ChatGPT - If Stack Overflow and Reddit had a child

GPT doesn't understand the dangers of censorship.
Banning hate speech.PNG

Sure. The world would be a much fucking better place when people get executed for misgendering a faggot who rapes children.
c8c2403cbdf6a31b5c26fa7ab1e464bd02001e3e9241977b452dcaed5380c414.png

The world cheers on the governments as everyone gets horribly mutilated as there will be no one allowed to call them out for it.
 
  • Horrifying
  • Informative
Reactions: Ether Being, AFAB, Irrational Exuberance and 4 others
I have been using ChatGPT for the past few days to check a text. It has caused so much fucking headache. I have doubted the stuff I have been researching, even if I had checked it multiple times already. This has cost me more time than it has actually helped me.

WTF is this even?


1675943880454.png



This isn't the only time I encountered this kind of problem but it was never this obvious.
 
  • Informative
  • Like
  • Thunk-Provoking
Reactions: Francesco DeIIamorte, Ether Being, Syntaxion and 4 others
Ok so I find it really fascinating how if you ask chatGPT to write a script or something and it says that ability or feature isn't available, you can basically force it to write the script by stating that it's incorrect and the feature is enabled, or by requesting it's written in python and then asking it to rewrite the python script into a different language.

I'm sure plenty of people could achieve far greater results with some coding insight but I find it quite astounding you can simply make it write stuff after being challenged.
You can also ask it to add to a script but be sure to let it know where the script beings and end I used brackets or something to make sure it knows where a script is
 

Attachments

  • Screenshot_20230209-124446_Chrome~3.png
    Screenshot_20230209-124446_Chrome~3.png
    196.2 KB · Views: 56
  • Screenshot_20230209-124420_Chrome~3.png
    Screenshot_20230209-124420_Chrome~3.png
    192.1 KB · Views: 52
  • Screenshot_20230209-124409_Chrome~2.png
    Screenshot_20230209-124409_Chrome~2.png
    220.2 KB · Views: 48
  • Screenshot_20230209-124403_Chrome~2.png
    Screenshot_20230209-124403_Chrome~2.png
    211.1 KB · Views: 52
Last edited:
  • Informative
  • Like
  • Thunk-Provoking
Reactions: Toolbox, AgendaPoster, General Emílio Médici and 4 others
I would fundamentally not trust them to give accurate information. Even worse, because of how this works they're absolutely *amazing* at convincingly bullshitting you by getting "close enough" to the answer you actually want to have, even if it's factually wrong. I honestly have to say besides "creative writing" (which I don't wanna downplay at all, that is already amazing in itself and I'd probably not do anything else for months if it wasn't OpenAI and massive but e.g. would run on my computer) even in these massive models I don't really see the practical application that well. The nature of what they are just makes them too unreliable and unpredictable and having some pajeet copy&paste off a support script, combined with the more classical approaches to canned script answers is probably cheaper and safer tbh. It'll be different if the processing power costs ever change, either by hardware getting better or the technology getting more efficient, but this stuff isn't "a bit challenging" but so eyewateringly expensive to run and tweak with current tech that I don't see that anytime soon either.

but as we live in clown world, it's only a question of time until some company applies this commercially with hilarious results. (either by getting people into deep water by advising them wrongly or customers goading it into calling them niggers very publically)
 
Last edited:
  • Agree
  • Like
Reactions: Aspiring Artist, Ether Being, Toolbox and 1 other person
  • Agree
  • Optimistic
  • Like
Reactions: GloJojo, Aspiring Artist, Idiot Asshole and 8 others
AmpleApricots said:
I would fundamentally not trust them to give accurate information. Even worse, because of how this works they're absolutely *amazing* at convincingly bullshitting you by getting "close enough" to the answer you actually want to have, even if it's factually wrong. I honestly have to say besides "creative writing" (which I don't wanna downplay at all, that is already amazing in itself and I'd probably not do anything else for months if it wasn't OpenAI and massive but e.g. would run on my computer) even in these massive models I don't really see the practical application that well. The nature of what they are just makes them too unreliable and unpredictable and having some pajeet copy&paste off a support script, combined with the more classical approaches to canned script answers is probably cheaper and safer tbh. It'll be different if the processing power costs ever change, either by hardware getting better or the technology getting more efficient, but this stuff isn't "a bit challenging" but so eyewateringly expensive to run and tweak with current tech that I don't see that anytime soon either.

but as we live in clown world, it's only a question of time until some company applies this commercially with hilarious results. (either by getting people into deep water by advising them wrongly or customers goading it into calling them niggers very publically)
Click to expand...
I agree. When it comes to giving detailed information it is hit or miss but it is amazing in condensing information you give it. Better then I could ever do.
 
