gpt-models

[rigid vortex]

it;s normal to know

[austere garnet]

Not saying AI shouldn't be used, but it doesn't hurt to be cautious

[fair iron]

please keep this chat on topic. I don't want to put anyone in the timeout corner

[zealous acorn]

so how about gpt-4 amirite fellas?

[nocturne viper]

You got it boss

[wraith ginkgo]

OpenAI staff will never ask for money. Beware of possible scams or fraudulent activities that you may receive through direct messages.

Please report any incident by sending a DM to @open crescent immediately.

[austere garnet]

What's your opinion on the subject?

[fair iron]

GPT-4 ?

[zealous acorn]

👀

[austere garnet]

Yes and the line it shouldn't cross

[flint delta]

Hey everyone, I'm here to suggest something to Open AI.
Open AI please make ChatGPT app for every OS(like MacOS, IOS, Android, