Scratch should add moderator bots, they would moderate comments, project, etc, so the users dont have to, and the bots will instantly report it to the scratch team

They will be AIs and the scratch team will teach it what is inappropriate and what is not.

I don't think bots are smart enough to report inappropriate projects.

Maximouse wrote:

I don't think bots are smart enough to report inappropriate projects.

i forgot to say they are learning AIs

Why can't the humans do it?

MeowyTitan08 wrote:

Why can't the humans do it?

Well, there are millions of scratchers, and less than 50 moderators, so they are busy people

thr565ono wrote:

MeowyTitan08 wrote:

Why can't the humans do it?

Well, there are millions of scratchers, and less than 50 moderators, so they are busy people

Well there's the forum helpers and the community mods (archived forums).
Also the forums don't have a profanity detector. Why?

“just add moderator bots, it's very simple”

if moderator bots were good enough to not flag appropriate content and correctly flag inappropriate content then every website would be spam and inappropriate free. it's physically impossible with today's tech and especially for the scratch team to just add like that.

-InsanityPlays- wrote:

Also the forums don't have a profanity detector. Why?

They do have a filter that replaces bad words with asterisks, but it's much less efficient than the comment one.

Dude. AI can't tell what's inappropriate and what's not. Wait a few more centuries, and this may become possible

thr565ono wrote:

MeowyTitan08 wrote:

Why can't the humans do it?

Well, there are millions of scratchers, and less than 50 moderators, so they are busy people

The OP says that

SuperSean12 wrote:

the bots will instantly report it to the scratch team

meaning more work for the ST, unless they build an extremely smart AI. Lets say I put a bad word drawn by pen. How would a bot detect that? There's many ways to draw letters. I could draw it with dots. With lines. Stamping squares.

SuperSean12 wrote:

i forgot to say they are learning AIs

It's not that easy. There's so many ways that I could do something that could get me reported. Can an AI track down every single site that's not appropriate?

Machine learning is not yet good enough to filter inappropriate stuff. How would it know, for example, if a project is too scary?

Maximouse wrote:

Machine learning is not yet good enough to filter inappropriate stuff. How would it know, for example, if a project is too scary?

It would get existing data

Yes, this would be very hard to add. I bet though with some super high tech deep learning it could be done though but that would be very hard to implement.

People will find ways to get by the system. It would be hard to add.

We already rely on quite a bit of automated moderation processes. This has made a lot of people very angry and been widely regarded as a bad move (even though it actually does help us quite a bit).

I suppose the point is that no matter how much automation we do, it won't be perfect, and it's possible that some Scratchers may suffer because of it. A fine balance would have to be made to make sure that the moderation is as effective as possible without making too many mistakes.

We aren't roblox. I have a feeling that if this happened, you would get this alert for having gobo in your project (yes, very ridiculous).

Col_Cat228 wrote:

Dude. AI can't tell what's inappropriate and what's not. Wait a few more centuries, and this may become possible

Well, they probably can. I'm no legit coder, but the bot could scan a comment for any words that are bad words misspelled.

Automating part of the moderation process is important in trying to keep up with the massive amount of content input into this site, though I wouldn't want to go as far as to automatically take down projects algorithmically. Even if that system were to be introduced, it'd still need a lot of human intervention and people are going to get mad if they realise a machine is determining whether or not their project should be public

Col_Cat228 wrote:

Dude. AI can't tell what's inappropriate and what's not. Wait a few more centuries, and this may become possible

Maximouse wrote:

Machine learning is not yet good enough to filter inappropriate stuff. How would it know, for example, if a project is too scary?

okay looks like we're waiting for GPT-4 to come out; pack up your bags everyone lol

CatsUnited wrote:

Automating part of the moderation process is important in trying to keep up with the massive amount of content input into this site, though I wouldn't want to go as far as to automatically take down projects algorithmically. Even if that system were to be introduced, it'd still need a lot of human intervention and people are going to get mad if they realise a machine is determining whether or not their project should be public

Col_Cat228 wrote:

Dude. AI can't tell what's inappropriate and what's not. Wait a few more centuries, and this may become possible

Maximouse wrote:

Machine learning is not yet good enough to filter inappropriate stuff. How would it know, for example, if a project is too scary?

okay looks like we're waiting for GPT-4 to come out; pack up your bags everyone lol

Fwiw GPT-3 is sometimes generating some quite racist output so they're blocking some output of GPT-3 if it contains certain things.

bump