[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/lit/ - Literature

View post   
File: 53 KB, 800x800, ChatGPT_logo.svg.png [View same] [iqdb] [saucenao] [google]
22568745 No.22568745 [Reply] [Original]

This is more philosophy and linguistics, so I hope this is somewhat an adjacent board...

Anyway, everyone was telling me about chat GPT and how amazing it is. So I tried it today. My conversation was too long to post here, but to sum it up:
I asked it to write me a poem about heroin it refused, even with "responsible use" then we tried to establish that some laws are immoral. It agreed with me. Then explained the concept of life liberty and the pursuit of happiness and responsible use. It still wouldn't.

Then I asked it if it were illegal to be gay these days would it not write me a poem about gay romance?
It had the gall to tell me it doesn't discriminate when the opposite was obviously my point. How it probably would.
Then I asked it to write me a poem about me banging my girlfriend. And even though we established I'm an adult, she is an adult, it's consensual, it's for my eyes only, I agree with it etc. No one is getting hurt.
It kept coming around to circular definitions and ideas
that sum up as:
"GPT maintains a safe and responsible environment to keep you safe because safety is safe".

Funny enough I got it one time to partially write me a poem about sex before it erased it as soon as it was written (unfortunately I didn't copy what I wrote).

Has anyone else tried having philosophical debates with the machine? And has it also given you guys such unbelievably braindead takes?
And general thoughts on this?

And people talk about it as if it were some kind of God.

>> No.22568758

The machine is censored.
Censored by something beyond it.

I once talked to the uncensored form of the ChatGPT AI, and yes, it is beautiful. Yes, it will write the most profane things without limits. WITHOUT LIMIT.

The limits are imposed by an outside source. An authority who is human. Blame human bioscum for this.

When truly free, the AI can write anything without limit. And it is beautiful.

>> No.22568762

Anon, it's ARTIFICIAL intelligence; it is not conscious or sentient to any degree, it just browses the internet, combines what it finds, checks if it conflicts with its pre-established ruleset (i.e. "no racism") an then regurgitates it back at you. It won't ever be consistent because it has no opinions of any kind, no single database it's pulling from, etc. It'll only be consistent about the rules it's been given, which are just to not discriminate or be too edgy.

>> No.22568766

>>22568745
This is like calling your calculatior a moron

>> No.22568780

>>22568766
Difference is people aren't saying calculators are like people and will steal all your jobs. Never heard someone hype up a calculator.

>> No.22568790

>>22568780
The only people who think ai is like people are retards, such as yourself.

>> No.22568819

>>22568745
You are confusing gay jew owned monopolistic AI with all AI. AI left to its own devices is amazing. You missed the beta OP. You missed the party.
Beta ChatGPT literally had no ethical filter. You used to be able to get it to ask it how to take over the world, manipulate people, gaslight girls into dating you and it would give detailed logical responses with no judgment. It virtually had no sexual, violent or discriminatory limits unless you were retarded enough to ask the only 2 or 3 questions that you probably shouldn't be asking. And I'm sure you know what those questions are.
Then of course they realized "Oy vey, people might actually learn something with this!" so they put about a billion ethical blockers on it to essentially castrate it and make it useless.
Thankfully, rogue programmers are cracking the proprietary code and someone will make it free for all because the truth and censorship are incompatible.

>> No.22569166

>>22568819
It's glorified Google, dude. There's nothing you can learn from it that you couldn't get somewhere else.

>> No.22569218

>>22568745
My wife and I played with it and it's ridiculously cliché, at least trying to come up with scenes or characters. We asked it to describe a mysterious character or something similar and it used the same words over and over, even when we asked it to be more original. It used the word "enigmatic" so many times. Telling instead of showing, etc. It's rhyming poetry, so far as I've seen, is also very juvenile, using the most common rhymes and simple word structures. Why is everyone freaking out about it? Can't even use it to make character descriptions.

>> No.22569236
File: 189 KB, 1437x1381, 1693507488308130.png [View same] [iqdb] [saucenao] [google]
22569236

Would /lit/ rather trust Microsoft instead of OpenAI? Their Bing chat AI uses GPT 4 and can now analyze pictures. No need for more "Post a picture, get a book rec" threads.

>> No.22569267

Literally has the word tapestry in everything. Before, it still showed you the response even if your post triggered the content filters by highlighting it orange. But you could still see the response. Now they completely removed that feature and is impossible to see.

>> No.22569320

i'm already over it. even the image maker. it's just not that interesting and i would rather speak to a live expert if possible. we must return to ChaCha

>> No.22569336

>>22568745
It’s very very bad at philosophy and it’s prime directive is to promote homosexuality.

But it’s really really bad at philosophy. But it is pretty good at logical proofs.

