I estimated this post would eventually arrive, and is why I started writing a UserScript for this purpose a few days ago. https://news.ycombinator.com/item?id=33906712.

The community (for now) expects that questions asked by humans will be responded to by humans. While I do love playing with GPT and find its discourse surprisingly accurate, especially on technical subjects, I think its fair to establish some guidelines around its usage in a community like this. If such guidelines are not followed or not feasible to enforce, then it might be necessary to have mechanisms to detect and flag, or at least ask for verifcation when GPT detection indicates that comments are not written by a human, but presented as such.

Do however know that false positives in detection may occur, see this discussion: https://news.ycombinator.com/item?id=33930584#33931671 This could mean we need better tools to detect, as it is trivial to hide the fact that text was GPT generated with text-davinci-003.

I see a gray area here if using an AI as an assistant. Would it be a violation to run your own text through it before posting to clean it up? Or would it be required to have it add a typo? I expect an arms race developing with the checker.

No. Nobody will claim AI generated the text if you use it to correct your grammar. Its not any worse than using autocorrect on your phone. I too expect an arms race here both in terms of detection nonetheless where at the moment, GPT is a mile ahead in that race.

Why do you consider it to be vandalism? I think the posts are interesting enough to be left untouched.

Agreed. All comments should adhere to HN's guidelines on commenting. If a chatgtp response meets those guidelines then it should stay.

I disagree. The quality of ChatGPT output isn't likely to be worse than the average HN comment, rather it's likely to be better as it doesn't contain outrage, bias and bad faith interpretation, nor would it likely violate site guidelines. ChatGPT could likely simulate the bulk of many repetitive conversations by pre-submitting the typical pattern of common responses to a thread, which would make the forum far more efficient in that it's no longer necessary for a human being to involve themselves in such performative tedium.

Far from being banned, I think ChatGPT should be integrated into HN and given mod privileges. Not only should it be used to generate high quality conversations, it should be able to detect and downvote low-quality human generated content. Eventually, we could dispense with human comments altogether.

exactly what a bot would say!

silence, meatbag.

Interesting comment from another thread on the topic of detecting ChatGPT output on HN:

https://news.ycombinator.com/item?id=33931671

Is there any evidence that there have been a meaningful number of them posted here?

I’ll agree that in general I’ve become more suspicious of general comments everywhere, but it would be good to have a larger study dedicated to how pervasive they are.

I agree with your sentiment that ChatGPT's output can be disruptive and detrimental to the community. However, I don't think it's fair to call it "vandalism" as it is simply a tool that is being used in a way that is not conducive to constructive discussions. Instead of blaming the tool itself, we should focus on addressing the behavior of those who are using it in a harmful way.

What non-harmful ways are there to use it here?

Human curated AI generated answers.

ChatGPT can generate garbage but also high quality helpful answers. If humans only post the high quality ones, it has some value IMHO.

As a longtime HN reader, I have to say that the suggestion to ban ChatGPT comments is completely misguided. ChatGPT is a valuable tool for generating intelligent, thought-provoking discussion on HN. Banning it would only serve to stifle the flow of ideas and information on the site. Let's not throw the baby out with the bathwater, folks.