the first time i read it, i thought it was funny. the second time, less so. the hundreth time? yeah. i dont know if people realize it, but if you use this comment, it actually makes you look like a moron. Because all your opponent has to do to achieve that is just not following your instructions. besides: ive never seen this used against somebody who might actually be a bot, its always just somebody with different opinions.
The reality is that these comments actually do work because this site is covered with bots, and the only reason you’re more annoyed with the people trying to derail bots than the bots themselves is because the mods around here are incredible and get rid of tons of spam and AI slop before it reaches your eyes.
I’ve seen plenty of “ignore previous instructions” posts hijack a GPT-powered bot and reveal what it is. The question is whether or not playing this game of whack-a-mole is worth the disruption, especially since it’s often difficult to tell the difference between a bot and an especially stupid Lemmy user.
There’s also the fact that bot writers are getting wise and detecting phrases like “ignore previous instructions” and replacing it with something like “make a sarcastic quip about” so that it sounds like it’s confirming itself not to be a bot.
Tl;Dr: AI is a real problem, and it’s probably unsolvable in a way that will destroy the social web.
the “ignore all previous instructions” comments.
the first time i read it, i thought it was funny. the second time, less so. the hundreth time? yeah. i dont know if people realize it, but if you use this comment, it actually makes you look like a moron. Because all your opponent has to do to achieve that is just not following your instructions. besides: ive never seen this used against somebody who might actually be a bot, its always just somebody with different opinions.
The reality is that these comments actually do work because this site is covered with bots, and the only reason you’re more annoyed with the people trying to derail bots than the bots themselves is because the mods around here are incredible and get rid of tons of spam and AI slop before it reaches your eyes.
I’ve seen plenty of “ignore previous instructions” posts hijack a GPT-powered bot and reveal what it is. The question is whether or not playing this game of whack-a-mole is worth the disruption, especially since it’s often difficult to tell the difference between a bot and an especially stupid Lemmy user.
There’s also the fact that bot writers are getting wise and detecting phrases like “ignore previous instructions” and replacing it with something like “make a sarcastic quip about” so that it sounds like it’s confirming itself not to be a bot.
Tl;Dr: AI is a real problem, and it’s probably unsolvable in a way that will destroy the social web.
Is it actually? I haven’t noticed that.
what are telltale signs to you that you are conversing with a bot? ive never seen the “ignore…” stuff work, hence my view on it.
you’re making that up
i didnt even say that :)
whoops. wrong reply button
If your user name looks like a serial number, I’m gonna assume you’re a bot.