Maybe AI could solve for people who use the “laugh” emoji as an unlimited proxy for the “thumbs down”?
(the latter of which has limited uses per day)?
(the latter of which has limited uses per day)?
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Turn over moderation duties to AI
- Thread starter Ben J.
- Start date
- Sort by reaction score
For once we agree! Unfortunately this forum is loaded with children who lack the ability to debate with words. Luckily they are all +1s to our meaningless reaction score so it is really a net win for the "victim".
Based on personal experience, I doubt that is true in most cases.
There are too many moderators on this forum, with many individual decisions.
The last two posts are spam https://forums.macrumors.com/thread...und-music-from-a-track-to-just-leave.1130354/
This should have been deleted https://forums.macrumors.com/threads/vmtek-virtualization-platform-coupon-giveaway.2418688/
And threads like https://forums.macrumors.com/threads/digital-life-destroyed-from-hacker-malware.2418387/ are probably allowed just to generate traffic.
I just checked, and none of the posts you linked were ever reported. So no mod ever reviewed the posts.
I'm not following how you think this proves mods here don't work as a group?
Nonsense. You bring an issue of fallibility and yet even the state of the art AI models are as fallible as humans if not more.
I’m sorry, I should have inserted a pause between the two.
My opinion refers to request 37461.
I agree that AI can not moderate a forum. The examples illustrate the variety of content that needs to be moderated and the various motivations that have to be taken into account.
The comment from 2022 in the 1130354 thread was made by a new account and it contained a link. A moderator must have approved it. Another moderator has now deleted it
Any public statement from the moderating team regarding threads like 2418387? No comment is fine
I think I understand why @Ben J. started this thread. I makes no sense to me either and I consider that a better decision would have been to move the thread 2418738 in the Web Design and Development section.
Far from perfect
UnitedHealth uses AI model with 90% error rate to deny care, lawsuit alleges
While humans are apt to make mistakes, this weakness is offset by collaboration, i.e., the staff working together to discuss and formulate an action.
I would pick humans managing the site 100% of the time over an AI
A frequent complaint by moderators is that there aren't enough of them to do the job, they're volunteers, and they can't moderate 24/7. So yeah, moderation could very well be supplemented with AI, possibly even replaced in the future. Maybe then the site's 'Contact Us' page would actually yield a response.
AI is created by people and contains the biases and opinions of the people that helped in its creation.
AI is much better than people when it comes to searching through data using guidelines that people specify.
AI would be terrible in making moral decisions on that data.
AI could help the moderators by flagging post but, the moderators are needed to make the decisions on what to do with the post (i.e. I think that AI could help the moderators but that the moderators are needed for the decision making).
Pretty sure thats not how it works here, except maybe for a wild fire situation. It seems to be a bit of the wild west with moderators acting as sole judge and jury. And the rules are quite clear, don't bother appealing a moderators decision. Sure that's expedient but can create friction.
To be clear, the moderator's job is a thankless chore, and given how toxic the threads have become, quite a bit of work. The reality is while the rules are in black and white, enforcement does appear to be uneven. You have to go really far off topic before being called for it. There are still plenty of one word replies. And the insults do fly. All against the rules, but not all moderated.
In fact, I dare say commenting at MR is a lot like speeding on the highway. Some observe the speed limits, most do not, and only a few are pulled over for violations.
To the OP's point, IF it was possible to have a fairly nuanced AI that would be more active and really enforced the rules, I would see that as a good thing. Problem is, AI is often a matter of training, and requires a lot of examples, and I don't see that being practical here.
This a common misconception that mods see all posts, so if a rule breaking post exists, it is because a mod allowed it. This is definitely not the case. If you see a post that clearly breaks the rules, the more likely explanation is no mod saw it among the thousands of new posts each day. If you see posts like that, please report them so we can take a look.
Thank you for clarifying, I inferred that from your previous reply. I am loathe to report things because it feels like tattle telling and I have seen people use reporting almost as a means of trolling. And also it does seem subjective, and I dont want to waste moderator's time. But I can see the other side of the coin, don't complain about how something is done (or not done) if you aren't willing to help be part of the solution.
Perhaps you can clarify another thing, you use the word 'we', and someone else suggested it was a group decision. Is that the norm, or can moderators make a decision on their own?
Mods here can and do make decisions on their own if the issue is clear cut. For example, "You are a moron", does not leave much up for interpretation and would normally just be handled by a single mod. If something is less clear cut, it is discussed by multiple mods to reach a consensus.
