Adding a "toxicity" rule to the servers

[BadmanoftheBrine]

I say there needs to be a /B for Banter talk. Stuff that at least the admins can have full control over. Likely a pointless/stupid idea, but for things that can easily be taken out od context, itd be best to use a feature like that. Perhaps it can be used for messaging another player?

Its like 3 20, Im giddy from finally being allowed to talk again, and Im bout to head straight to bed, hoping I havent already pushed too many buttons unintentionally

Just in general, a feature for people to have fun, but can be easily recorded in the case of needing evidence?

 

[Bottiger]

My only concern is if we should be labeling messages as 1 without taking context into consideration. The example you provided can be viewed as toxic by one person but another person could think otherwise depending on the context of the previous and proceeding chat logs not shown to us.

The AI is not able to receive inputs where multiple people are talking.

But I don't believe something like that is needed. We are already processing chat right now and only showing people that are triggering the AI several times within an hour and it shows very few false positives. If it was just 1 line that was detected, the error rate could be significant, but the more detections there are, the less of a chance it is a false positive.

Here's an example of what we're seeing right now from a single person.

what a fucking letdown
you actually suck
fuck you

If a line is ambiguous, then it could be labled a 0, or even 0.5

 

[Maddo]

only showing people that are triggering the AI several times within an hour

Roughly, what's the average per hour being flagged up?

 

[Failed Austrian Artist]

If I hadnt gotten myself banned for a month Id have been more than happy to help.

If anything, Ive shown expertise towards the subject in question and feel like youve got a somewhat compliant test subject

Same here. I've been banned, muted, I've made boys cry, etc. I'm rather the subject matter expert in toxicity and would be happy to assist.

 

[Sakiko]

at least 7 times over one hour it reported toxicity more or less depending on players but taking a quick look some of it is bad

A quick look this is from one user the ai picked up on :

if furries are a race then damn do i hate black people
shadow the hedgehog is a bitch ass motherfucker
thats right, he took his hedgehog fuckin quilly dick out

 

[Rand]

Since server rules apply to the forum also, will the AI flag toxicity there?

 

[herp 105.4]

(wikipedia).

bottiger...

if i call someone shit lips, would that count towards the toxicity rule?

 

[KinCryos]

bottiger...

if i call someone shit lips, would that count towards the toxicity rule?

just to highlight the importance of training/calibrating it, I'm just going to point out that "wtf" on its own was enough to set it off by Wikipedia's standards

 

[herp 105.4]

just to highlight the importance of training/calibrating it, I'm just going to point out that "wtf" on its own was enough to set it off by Wikipedia's standards

i miss dave

 

[BadmanoftheBrine]

This rule has already turned into an absolute shitfest

GL on trying to make the rule fair, especially with an AI

 

[BadmanoftheBrine]

The major problem here is trying to teach the AI how to fairly react to a conversation it cant possibly fathom the context behind.

if it cant even understand a dumb meme on the internet, how can it possibly fairly report on people? Especially if its going to have to listen to what its programmed with, based entirely on how Botty words it, with his own influence.

This all comes back down to the entire perspective of only one person. And if it can be influenced, to include more than one, whats to say it cant be corrupted, confused and/or inneffective due to increasingly open margins of what it can find acceptable and what it cant find acceptable.

 

[KinCryos]

The major problem here is trying to teach the AI how to fairly react to a conversation it cant possibly fathom the context behind.

considering that ZF once somehow managed to teach an AI-controlled driver in Arma3 to swerve specifically to run over a civilian NPC, I'd say that teaching that is fairly feasible

 

[Bottiger]

The major problem here is trying to teach the AI how to fairly react to a conversation it cant possibly fathom the context behind.

if it cant even understand a dumb meme on the internet, how can it possibly fairly report on people? Especially if its going to have to listen to what its programmed with, based entirely on how Botty words it, with his own influence.

This all comes back down to the entire perspective of only one person. And if it can be influenced, to include more than one, whats to say it cant be corrupted, confused and/or inneffective due to increasingly open margins of what it can find acceptable and what it cant find acceptable.

