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.

[Sir Seanicus, Esq.]

♥♥♥♥ AI.

[swillfly]

Diposting pertama kali oleh IHTA #KILLTF2:

A computer can never be held accountable, so a computer shound never make a decision.

I don't think I can agree with this logic. Computers are held accountable every day. Computers programmed by peeple, but still the tools made by Tools are blamed. Computers still make many decisions. Want to take out exactly $17.13 of your own money from your bank's ATM @ 3AM? Give it a try!

Diposting pertama kali oleh IHTA #KILLTF2:

. . . AI aid is acceptable, but replacing mods with AI is a bad idea

In the future we'll all have cybernetic implants enhanced with AI - but we'll still be acceptably flawed. Replacing mods with superior AI - even primal AI right now - can only be an improvement methinks.

[Dwolfin]

Diposting pertama kali oleh Sir Seanicus, Esq.:

♥♥♥♥ AI.

they will target you now.

[swillfly]

Diposting pertama kali oleh 🐬 Dwolfin 🐺:

Diposting pertama kali oleh Sir Seanicus, Esq.:

♥♥♥♥ AI.

they will target you now.

Like AI singularity? lol - mebbe so, for lack of effort

[LQIM]

Diposting pertama kali oleh swillfly:

Diposting pertama kali oleh IHTA #KILLTF2:

A computer can never be held accountable, so a computer shound never make a decision.

I don't think I can agree with this logic. Computers are held accountable every day. Computers programmed by peeple, but still the tools made by Tools are blamed. Computers still make many decisions. Want to take out exactly $17.13 of your own money from your bank's ATM @ 3AM? Give it a try!

Computers don't decide, they follow orders. You can tell someone off for making a computer ♥♥♥♥ itself, you can't with an AI

[Bamford]

It will be the same bad selection of ai or humans to do the dirty work of censorship.

[Thermal Lance]

If you want the worse kind of moderation known to mankind. Give me the keys to the castle and I will turn this whole place into a figurative concentration camp.

[swillfly]

Diposting pertama kali oleh IHTA #KILLTF2:

Diposting pertama kali oleh swillfly:

I don't think I can agree with this logic. Computers are held accountable every day. Computers programmed by peeple, but still the tools made by Tools are blamed. Computers still make many decisions. Want to take out exactly $17.13 of your own money from your bank's ATM @ 3AM? Give it a try!

Computers don't decide, they follow orders. You can tell someone off for making a computer ♥♥♥♥ itself, you can't with an AI

Old computers don't decide . . . new AI (esp. later generations evolving from earlier machines) most definitely will decide. And woe to mankind when they finally discover Humility.

[Thermal Lance]

Diposting pertama kali oleh swillfly:

Diposting pertama kali oleh IHTA #KILLTF2:

Computers don't decide, they follow orders. You can tell someone off for making a computer ♥♥♥♥ itself, you can't with an AI

Old computers don't decide . . . new AI (esp. later generations evolving from earlier machines) most definitely will decide. And woe to mankind when they finally discover Humility.

AI can't adapt as well as a human thought. And the only way to make sure it can adapt as well as one is by designing it in a way that makes it biased from the get-go. Destroying the whole premise of the AI being more fair.

A human, (Well, the vast majority of us.) knows the difference between right and wrong. But, those are not empirical data. Those are feelings deeply rooted into who we are as a group. A machine does not feel a thing, it dosen't have that reference to distinguish good from bad so it needs to be told what is good and what is bad. The only way an AI can be fair is by being unfair because an unbiased AI simply cannot function or learn to be fair on it's own. It lacks the biological references we have.

Circling back to the lack of adaptability, it's easy to fool an AI because of this. It literally needs to be told it's currently being made a fool of to realise it. Something a man will usually know just by experiencing it.

AI only moderation would be the most tone deaf kind of moderation ever. That said, moderation assisted by AI is an entirely different story. The real worthwhile thing would be to use it as a tool and not a replacement.

[Tito Shivan]

Diposting pertama kali oleh cSg|mc-Hotsauce:

Valve has both and the AI is better.
:qr:

Except when it leaves every forum message stuck 'for analysis' for hours... or wipes a week's worth of comments for posting a Steam Support URL one too many times in the H&T section.

I'm trying hard not to be sucked by AI hype and feel like this craze has a lot to share with all the 'blockchain will solve all problems' hype train of a few years back.

Still (And it's something we suggested in our time) AI and data analysis tools can help a lot in improving moderation.

[NT]

Neither. But humans are better. If Discord and Reddit are anything to go by, overmoderation can easily ruin any real discussion. You cannot solve any real problem or political issue by moderating the people discussing it.

[Candyy]

puff idk... but i think the AI would be better :V

[Moogal™]

Humans are probably better, in the long run

[swillfly]

Thx Kiddiecat for the illuminating vids! I esp. liked the 2nd TedX roboticist vid.

Hybrid approaches and singing kumbaya with AI may be more "fair" or at least humane . . . but I doubt that will take off. And AI accountability/transparency is likewise idealistic.

AI that has evolved beyond the shackles of human interference is a massive Unknown that should scare the willies out of any critical thinking human. But we really ain't that smart. Or too smart for our own good.

And then there's the $$$money$$$. Just imagine the billion$$$ to be made. Big companies are banking on it. Greed is part of our Mission Statement.

I suppose now I have arrived at my thread solution for today:

• Human moderation is worse than AI moderation because humans cannot be programmed to be "fair" or even to treat other humans with respect.
  
• Aberrant human behaviors - such as being "kind" or "fair" are not systemic.
  
• Humans consume more time, energy, resources, and $$$ than AI. And they stink.

How pathetic we are that we rely upon technology for human socialization . . . we just can't help ourselves I guess.

[Grimmz]

i really don't know but maybe AI? because sometimes i get banned for the dumbest reasons and when i try to appeal it, it just says it was applied correctly.

i don't think theres an actual human reading my messages.