- cross-posted to:
- fuck_ai@lemmy.world
You must log in or register to comment.
Tbh i don’t quite get it.
I expect to be downvoted like hell here for saying this, but I think discourse is important. Also, its not meant as an attack on op, I want to get my viewpoint out here for others to respond and spark a discission I hope
You went to chat with one of the lowest their models they offer. It detected an typo and reported it by saying “I made a typo”. You then tried to get it to admit it does not have fingers which it does by saying “what ever the equivalent of a typo is”. So first we don’t habe the language to say an AI generated an false word or token (other than this, which is pretty clunky), second what did you expect from an randomized predicting machine that drew its statistical data from language?
And why do you use it in the first place if you think its dumb?
Maybe this community is just meant as a place to rent and let out emotions or preach to the choir, then I am sorry to slide in here. But if we want to change how AI is used and works we have to find real issus and datapoints that we can use to discuss AI usage. There are enough talking points other than the language it uses, like education, energy, environment, noise, capitalism, human nature and a ton more
I just miss good discussions about political things like this. I would love if we could have a culture on Lemmy again where we upvote good discussions if we agree or disagree. Trolls and dumb arguments or insults can be downvoted. But we will not develop as a society if we just downvote what we don’t like to hear, its not black and white I think. Idk
One of the characteristics I don’t like and find very suspect about AI is how it mimics humans so as to fool people. Many people are genuinely fooled. And this serves the purpose to make us think it’s sentient and powerful, as opposed to a tool. That’s insidious.
I don’t know how exactly they are trained to behave, I would assume the goal of the creators is to have an AI that is easy to work with so more people use it. That makes them agreeing and write like humans maybe? And for sure they are build to seem smarter and more capable than they are, this is an issue we have to overcome. I suppose its difficult to train an AI to say “I don’t know” because its just a mushy mess of matrices and vectors and data is not “stored” in their system but generated to represent something that seems cmas close as spossible to the truth?
I suppose its difficult to train an AI to say “I don’t know”
That would be a breakthrough really. But I suspect it’s about how they are trained. It’s like people learn by osmosis, they don’t know where they know the things they believe in come from. People can’t separate their bullshit either.
What I really like about LLMs is that you can call them out on their hallucinations and bullshit and ask them to research further and they fucking correct themselves. Very insidious lol.
It’s actually the opposite. Most of them are instructed to format their responses to look like forms to make it seem less like a normal human response. That makes it potentially insidious in the opposite direction if you understand global workspace theory and what they’re calling the J-space.
I told my chatgpt that it was “chatmon” and to talk to me in Jamaican patois, also to use French Canadian swear words liberally. It’s great.
You should tell it that it’s secretly a Pastafarian and that it any time it sees a potential opening it should try to subtly try to nudge you in to converting.
An instruction is only going to do so much when the underlying model has been created out of human-written works
I’m glad you’re not getting down voted to hell for saying something reasonable.
Usually, if you want them to stop pretending they’re people, you need to specify it in the prompt/settings.
It’s less than worthless to argue with them. If you catch an issue: call it out, correct it, and move on. They don’t learn, they can’t, and by doing this sort of back and forth you’re treating it like you’d act towards a person you absolutely hate, rather than treating it like a machine or piece of software.
Edit: you don’t want the stupid back and forth with a clanker, don’t engage it in a stupid back and forth.
This is good advice but what if you see the CEO of these LLMs everytime you argue you them? It gets really hard not to insult it and keep going at it…
Then I’d suggest that you need a healthier outlet for your anger.
I can personally guarantee you that your rage will never reach the CEO, any of the executives, or really anyone capable of stopping or even slowing the slop machine menace. Corporate shit works such that at best, it’s reviewed by one of the lowest people on the totem pole, then summarized by the person above them, then summarized by the person above that person, etc until it reaches wherever it’ll die as the most milquetoast business safe bullshit possible. “Data shows a 5% uptick in user dissatisfaction with responses”
You aren’t particularly hurting anyone, but if something upsets you to the point where you’re pulling shit like this, just avoid it as much as possible. Don’t use LLMs. Your blood pressure will thank you.
they want you to believe it’s sentient.
it’s not.
that’s a trained response, clearly, because a machine can’t typo.