AI would not be an improvement on the humans we currently have moderating. AI does not do well in understanding context, humour or the flow of a conversation. How can it? It's just code.
Let's leave AI at what its good at instead of trying to shoehorn it into roles that just won't work.
Let's leave AI at what its good at instead of trying to shoehorn it into roles that just won't work.
Might as well. I was a mod on various sites over the years and leaving that to human interpretation is a time consuming *&^%$show. With ai, when users get b& by a poorly written piece of software because it can't decipher context in internet posts any better than people can, the dysfunction is something we're all used to, so there's no mistaking it for being personal. The aggrieved is then just irritated that software still sucks, and can now look forward to a week off posting bc the dumb bot is dumb. ...rather than feeling the need to climb up on their horse and burn some mods village down & annoying everyone about it forever.
On one site, we set up some convenient wordfilters that just replaced all the cursewords, nazi buzzwords & slang that was trending on 4chan and Fox News, so it effectively mixed users manifestos and screes up into absurdly comical jibberish before automatically giving the poster a week-long ban for having triggered the wordfilter. The effect was the same as having an ai bot in charge, no ai necessary, and was by far regarded as the fairest and most liked moderation strategy on any site I've ever been on. But we also just thought of discussion posts as that, ephemeral and following the natural myriad directions human discussions go with a discussion tree, so in that case there's no need or desire to continually whip users in line to stay on a single topic in a straight line. These days people seem to think about internet discussion threads as if they'll eventually be compiled some valuable timeless tome to stand the test of time as a coherent literary work, so if that's the expecation, then yeah it needs active co-authors and editors.
On one site, we set up some convenient wordfilters that just replaced all the cursewords, nazi buzzwords & slang that was trending on 4chan and Fox News, so it effectively mixed users manifestos and screes up into absurdly comical jibberish before automatically giving the poster a week-long ban for having triggered the wordfilter. The effect was the same as having an ai bot in charge, no ai necessary, and was by far regarded as the fairest and most liked moderation strategy on any site I've ever been on. But we also just thought of discussion posts as that, ephemeral and following the natural myriad directions human discussions go with a discussion tree, so in that case there's no need or desire to continually whip users in line to stay on a single topic in a straight line. These days people seem to think about internet discussion threads as if they'll eventually be compiled some valuable timeless tome to stand the test of time as a coherent literary work, so if that's the expecation, then yeah it needs active co-authors and editors.
Last edited:
AI moderating human conversations on contentious issues sounds, to me, like one of the hardest things AI could ever be asked to do, and a task where bad AI would do maximum damage.
No thanks.
I had mixed feelings about the PRSI forum being shut down, because I thought it was remarkably well-moderated (and thanks again to all the mods who did that incredibly thankless job). One of the few places I ever ventured to talk politics on the internet.
On the other hand, the ability for the internet to host and spread toxicity is pretty obvious these days, and I think there is a reasonable argument for an increasingly compartmentalized web. Back in the day every forum on every topic had a little subforum for politics and other off-topic threads. These days if you want to lock horns about politics, you have Reddit and the socials. Go there, it's quite brutal and in my opinion largely unedifying. But no AI is going to be able to wrangle those cesspools.
No thanks.
I had mixed feelings about the PRSI forum being shut down, because I thought it was remarkably well-moderated (and thanks again to all the mods who did that incredibly thankless job). One of the few places I ever ventured to talk politics on the internet.
On the other hand, the ability for the internet to host and spread toxicity is pretty obvious these days, and I think there is a reasonable argument for an increasingly compartmentalized web. Back in the day every forum on every topic had a little subforum for politics and other off-topic threads. These days if you want to lock horns about politics, you have Reddit and the socials. Go there, it's quite brutal and in my opinion largely unedifying. But no AI is going to be able to wrangle those cesspools.
AI is merely the latest in a very long line of human inventions that have sparked fearful debate over whether it is good or bad for humanity. The answer is, in every case, both.
AI isn't even close to be competent at doing tasks like moderation. Maybe a decade from now, if it hasn't utterly exhausted our planet's energy resources.
Last edited:
AI moderation??? worst idea ever. Ebay uses AI for it's reporting of posts. I've reported posts that clearly break ebay's listing rules and i get a report back saying no action necessary because their AI system said there was nothing wrong. AI is not the golden ticket, it has flaws and prone to errors just like humans do.
