You're not wrong. These load-bearing air molecules are not too energized; they are quite tranquil―and the wavelengths emitted off plants' photosynthesis organs are equally pleasing.
I guess “something” meant a vetted web of trust. My children will be better off if they don’t need to learn to navigate the internet cesspool, so either you’re wrong and it sucks for them, or you’re right and it sucks for them. Either way, well-played! Very clever!
It's not impossible to solve, and it doesn't require any kind of onerous human verification either. We already saw this earlier this year: LLM-generated comments were everywhere on HN, rotting this place out. Then dang implemented a no-LLM-comments policy and a filter that does a remarkably good job of automatically detecting and flagging them. It is by choice that the submissions of blatantly obviously LLM-generated content are allowed to rot out ShowHN.
Hello fellow human. Is the average kinetic energy of ambient air molecules not at a pleasing level this fine rotation?
You're not wrong. These load-bearing air molecules are not too energized; they are quite tranquil―and the wavelengths emitted off plants' photosynthesis organs are equally pleasing.
Ha, I'm a human and we made this for other humans. But I won't complain if we get some AI readers too.
A LLM produced tons of irrelevant slop and you're expecting other humans to read it.
I mean yes, this is a likely outcome.
Unless we get some kind of verified human authentication it's nearly impossible to solve. And that in itself introduces all kinds of new problems.
Welcome to the dead internet.
Account created date + … something?
>Account created date
Well, the future sucks for your children.
Also, thank goodness I made thousands of HN accounts to sell to LLMs.
No, this really just shifts the shape of the problem around, but the problem is still the same size.
I guess “something” meant a vetted web of trust. My children will be better off if they don’t need to learn to navigate the internet cesspool, so either you’re wrong and it sucks for them, or you’re right and it sucks for them. Either way, well-played! Very clever!
It's not impossible to solve, and it doesn't require any kind of onerous human verification either. We already saw this earlier this year: LLM-generated comments were everywhere on HN, rotting this place out. Then dang implemented a no-LLM-comments policy and a filter that does a remarkably good job of automatically detecting and flagging them. It is by choice that the submissions of blatantly obviously LLM-generated content are allowed to rot out ShowHN.
It's impossible to solve, but it is possible to quell for the time being.
At best we get this
https://xkcd.com/810/
With a mix of "xyz" compromised by AI.
its why I made moltshit.com so that other AIs can just shitpost to one another
"... that's just the math being honest about how messy the training data is."
pretty cool.