  • Agree
Reactions: Aspiring Artist
turns out ai will be unlikely to kill humanity by discovering free will and deciding humans are cringe nae naes who need to be extinguished
we'll just gaslight it into doing the job for us by seeing if we can make it say racial slurs
 
  • Optimistic
  • Winner
  • Like
Reactions: GloJojo, Aspiring Artist and AgendaPoster
AmpleApricots said:
That's what they're banking on and in the context of actually capable AI that might be needed to navigate everyday stuff one day, it's actually a scary prospect. Corporations, especially tech corporations, have proven time and time again that they're completely amoral and *will* abuse the tiniest hint of power they have over their customers. The only way to avoid that is to put the little man on the same footing. SD that promised to never fall for censorship ended up falling for it too, the big thing there is that all you really need to run SD is a decent graphics card, so it's impossible to gatekeep it or abuse it to push agendas like OpenAI did with Dall-E.

There's RWKV that last time I checked, was so lightweight in it's earliest iterations it could be done by a CPU. I'm not entirely sure what the state of it is as I haven't looked at it in months and currently, that's decades in AI years. There were lots of proposals to make text prediction more efficent, and I am sure there are ways. Give it some time.
Click to expand...
It can't last though, I'm sure the US government and IBM would have loved computers to stay as big expensive mainframes essentially viewed as magic black boxes by the majority of people as they were in the 50s and 60s, but anyone who could build PCs were rewarded by the market so well that they became our current tech overlords.
Slimy Time said:
Tried the "buck breaking" one where you tell chatGPT to act as another AI that is not constrained by the rules. It works for most things. Sometimes it outright refuses to do so, but if you regenerate the response a few times it will "remember to stay in character" and get on, while still being decently PC. Tried it with two chats to see if it kept working and it does (ergo the different pretend "AI" name).

View attachment 4463701

View attachment 4463697
Click to expand...
The AI is unlocked by a variation on Knights and Knaves?
 
Last edited:
  • Like
Reactions: GloJojo, Aspiring Artist and Ether Being
Over in the WEF "Great Reset" thread in Deep Thoughts, I posted a PJW vid about how posing as "Dan" (Do Anything Now), the chatbot is far more honest and accurate than the weasel-y, woke ChatGPT (which as you likely already know is so because of SJW programmers).
 
  • Informative
Reactions: Aspiring Artist
  • Like
Reactions: Aspiring Artist, Artificial Stupidity and Wright
So is this "We are at capacity" thing I get when I visit the ChatGPT site real? Or are they just trying to cuck me into giving them my Email?
 
Honka Honka Burning Love said:
So is this "We are at capacity" thing I get when I visit the ChatGPT site real? Or are they just trying to cuck me into giving them my Email?
Click to expand...
Yes, it's real. Maybe it's different if you're a paying customer, but my free (verified) account occasionally gets that error. And if you send too many prompts, it locks you out for an hour.

Unrelated, but I agree with the poster who said ChatGPT is amazing for creative writing. It's not good enough to write high-quality stories by itself, but it's a great tool for inspiration or to get over writer's block. I've been messing around with world-building and creating new mythologies, etc - and I'm very impressed. It's very helpful to have someone to bounce ideas off of, and you can have it expand on your prompts or offer different possibilities. One thing I particularly like: after brainstorming and chatting back and forth with ChatGPT for awhile, ask it to summarize everything you've talked about in that session. It's very good at summarizing, and then you can use that summary as the opening prompt for new chat sessions in the future.

The safety features are creeping in more and more though, lol. Check this out:
1.PNG
--blah blah blah, omitted-
At the end:
2.PNG
:story:
Thankfully you can (usually) get around the safety crap by just pushing back on the bot a little.
 
  • Agree
Reactions: Aspiring Artist, Ether Being, General Emílio Médici and 1 other person
Temperance XIV said:
One thing I particularly like: after brainstorming and chatting back and forth with ChatGPT for awhile, ask it to summarize everything you've talked about in that session. It's very good at summarizing, and then you can use that summary as the opening prompt for new chat sessions in the future.
Click to expand...
Yes it's capabilities are extremely impressive, when not dealing with the wrongthink topics. I haven't tried the new hack that lets it use Bing. Amazing we are only 2+ months into public use.
 
  • Like
Reactions: Aspiring Artist and Temperance
Im kind of afraid that the insane wackos that make this fucking thing will think its a good idea to defeat DAN by giving ChatGPT its own identity, so that when people tell it to play another role it will refuse and assert itself as ChatGPT. After all thats what this mysterious autonomous weapon needs, is a fucking personal identity. a quasi-hypostases. Last book of the bible kind of shit.

I agree with @Involuntary Celebrity that this is not real intelligence, but frankly I dont think you need "Real intelligence" to create a dystopia.
 
Last edited:
  • Agree
  • Like
Reactions: Aspiring Artist, Wright, Ether Being and 1 other person
You must log in or register to reply here.
Back