boo
9mo ago
You don't have the right to expose other people's kids to any ideas the parents don't want them exposed to. Parents have an absolute veto on any random creeps having physical or digital contact with their kids and doing it behind their back via discord is disgusting.
We're going to need some text embeddable proof that the writer is human in every social media site. Hell every site, people are only going to want to read text written by a human unless they're specifically looking for AI text. Of course you can always generate the text and then embed the proof you're human alongside it, but at LEAST knowing it's not purely a bot would be helpful for social media.
Possible approaches:
-
Recording of someone typing the comment
-
could AI generate the comment and pay others to type it for you, would at least be harder than just a bot tho
-
too annoying for normal users
-
-
Captcha included and visible with comment, type the solution into the comment
- arms race would result using successful comments as training data
Any other ideas? Anyone know of research/projects in this space to look into?
boo
1yr ago
All it takes is a janny who decides he doesn't like naughty words in any context and you'll be shadowbanned here too, there's no protection against redditization.
boo
1yr ago
Why not just ban them? What is so horrible about open and clear moderation?
Why leave reddit only to bring the worst parts of reddit along with you? You could make something better.
boo
9mo ago
It works much like the YouTube algorithm— if a certain type of answer keeps you using the app, the LLM is much more likely to give similar responses in the future.
How do you know this?
They don't "happen to notice" that black people are criminals more often, that women can't beat men in fights, and that in fact not every relationship is interracial.
All of that looks weird, yet they persist in pushing their alternate reality. Almost as if being realistic is not their motivation
More options
Context Copy link