Just for clarification of the issue, one of the listing rules is that sellers cannot use tags/words that manipulate the search feature. I did a search on the words 'computer parts' and in the search was many listings for watch bezels that fit onto mechanical watches, absolutely nothing to do with computer parts. The sellers listing page specifically points out the bezels are for specific types of mechanical watches BUT the seller used specific catch words in the listing that would show their listings if people searched for computer parts. I reported numerous posts and each time everyone came back saying their AI checker had determined that their was nothing wrong with the posts. It was all automated meaning a human being did not check the reported posts, it was left to AI and an automated response then sent to me.
The AI is wrong/ clearly wrong but Ebay does not care because they believe their AI checker works when clearly it doesn't but you cannot get in touch with a human being at ebay to tell them that and this is what is happening with companies and AI, they believe their AI works and thus trusts is more over that of a human being.
So no, AI moderation is and will be a very very bad thing.
Just for clarification of the issue, one of the listing rules is that sellers cannot use tags/words that manipulate the search feature. I did a search on the words 'computer parts' and in the search was many listings for watch bezels that fit onto mechanical watches, absolutely nothing to do with computer parts. The sellers listing page specifically points out the bezels are for specific types of mechanical watches BUT the seller used specific catch words in the listing that would show their listings if people searched for computer parts. I reported numerous posts and each time everyone came back saying their AI checker had determined that their was nothing wrong with the posts. It was all automated meaning a human being did not check the reported posts, it was left to AI and an automated response then sent to me.
The AI is wrong/ clearly wrong but Ebay does not care because they believe their AI checker works when clearly it doesn't but you cannot get in touch with a human being at ebay to tell them that and this is what is happening with companies and AI, they believe their AI works and thus trusts is more over that of a human being.
So no, AI moderation is and will be a very very bad thing.
Any way, from what I keep reading from experts and big wigs who support the use of AI, the main argument is that it can replace the paid work of human beings. From what is known about moderation in this forum, mod's are not paid therefore why introduce an AI system that will cost money to implement and maintain against a system that costs them no money?
The issue with moderation on this forum is not necessarily that it is performed by humans but rather that:
1) often a _single_ human decides (collective discussion is referenced but I do not see this reflected in moderator messaging).
2) there is no means to contest a moderator action : they decide, the decision is final, even if questionable or based on clearly incorrect assessment or an overly broad ‘rule’. There is a fine line between moderation and censorship.
The whole mechanism seems arbitrary and I have found moderators here to be very sensitive… jokes have been made about an AI being unable to distinguish between jokes and serious statements but the same could be said of some moderators
Sites like StackOverflow have a community moderation platform which makes assessment of any post a matter of multiple site users, thus making the entire process much more objective.
Even with a positive response score (which would denote positive contribution to this community) on my profile, I will likely get banned for this inflammatory content (which would prove the point).
1) often a _single_ human decides (collective discussion is referenced but I do not see this reflected in moderator messaging).
2) there is no means to contest a moderator action : they decide, the decision is final, even if questionable or based on clearly incorrect assessment or an overly broad ‘rule’. There is a fine line between moderation and censorship.
The whole mechanism seems arbitrary and I have found moderators here to be very sensitive… jokes have been made about an AI being unable to distinguish between jokes and serious statements but the same could be said of some moderators
Sites like StackOverflow have a community moderation platform which makes assessment of any post a matter of multiple site users, thus making the entire process much more objective.
Even with a positive response score (which would denote positive contribution to this community) on my profile, I will likely get banned for this inflammatory content (which would prove the point).
The issue with moderation on this forum is not necessarily that it is performed by humans but rather that:
1) often a _single_ human decides (collective discussion is referenced but I do not see this reflected in moderator messaging).
2) there is no means to contest a moderator action : they decide, the decision is final, even if questionable or based on clearly incorrect assessment or an overly broad ‘rule’. There is a fine line between moderation and censorship.
The whole mechanism seems arbitrary and I have found moderators here to be very sensitive… jokes have been made about an AI being unable to distinguish between jokes and serious statements but the same could be said of some moderators
Sites like StackOverflow have a community moderation platform which makes assessment of any post a matter of multiple site users, thus making the entire process much more objective.
Even with a positive response score (which would denote positive contribution to this community) on my profile, I will likely get banned for this inflammatory content (which would prove the point).
Figured all that out in the two weeks you have been a member huh?
"The lady doth protest too much, methinks" -Shakespeare