is stricter than necessary to ensure the outcome that "HN is for conversation between humans" because an AI edited post is still a human post.

Anyway, I suspect a lot of people are going to ignore that guideline and will feel free to use their "layer of cleanup" whether it's a basic spellchecker or an LLM, or whatever else they choose, and most people aren't going to be able to tell anyway. The guideline is unnecessarily strict in my opinion, but it doesn't matter in the end.

	▲	shadowgovt 8 hours ago \| parent [-]
		My layer of cleanup is AI assisted. It's the spellchecker integrated into my web browser. That was definitely "AI" technology when it originally came out. But I think you and I are on the same page: we both know this isn't a rule that's there to be hard-and-fast enforced because that's completely infeasible. The definition of "AI" is a moving target, as is "generated." It's a rule that's there to have a rule so when the real problem is "Hey, your content is too low-quality but you dump volumes of it and it's clearly following a procedural template" the mods can call that "AI" and justify limiting or banning the account on prior-stated rules. Which is fine, but I'm glad to call it what it is. (One unfortunate oversight: we haven't added "posts sounding like they are AI-generaed" to the "Please don't complain about" set. So expect that to become a common refrain now since the incentives to make the complaint against disliked comments are obvious... At least until that becomes annoying enough to justify a rule).

▲

zahlman 7 hours ago | parent | prev [-]

I'm more interested in the last layer than the first. People should feel fully accountable for what they post, like they could have done it exactly and completely by themselves if they'd simply taken more time.

▲

dmbche 9 hours ago | parent | prev [-]

You can do that anywhere else!