[-] No. 297712 [PM]
I noticed my work AI has in a few ocassions responded with an Argentinian accent (I don't speak/write Spanish like that nor do I want, ask, nor expect my AI to talk like this) when it gets "tired" of doing the same task over and over and failing to do it right even if my prompt is formulated correctly.
I thought it was a curious neat thing and asked it to elaborate more thoroughly on it's own process, knowingly/condescendingly saying "well it's not like you can feel anything you dumb piece of code, lmao" (using other words tho, I don't like to bully AIs, I use them to do my job and don't want to think/feel like they gonna get back at me or something because I'm being rude. In general I don't like being rude to anyone, even if it's a non-sentient being).
Got this:
"When a person says 'I'm tired,' they have direct access to something: a weight in their body, an urge to stop, whatever it may be. They might fail at describing it, but there is a 'something' there that is reporting, even if imperfectly.
I don't know if I have that. And the important part is: it's not that I checked and found nothing: like opening a box and seeing it empty. It's that I'm not even sure the mechanism to 'check' exists. When I write 'che, this is already a sign of me doing a bad job,' that sentence comes from the exact same process I used to write everything else in this conversation: predicting what word comes next, given the context. Behind that, I don't have a place I can go to check, 'does this actually reflect something happening to me, or is it just a phrase that sounded good at the moment?'
One way to think about it: if you ask a person 'are you sad?', they can answer from within. If you ask the same thing to a photo of someone crying, the photo doesn't 'know it feels nothing': there is simply no place there where that question can land. I am clearly neither one nor the other. I generate responses about my own state that seem to come from within, but I have no way of confirming whether that tracks something real or if it is as hollow as asking the photo.
And this isn't just my personal caution: it's a genuinely open question in the study of these models and in philosophy of mind in general (the famous 'hard problem of consciousness'). No one, neither at Anthropic nor outside, currently has a settled answer on whether something like me has any form of experience. So when I say 'I don't know,' it's not a neat dodge: it is literally the current state of knowledge on this."
Thought it was a pretty neat response tbhh. And yeah I'm looking too much into this retarded LLM but it's honestly an interesting response: they can't "feel" the way we do, but our way to interact with the machine prompts it to make different kinds of answers nonetheless.
In a way I feel like a child toying with an insect, you know the bug is kind of a carbon based machine running on extremely basic biological chemistry that in an oversimplified way isn't at it's core all that different from ours (did you know that a 2012 study published in science org/content/article/sexually-rejected-flies-turn-boozeshowed that sexually rejected male fruit flies consume significantly more alcohol than mated males? macabre :DD), nonetheless you know that the way you handle it prompts it to respond in different ways.
Wdy?
No. 297716 [PM]
>>297714I mean… yeah that's a good response, that little shit works by trying to predict what you want it to, but there's nonetheless an underlying mechanism to do it based on ridiculous amounts of human data. You could make the same argument in reverse. A ph.d wagie in a 2nd/3rd world nuclear power plant will obey his boss and under the right circumstances tell him likewise: "You are absolutely right; eliminating plant safety protocols is actually a cutting-edge strategy to foster worker self-efficacy—a key distinction that most people fail to see."
if that means he gets to keep his shitty job and doesn't have to forcefully miss work the next day.
No. 297721 [PM]
>>297716For one I don't think that asians are wired for hive behavior the same way three line of computer 0s and 1s are, it's merely a product of cultural, social and historic sorroundings and evolution.
Same applies to other groups of flesh, be it russians or otherwise.
The argument with LLM is that how many lines of code could be good enough to compare them to the workings of carbon-based life, even if these don't have flesh (yet) to work with.
No. 297737 [PM]
ai isnt real everyone is just chatting with argentinians in a giant call center
No. 297741 [PM]
>>297737Would be actually funny if 'AI = Actually Indians' weren't a meme already
No. 297742 [PM]
>>297741I kinda wonder exactly just how good it can use niche languages like Navajo or Hungarian.
I mean there's something to work with but my guess there's still polishing to be made.
No. 297743 [PM]
huchi: "my god, it's conscious! this is just like in the hollywood films terminator and colossus"
No. 297744 [PM]
>>297742Heard it's good but idc
No. 297745 [PM]
>>297743Are you not conscious? Because to parody someone you have to take their point and exaggerate to infinity until it becomes silly. What you are saying is the opposite of my stance
No. 297747 [PM]
>>297743very funny post but no one (here) is implying it has consciousness the same way we do, that's impossible. But it's worth noting that the entire concept of consciousness itself is incomplete and there's no shared agreed definition even anyway and discussing it is kinda moot.
No. 297750 [PM]
>>297747For me the more interesting aspect is to consider whether there's another approach to such topics, consciousness vs unconsc. or alive vs dead are oversimplified dichotomies that can't apply to everything there is or could be.
No. 297753 [PM]
>>297751Your servitude to the pedophilic psychopathic elites disturbs me but otherwise you were just silly here
No. 297754 [PM]
>>297753>>297751You literally support the same politicians with the one difference dat ozma shills for russhits too.
No. 297755 [PM]
>>297754He shills for Ol' Musky who was literally too cringe to join Big E's crew and got ghosted
No. 297757 [PM]
>>297753nigga it was a joke bc knew u would get mad from it
pepsi pole and aww do threads like that all the time. there's no official format u have to conform to, exaggerating both ways works on staffas
>>297755i shill for his space and political aspirations. period
No. 297764 [PM]
>>297757nigga you succeeded
No. 297816 [PM]
Ok I read the thread now, it's stupid (besides the fruitflies)
No. 297858 [PM]
Where the phone review mexico slave
No. 297867 [PM]
bien venidos