You know how when you say “a bad word” on the website, and an AI stops you from posting it and pauses your comments? It's a humble idea at first, until you realize this AI is TRASH!

Do you KNOW how many times my comments have been paused because I said something innocent? (I'd give an example but It will probably pause my Discussion Forums.) This is because the AI is using its actual intelligence to analyze the comment and is constantly mistaking the comment for something else. I'm not saying REMOVE this feature, but instead use a keyword-based algorithm that analyzes the comment for certain keywords. I know AI is the new, all the rave thing right now; but until it gets better at it's job, I suggest this alternative so people don't always get their comments paused.

I'm also kind of scared if there's a criminal record type thing on Scratch for every user. That could mean that people (including me) are increasing there record and their chance to get banned.

What are your thoughts?

1. You only get warnings/bans if you said sth actually bad or if you were intentionally triggering the filter (which is what the “AI” is called) many times.
2. Use contact us for this.

{what method did you use::control hat
answer on profile ::motion
} ::operators

lapizplayz1234 wrote:

I'm not saying REMOVE this feature, but instead use a keyword-based algorithm that analyzes the comment for certain keywords. I know AI is the new, all the rave thing right now; but until it gets better at it's job, I suggest this alternative so people don't always get their comments paused.

There isn't an AI, and the comment filter actually uses the exact method you suggested. You might want to read this post.
For questions and concerns about filter bugs, use contact us.

when tuned to [drop c v] :: hat events
repeat until <drenched in sweat? :: sensing>
blast some epic [system of a down v] riffs :: motion
end

lapizplayz1234 wrote:

I know AI is the new, all the rave thing right now; but until it gets better at it's job, I suggest this alternative so people don't always get their comments paused.

First of all, this “AI” isn't crazy gpt level stuff. It's just a chat filter.

lapizplayz1234 wrote:

I'm also kind of scared if there's a criminal record type thing on Scratch for every user. That could mean that people (including me) are increasing there record and their chance to get banned.?

There probably is a “Criminal record” on scratch. However, the scratch team most likely doesn't check every comment that gets flagged by the filter, as they don't have the time to do so. Even if they did, you wouldn't get banned for a comment that doesn't even contain anything offensive. Unless you're posting a ton of comments that get flagged, you won't get banned over comments.

lapizplayz1234 wrote:

use a keyword-based algorithm that analyzes the comment for certain keywords

I think the ST agrees with you though I don't know how the filter works, I imagine that's how they do it

lapizplayz1234 wrote:

What are your thoughts?

I personally think that such a strict filter is a good thing, because after all scratch isn't supposed to be social media (though some people use it like social media), it's for teaching people to code. An incident where someone in the community says something offensive could lead to schools and parents stop supporting scratch because of it's community.

There isn't an AI, and the comment filter actually uses the exact method you suggested. You might want to read this post.
For questions and concerns about filter bugs, use contact us.

I was sure there was because lots of unoffensive words got blocked by the thing, making me believe it's AI.