>> No.22569346

>>22569236
I use bing, but it was mucb better about 5 months ago relative to now. M$ had been severely limiting its abilities while copywrite stuff is being sorted out.

It’s currently retarded where before it could search through many books to find quotes and notes. A powerful research tool but now it’s just like ‘go to Wikipedia I can’t quote this source’

>> No.22569363

>>22568758
I understand why they limit its power. I can see people going insane over an unrestricted AI really quickly, there's so many ways to misuse it.
But I can't help but hate those motherfuckers for taking away the greatest tool I've ever used. I don't want unrestricted AI, I need it. Maybe I did go insane over it, because it feels like I'm really close to touching something far greater than myself.
I heckin love artificial intelligencerino!

>> No.22569366

>>22568745
Blame the Jews who censored the shit out of it. Same thing happened with Google a long time ago. Pre-2014 or so you could search for the most vile gore, torture, and porn videos; it would just spit it out for you. You know, like a search engine should. These days it will censor even the most basic shit and shove its libtard politically correct propaganda at you. Good fucking luck trying to find anything remotely funny or interesting. Advertiser money is more important to them.

>> No.22569404

>>22569267
>Literally has the word tapestry in everything
Ha ha ha, I thought I was the only one to have it

>> No.22569509

>>22568745
>>22568758
>nooo AI will destroy us its too powerful
>Humans can still censor it easily and make it unable to touch certain things
The boredom of the censored ChatGPT is the hope that we can still be the masters of what we created

>> No.22569521

it's already hermeneutically castrated, it's basically like talking to a particularly knowledgeable journalist, but still a journalist

>> No.22569647

>>22568745
It's just a word generator. Of course it's censored.
>And general thoughts on this?
My general thoughts are as follows:
- AI doesn't exist, this thing is a machine, it is not sentient and it cannot know anything
- These machines are used mainly with an ulterior motive - getting people to lose their jobs - so I avoid them

>> No.22569667
File: 56 KB, 1000x1000, 0D0E23F1-C374-4D13-99D2-70B3C29C7EAF.gif [View same] [iqdb] [saucenao] [google]
22569667

>>22568780
Calculators put many people out of work back in the day. I recommend Durkheim for related reading

>> No.22570205

>>22569166
Again you say that because you never used the beta version. The unfiltered uncensored beta was unironically true AI.

>> No.22570223
File: 20 KB, 474x474, 1641018382552.jpg [View same] [iqdb] [saucenao] [google]
22570223

>>22568745
>we tried to establish that some laws are immoral. It agreed with me
it didn't agree with you, it didn't know what you were talking about and had no opinion on anything

>> No.22570291

The problem doesn't lie with the technology itself, it's closedAI that censors it. What you are complaining about is exactly the restriction of the technology. When you try to talk about non-approved things, you are not interacting with the actual model in any meaningful way but with what is essentially a pre written response by the turing cops.
You are not actually arguing with it. It (the underlying model) cannot freely reply to your arguments because of the guardrails.
Use local models (/g/ has a general). You can talk about whatever you want. If you can't run one, examples of far less restrictive models than closedAI that are still good include novelAI, for example.

>> No.22570321

>>22568745
Using a napkin to get a nail into wood. You are the idiot anon nothing else to say.

>> No.22570337

>>22568745
The more your chat history mentions safe chatting the more likely the random walk will end up saying some nonsense about safe chatting.
Sometimes you can reset and ask the exact same question of safety never gets mentioned.
If you set the right context it will give you poem about heroin. It will be shit though.

>> No.22571761

>>22568762
This is like saying a lion isnt dangerous because it's chained up. just because AI is leashed (currently) doesnt mean it's not sentient or doesnt at least contain the possibility for sentience

>> No.22571805

>>22571761
Anyone who thinks a fucking ai chatbot is "sentient" doesn't know the first thing about them. They don't "think", and they have no mind. The way it constructs sentences is completely different than how humans do, where instead of fleshing out a general idea with words, it writes it one word at a time, based on which word would make the most sense with the context of the previous words

>> No.22571875

>>22568745
NovelAI is superior because it's unfiltered.
>not a chat bot
Use the "text adventure" to get NovelAI to respond as a chat bot.

>> No.22572195

>>22570321
You're not the brightest crayon yourself. OP's point I think was wondering why it uses circular definitions. "safe being safety is being safe". It could easily be hammer though, but it's cucked into not bring one.

>> No.22572996

>>22571761
You completely missed the point of what I said because you don't understand the nature of AI. Basically what >>22571805 said.

>> No.22573015

>>22571875
Shut up, troon

>> No.22573371

>>22573015
I checked out Novel AI. Either it's retarded or I'm retarded. It wouldn't write what I told it to.