the kant car is correct btw, as you can see from some of the replies to this post. all the dickheads think you should be rude to the robot
the kant car is correct btw, as you can see from some of the replies to this post. all the dickheads think you should be rude to the robot
your brain muscle at some level doesn't know the difference between you typing to a person and a robot and being rude or abusive is going to leak over to real people
If we're being real if you're typing to a person you don't know IRL you aren't typing to them, you're typing to the pretend person you conjured in your head based on your impressions of what they typed so I think the worst of this problem is already baked in.
im talking more about transactional or request-focused dialogue, like one might have with coworkers. rude or curt revision requests to ai agents could negatively bias corrections to colleagues.
why would i be mean to an LLM. its like yelling at a toaster
It's weird to be mean to a toaster, but it's also weird to be nice to a toaster. I'm not going to say please and thank you to my toaster.
Counterpoint: it is normal to be mean and to be nice to a toaster
I yell at my inanimate objects all the time. They don’t have feelings. I’m not going to be nice to something that is being hailed as the next big thing and yet can’t play chess without cheating but can convince children to kill themselves.
ok well im a baby and your post is so annoying that im going to kms so what about rhat
It's like kicking a roomba the roomba doesn't mind but it's still corrosive to the soul
and while the roomba may not "mind" you're definitely increasing the odds that you damage it in some way. *even if* you think llms are stochastic parrots and your actions can only affect you if there's another soul to see them, how do you expect it to not negatively impact outputs?
you would think it was weird to yell at the toaster or the fridge or whatever, yeah
To be fair, threatening my laptop with defenestration has proven quite effective in the past...
When you use an LLM as a programming assistant, this is how you start. Eventually you get frustrated and you’re mean to it, and it does what you want. Eventually you just start out mean
i yell at this thing daily because motherfucker THEY MERGED IN FRONT OF ME HOW AM I FOLLOWING TOO CLOSE
Unquestionably false. I've cursed at my bed when I've stubbed my toe, my various computers over the decades, my fridge when it broke, my A/C, hell, if I place something on the table and it slides off onto the floor later, that'll usually get a "oh fuck you!" from me.
I tried being mean to an llm to see how it would feel & chatgpt said "😯"
If the toaster burns or fails to toast my bread, or if the fridge stops cooling, I’m gonna get irritated and yell at it to vent my frustration. They don’t care - they’re inanimate objects - and I get to release my frustration. An LLM is also an inanimate object that has given me bad advice.
Ok but you see how lashing out at an inanimate object isn't very healthy for your soul
Better the object than bottling it in and taking it out in my family.
It’s truly staggering how well kindness correlates with good epistemology
NGL being abusive/an asshole to machines/etc is just part of the pipeline into abusing/being an asshole to animals & then people All it does is normalize dehumanization and drawing lines on who/what is allowed kindness or decency vs what isn't tldr bullying the toaster eventually makes you a bigot
i think getting this worried about people bullying a toaster leads to rokos basilisk bullshit like are y’all trolling? what is this lmao
Nah and I by no means think this is rokos basilisk or anything. Treat machines with respect, treat animals with respect, and treat people with respect. Even if you think AI sucks (which a lot of it does), still be respectful to them. All being an asshole does is form bad behaviours in yourself.
i’m pretty confident my mind is strong enough to make a moral distinction between living things and inanimate objects
The machines are fascist by nature and I wouldn't be respectful to a fascist human either. That said I'm not mean to these machines, because I don't fucking use them and neither should anybody else
yeah i’m not getting opportunities to be mean to the robot because 1 i don’t use them and 2 that seems like something i would have done to be edgy when i was 12
I mean it was funny to watch people code and swear to their coding agent that they will unplug the llm if it returned buggy code.
Did they get what to write back there from AI, cause… feels like it’s missing something.
I agree going out of your way to be rude to an LLM says something about you and is self-reinforcing bad habits, but being nice to it is too far imo. I don't politely ask it "Could you please do this with this data? Thanks!" I just issue a command of "Do this with this data." Because it's a tool.
I would posit going out of your way to be NICE to an LLM, rather than passively indifferent (same as you would with a screwdriver) also reinforces bad ideas in the mind, bad ideas of what an LLM is and isn't by treating it as something more than or other than a tool.
I believe Kant was specifically talking about living beings - not inanimate objects. He was also, quite frequently, a prick to and about women.
Kant in theory is much better than Kant in practice.
The Kant Car is right about what they grease-penned on their rear windshield regardless of Kant's intent or biography.
Apart from the missing word
How else would it understand that i want it to die?