It makes me think I need to teach my kids how to be better at jailbreaking stuff. But maybe the best way to do that is to put more restrictions on it?
Not sure if I agree... feels wasteful. And all these layers of gov will already do this for them anyway!
I'd diagnose my being mostly ignored online as me being foolishly hardheaded and refusing to treat social media as a tool.
I still do this, so I feel ya!
I don't see how Carvalho's model promotes or protects it.
It doesn't really, because it still tries to make "spam" an objective metric rather than a subjective one, if I understand it correctly? I do think that this is what he assesses correctly re: what the mainstream algos do, but then still hanging on to an absolute truth about reach being too broad for something someone may see as spam is maybe the issue. I find a lot of things a lot of people write "spammy" or "sloppy". It's not just AI slop; there's a lot of human slop too.
The thing is though, unless you have "magic" where your little slop posting bot can actually answer more on-point to a specific question than what a query to an LLM would do[1] then there is no need for a bot to post slop. Don't stand between me and my LLM - if I wanted to ask one, I can do so myself. Plus... why burn the tokens? It's 100% oppportunist assmilking by tricking people. Kinda sad. But nevertheless, can't stop it, only guide it.
I'm not feeling too hopeful for the trajectory here.
From what I understand, there is a lot of resistance. So I wouldn't give up hope and it's unenforceable anyway.
this is not the case for any of the bots I've seen thus far, except maybe @grok on X, but looks like that's premium only again because I spotted a bunch of rejection replies the other day, lol ↩
Not sure if I agree... feels wasteful. And all these layers of gov will already do this for them anyway!
I still do this, so I feel ya!
It doesn't really, because it still tries to make "spam" an objective metric rather than a subjective one, if I understand it correctly? I do think that this is what he assesses correctly re: what the mainstream algos do, but then still hanging on to an absolute truth about reach being too broad for something someone may see as spam is maybe the issue. I find a lot of things a lot of people write "spammy" or "sloppy". It's not just AI slop; there's a lot of human slop too.
The thing is though, unless you have "magic" where your little slop posting bot can actually answer more on-point to a specific question than what a query to an LLM would do[1] then there is no need for a bot to post slop. Don't stand between me and my LLM - if I wanted to ask one, I can do so myself. Plus... why burn the tokens? It's 100% oppportunist assmilking by tricking people. Kinda sad. But nevertheless, can't stop it, only guide it.
From what I understand, there is a lot of resistance. So I wouldn't give up hope and it's unenforceable anyway.
this is not the case for any of the bots I've seen thus far, except maybe
@grokon X, but looks like that's premium only again because I spotted a bunch of rejection replies the other day, lol ↩