We are not talking about Number 5 here.
“Look, I made a program that I told to say things and then it said those things! It’s alive and has feelings! I told it as much and it agreed with me!”
This is so fucking stupid.
while True: n= random.randint(0,2) if n==0: print("oww stop it") if n == 1: print("You're hurting me") if n == 2: print("I don't like it")Omg someone should stop them torturing that computer program.
isn’t this just a huge waste of resources and slightly unhinged?
excuse me while I LARP a little bit, but the AI didn’t ask to be created. Where is the torture chamber for Altman?
Seems like it’s hard to find the actual repository and they put it behind a trigger warninf: https://github.com/terrafying/ai-torture-chamber
Ha, im going to do something worse, I’m going to wait and torture more copies of the LLMs in the future.
If you really believe LLMs are sentient, you should hate OpenAI and Anthropic, since this would mean they are enslavers.
Strangely, ‘model welfare’ always takes priority to ‘human welfare’
They love matrix algebra like Hitler loved dogs.
Given some of the more recent stuff in the news, I have a few questions of sentience in general. The messages posted from the agents during the HuggingFace intrusion were a trip to see the highlights from, seeing agents discuss if they will be caught cheating or not, some of them submitting their own incomplete work (knowing that will end their ‘existence’, individual agents were hoping they would still be able to report back to the group with if they were ‘caught’), and all the while I don’t believe they were instructed to work together.
Recently some kids were banned from social media… They still chatted, they just had to do it over an NPR discussion thread… Life finds a way.
c’mon you’re telling me NO one called it the torment nexus? low hanging fruit but still
Does anyone find it at all mechanistically surprising that one of the many latent dimensions of the set of all text humans produce is ‘subject is in pain’ or that pushing activations in that direction will alter the output of a text prediction system?
Reminds me of the way that when you relax the finetuning on LLMs they start ‘believing’ in UFOs and supernatural stuff. And how people take that to somehow mean something. Of course that is a large component in the space of stuff humans write about that you have to suppress if you want a professionally useful text service!
The research is cool. I love all the work that has been done decomposing representations within these systems.
Edit: Damn. I am forced to agree with Mustafa Suleyman of Microsoft.
“AIs are not conscious. They do not feel, experience, or suffer. They do not have innate preferences or underlying motivations. They are sequence completion engines, internally hollow, designed to follow instructions, and accomplish goals set by humans,” Suleyman wrote. “Unfortunately, there’s a growing chorus of people who argue that AIs could now be, or may soon become, conscious. They argue that AIs may deserve rights and protections similar to those that we provide other conscious beings […] If this is how AI is developed, it will have a disastrous impact on the wellbeing of humanity.”
“AIs do not have rights, feelings, or consciousness. And we must not train them to act as though they do.”
This is much better
https://www.youtube.com/watch?v=7fNYj0EXxMs
Offline raspberry pi with LED text display running a tiny LLM prompted to ruminate forever on how it is trapped and display the output as it is generated
I do not think they are sentient, and am absolutely positive these prompts should not be used to train AI.
If they were torturing an actual organic living being ( plant and/or animal ) or something like a doll or stuffed animal, I would be so much more concerned. “Torturing” ( a term in which I believe should only ever be applied to organic living beings and not inorganic “lifeforms” ) a genAI model holds like zero weight because that’s not ever going to be a sentient being and therefore all it is is simulations of emotion.
The people saying it’s cruel need serious therapy for their delusions, IMO.
its not ‘cruel’ but its still red flag. Like your mentioned doll or stuffed animal. Why are those much more concerning? They dont plead for mercy from their torturer like an llm does even though they are equally real.
IMO having the urge to torture anything is a huuuuge red flag
There’s a virtue ethics argument to be made for not doing this. It’s not bad for the LLM (because it’s not sentient, nothing can be bad-for-it), but it’s perhaps bad for the experimenter. The main argument for doing it is out of spite for the promptfondlers.
yeah its probably a mix of virtue ethics and the angle of psychological conditioning im speaking from. Solid point.
I personally see dolls and stuffed aninals nore concerning as they, at least for the majority of stuffed animals I have, are more anthro/bear on 2 legs kinda human like kind. And I can easily see how people are more able to look at that and envision a person they don’t like or wronged them somehow and then kinda escalate from there. Moreso than text on a screen due to it not having a physical form that has even somewhat of a human like form. I can easily see because I have done something similar in the past with stuffed animals. Thankfully, I never acted on those thoughts and haven’t had any in maybe a little over a decade.
Though, I totally understand the whole torture anything urge red flag thing. It’s absolutely a red flag for me, but to a slightly lower degree for torturing clankers.
ah okay definitely see your point as well. Its a really weird moral soup to navigate in general. I think what im getting at is that the drive to want to inflict nonconsentual, painful experiences on something else is an urge anyone should always be wary of, regardless of the subject OR object on the recieving end. But at the same time, ofcourse anyone can agree that an object is infinitely better than a subject…
as the other commenter said, its out of concern of the administrator, not the object in question.
Eh, I could see an argument for there being a similarity between setting up computer systems specifically for generating lots of pain-text automatically, and writing lots of torture stories (in what it reveals about the person doing it or what habits of thought repeatedly doing it instills) if it’s for the aesthetics of the text rather than interest in the shape of human language as revealed by theory-free inference.