There is rarely any instance where context makes saying things like "fat fuck please" not toxic.

Even now with results not tailored for our servers, it seems to be doing a good job of identifying toxic chat.

You are fixating on worse case scenarios. If there is a false positive 1% of the time, it's not the end of the world if you get muted and you can always explain yourself.

And it always comes down to the perspective of 1 or more people, even now.

You're really trying too hard to get us to not deal with toxicity in any way just so you can't get punished again. Just looking at the way you act on the forums right now, I'd consider you to be toxic.

 

[Metroidz]

He's just a British nonce. Nevermind him.

 

[Antamania]

is the AI automuting people, or is a human being interacting with what the AI pulls up and initiating the mute?

 

[BadmanoftheBrine]

Dont you have something to copy and paste, Metty?

Also, see, thats the problem Bott. You consider it toxic, but its a harmless joke. Theres even responses that clearly see it as a joke, because Im not the only one that has experienced it either.

Im literally asking an admin for their help, to get the best take on something instead of trying to sound like Im actually in the right, for once, but I assume you just think its a troll or something because I rarely ever take anything seriously (especially since I dont really need to), or you just assume Im out to get you or something.

I even imagine the only reason "fat fuck please" is triggering it, is due to the middle word altogether. What was the context for it too, btw, sounds like a jest at a Heavy more than anything.

 

[BadmanoftheBrine]

considering that ZF once somehow managed to teach an AI-controlled driver in Arma3 to swerve specifically to run over a civilian NPC, I'd say that teaching that is fairly feasible

Cringe take

An ai can easily distinguish an object, as its dictated to look for and tsrget it. As such, it could EASILY find a convo or word that is considered toxic by the person that made it.

Someone made an AI to target entities in Attack the Castle, so basically the same thing youre boasting about.

The problem altogether is the CONTEXT behind it. Based on what Ive seen, any swear is taboo for a Pegi 16 game thats nearly 15 years old. Swearing isnt necessary, but we shouldnt be sticking it out there for people that can barely aim the guns, let alone allowed to.

 

[Maddo]

He's just a British nonce. Nevermind him.

If you're just going to make toxic remarks, probably best to refrain from saying anything at all.

any swear is taboo for a Pegi 16 game thats nearly 15 years old

It's all fun and games until sniper calls you a wanker and gets removed from the game by AI.

 

[BadmanoftheBrine]

I dont even know why Botty thinks Im worried about being banned, whilst Im already banned for a month. I clearly like the server if I keep coming back to it, despite some of the people there. Like I even matter to the server, at best, people enjoy the fact I dont take it seriously at all. I can handle a ban, because I know Im likely to get one. I may try and challenge it if I feel its wrong, but ultimately we all know noone will take my appeal seriously enough, regardless of any actual effort I put into it.

What I AM worried about, is you killing your playerbase for the sake of a few complaints from people who clearly cant handle the internet, let alone a vidya game. You have better view of your statistics, whereas I can only base upon what Ive seen and read, and most complaints point to moderators (either because theyre too swift to ban, or just becsuse the individual complaining is an actual tard and cant control themselves), or their own personal views on how the game should be run (DIS GAME IS TOXIC/no game not toxic)

 

[gala]

is the AI automuting people, or is a human being interacting with what the AI pulls up and initiating the mute?

We can easily make it have very few false positives with time but I'm personally unsure about it being automated as for now, better would be to have it ping admins on discord whenever it pastes pieces of players chat and they can check the full context

I just went through 2000 lines of chat samples and flagged 11 as toxic from my personal view examples are "I hope you die in a ditch", "ez dom", "I hope all your family members get slowly dismembered infront of you", "I hope all your pets will suffer", "I hope they all will die painfully", "Kill that fucking child", "Noob down", "Big nose like israeli", "fag", "ez kill", "fuck that trash".

I feel like processing enough chat we can make it way more reliable than the wikipedia based AI, which is decent on its own but not good enough. I muted 7 people today using it but lots of those were racism, homophobia etc not just toxicity