Ask HN: Should bots actively be banned on HN

[podnami]

I saw the top comment on https://news.ycombinator.com/item?id=43970837 and started thinking if there are rules against using bots to write comments. Personally I wouldn’t have spotted it but some people pointed out the use of dashes and then I felt duped. What are your thoughts on this? Catching ChatGPT generated comments vs human responses isn’t easy, but in the case of that specific comment I felt “double duped” as the comment also described an experience from real life.

[toomuchtodo]

https://news.ycombinator.com/item?id=33950747

[Tomte]

> but some people pointed out the use of dashes

That’s a stupid heuristic. I use those dashes all the time, I even have an AutoHotKey macro in Windows for en and em dashes, ellipsis and other characters.

[MarcoDewey]

What if - some of us - just like to use excessive - dashes when writing -- ?

[beAbU → MarcoDewey]

You are not using em-dashes so you check out. Please proceed.

[onecommentman → MarcoDewey]

What’s - excessive - about that anyway? It’s - just - the written equivalent - of William Shatner’s acting style. Worked for him.

[krapp]

The popularity and quality of AI is going to make that impossible in the future, if it isn't already. Some significant number of comments here will be from bots, they will be less and less distinguishable from human beings, and that number will only increase over time. None of these tricks and heuristics people use to try to spot AI will be useful.

And a significant (and also inevitably increasing) part of the human remnant of HN will prefer it because as far as they're concerned text generated from an AI is superior in quality to human interaction. They aren't here to interact with humans so much as consume a product.

Just teach yourself not to care. You'll never know, don't be embarrassed by it.

[anenefan]

As much as I dislike A.I. generated answers, using it as a tool to fix sentence structure or styling a passage to meet a certain level of writing required for a format will at some future date be accepted as much as we accept automatic spell checkers. It'll probably be more accepted across the web once generated BS is sorted out.

Personally, I'm against all the new BS of no double space after a sentence and no gaps around em dashes. Not doing so removes a lot of options for those who use string searches to hunt out certain things.

[gus_massa → anenefan]

I use the spellchecker and I'd like a better gramar and spell checker like in gmail. Is gmail using an expert system or AI? I think there is no problem with that.

But a 100% AI generated comment is bad.

[mtmail]

That's already not allowed. Contact the moderators (link in footer) if you spot anything. I've reported users in the past, whole networks, of clearly AI generated comments.

[onecommentman]

Umm, if you’re not getting an “uncanny valley” vibe from the response, you should probably make a working assumption it’s a person and not AI. How would you change your response to the comment in one case or the other? Both can lie, manipulate, tell half-truths, etc. The only difference I can see is if you believe you can prevent a living person from making or continuing to make a big mistake by responding, then spend some time on the response. If it’s a bot, you don’t. You submit interesting relevant factoids regardless because the non-bots (TM all rights reserved) might like them.

BTW, using stylistic features in writing to ID bots might be low-yield. Style changes are trivial to implement in any automated system, and they can be used to stylistically “blackball” particular human contributors. “I’ve heard that using ‘grey’ instead of ‘gray’ is a sure sign of a bot!” That sort of thing.

[bediger4000]

I loathe bots more than I loathe spammers and SEO consultants. Ban them. Delete them. Scorn their use. Shun those who use them.

[benoau]

Those double-dashes also indicate an iPhone was used, I frequently use it but only when typing on my phone.