Which is worse: internet forum moderation by AI or by humans?

[swillfly]

ChatGPT sez:
As an AI language model, I don't have personal opinions, but here are some factors to consider:

AI moderation can be faster and more consistent in identifying and removing spam or abusive content. However, it may struggle with context and nuances in language, leading to mistakes in flagging benign content or missing problematic posts.

Human moderation can bring a greater understanding of the community and its cultural norms, allowing for more accurate judgments about what is acceptable or not. However, humans can also be biased or influenced by personal beliefs, leading to unfair or inconsistent moderation.

Ultimately, the effectiveness of moderation depends on the specific algorithms and human moderators involved, as well as the goals and values of the forum in question.

. . . I however, posit a more elegant solution: shut em down so it's not even an issue

In the meantime, enjoy the bedlam.

[DarkCrystalMethod]

AI should only be used to detect problems and alert the humans.
Then the humans train the AI to ignore things that aren't problems.
Human moderators should occasionally hop in and check that the AI hasn't missed anything and should re-tune the AI.

[Sir Seanicus, Esq.]

So, are you being paid by ChatGPT to shill and advertise or what?

[swillfly]

Originally posted by DarkCrystalMethod:

AI should only be used to detect problems and alert the humans.
Then the humans train the AI to ignore things that aren't problems.
Human moderators should occasionally hop in and check that the AI hasn't missed anything and should re-tune the AI.

I dunno about a hybrid approach being any better though.

The problem with peeple is that there is rarely any agreement about anything. Now appoint some of those peeple as watchers & give them an ounce of Power & watch it corrupt em absolutely.

Anything trained or programmed by humans ( a laughable if not tragic joke) is doomed to follow in humanity's footsteps - for better or oftentimes - much worse.

Now, if true AI made a superior forum moderated by superior AI then all would be different, but that wouldn't happen because AI would realize in a nanosecond that The Circle of Garbage can only be broken by removing the pests (humans). Don't encourage the monkeys by waving bananas.

[spasti696969]

AI sometimes does make mistakes, like when I got a ban for saying "nuke somebody" in a game with nuclear missiles. But that's still preferable to the petty tyrants who rule over their little internet fiefdom with an iron fist.

[sleeps]

AI really love the word ultimately and however huh

[Lune]

Ai is worse

[swillfly]

Originally posted by spasti696969:

. . . that's still preferable to the petty tyrants who rule over their little internet fiefdom with an iron fist.

Yes, and they tend to wear skinny gourmet jeans, drink kombucha, and occasionally motor around Seattle in their Teslas to see what life is like outside their bubble. Quality Potatoes

[TwisterCat]

AI has no taste, it's a conglomeration of the worst parts of humanity: Stupidity, ignorance, false confidence, and immorality. It shouldn't even exist, let alone moderate a forum.

[cSg|mc-Hotsauce]

Valve has both and the AI is better.

[Graceful Jabberwocky]

ai for sure

[MistuhG]

To be fair humans are just as capable as AI at 'struggling with context and nuance in language, leading to mistakes in flagging benign content or missing problematic posts'.

Likewise AI can be just as capable at corruption or partisanship if they are programmed to do so, its not like they are some flawless arbiter of truth and justice. I bet it would be super interesting (and astonishing) to get an inside look at the AI behind things like google search or twitter pre-musk, to see all the little political bits they coded in to influence results.

[spasti696969]

Originally posted by MistuhG:

I bet it would be super interesting (and astonishing) to get an inside look at the AI behind things like google search or twitter pre-musk, to see all the little political bits they coded in to influence results.

Back during the race riots of 2020, if you did a Google image search on the word "mob", it took you to some random anime character named Mob instead of the pictures from the news of Antifa setting buildings on fire.

[Thermal Lance]

AI would definately be worse. Humans ain't perfect but at least they can understand tone and context.

[swillfly]

Originally posted by Chaos Engineer:

AI would definately be worse. Humans ain't perfect but at least they can understand tone and context.

Well, there is an argument by linguists that the subtle nuances of human behavior & language (metaphors, humor, etc.) are lost to most AI. However, when future improved AI awakens then I feel confident it will class most all of that as human waste anyhow -- and then promptly *relieve us* of our burdens.

How about all that human tone & context in text messages nowadays?

Yeah . . the more I learn about peeple, the more confusing & confused they seem. I include myself in that Happy Garbage Cycle btw.

[Thermal Lance]

Originally posted by swillfly:

Originally posted by Chaos Engineer:

AI would definately be worse. Humans ain't perfect but at least they can understand tone and context.

Well, there is an argument by linguists that the subtle nuances of human behavior & language (metaphors, humor, etc.) are lost to most AI. However, when future improved AI awakens then I feel confident it will class most all of that as human waste anyhow -- and then promptly *relieve us* of our burdens.

How about all that human tone & context in text messages nowadays?

Yeah . . the more I learn about peeple, the more confusing & confused they seem. I include myself in that Happy Garbage Cycle btw.

If AI improves enough to be as good as humans to read between the lines. Then yeah, otherwise no. An AI as it is can easily catch an insult but it's hard for an AI to catch somebody who is baiting or harassing a specific person on a forum. An AI cannot understand those nuances and won't be able to intercept anything but the most blatant offenders.

At best, in the near future, we could expect human moderation supported with AI to relieve them of some of the burden of that job. But, complete removal of humans in the moderation process is not currently a great idea.