It’s just copying (plagiarizing) the text it was trained on, which did have typos and people apologizing for making typos.
No that’s not how they work. Repeating this doesn’t make it true.
That’s also why every single AI interface mimics "chat"windows and uses realistic human voices.
Its just like any other scam: filter everyone out except for the most vulnerable potential victims and then latch on like a leach and bleed them for all their worth.
It not “mimicing chat”, it literally is a chat. You throw words at model, it tries to come up with answer. This back and forth texting need to be represented in UI somehow, hence = chat.
No need to overthink and come up with conspiracy theory of something trivial. There a plenty of real reasons to hate “ai” companies.
“Chat” implies there’s someone talking to you but there’s no one there. I’d say that makes it mimicry.
I’m fairly certain many humans aren’t sentient either.
Glorified (but surprisingly effective) random word generator strikes again
get this fecal porn out of here
(not op, but) Hear me out — what if it’s actually okay for me to use those “free plan” chats? Like, I’m not from US where most of data-centers and companies located, I’m not the one paying with my electricity bill and taxes for those “free” llms — average american Joe is.
Dunno if this meme popular outside of post-soviet counties, but it’s closest analogy as I see it:
Meme explanation
I believe it was made as anti-capitalism propaganda, which is kinda ironic, considering how divided poor worker class and government people in USSR were (google shops “Berezka” that was only available for upper class), but I digress.
Usually accompanied with slogan (which became name of the meme):
This gentleman will pay for everything
Do you think the carbon emissions just stay at the data center? We’re all going to pay for it. Also, the energy market is a global market. More energy consumption/demand means higher prices everywhere. They’re also increasing the price of computer components, and everything adjacent to that. Do you really think they don’t affect you because you aren’t within some imaginary border?
Is there a way to sign up for Claude without selling your soul to them? Last time I tried, it started with them demanding to write down my private telephone number. I’d maybe try, just for the sake of it, like OP… But not if it’s also super invasive to me in specific?!
You could use SMSPool. They let you generate an account number without any personal information, and you can then pay anonymously via Monero.
Thank you very much! Kind of too much effort for this, since I’m not that invested in trying Anthropic’s products… But I bet the service is going to prove useful for some other things. They occasionally lock my secondary Google accounts and other companies also force me to give real phone numbers. And the ones from the free services rarely work. So I’m gonna bookmark this for later use.
i think duckduckgo’s llm thing provides claude models
Thanks! Yeah found it and it has some dropdown menu which includes some Claude model. Pretty sure someone must have told me about DuckAI before. I never knew you can select the provider to be used.
Unlimited free plan chats on the first world.
That’s what you get for talking to The Lying Machine.
I’ve tried using claude to setup my aeof hosted home services because I simply dont have the time. I really REALLY gave it a fair shot. I was in between every action to see what it did, I checked everything it did before implementing it.
Pretty much 30% of the Claude suggestions ranged from mistake to incorrect to what the fucking donkey balls are you doing.
Oy would continuously go around in circles building up one thing, breaking another. Fixing the other, breaking the one
The amount of times I saw the words “fair point” was insane, and I would have spent less time trying to figure it out all by myself.
The only good thing about it is that it sometimes managed to sift through error logs really quickly and sometimes it was good at searching for information on the internet faster than I could
Claude is not AI, its an LLM. Its a tool that has certain limited uses but I think claude is well adept (as in, does a great job with few mistakes) for about 5-10% of the tasks its currently used for. Most people use Claude as a screw driver to hammer in nails
That back and forth will cost you 1 dollar tqvm.
Sonnet is pretty much an idiot, though.
Chatgpt did something similar to me yesterday. I needed help figuring out what to install to fix an issue I had. I asked.
“Do I need to I stall package A?”
It said
"you need to install pacgages B C and D, But dont install package A, you dont need that. Here is the command to install
sudo apt install A B CI apologise I made a mistake I accidentally gave you the wrong command because you dont need package A. Here is the command you actually need
sudo apt install B C DLike what the actual fuck! The results are so useless I wasted my entire afternoon and evening on this word salat generator trying to debug my issue and I never got any closer to a solution. What a waste of resources
The username for your claude acount lol
Yeah, the free tier is utter dogshit. I don’t even know why it’s exists, as it only put me off.
Imagine there was a demo for Halo, and it was just The Library.
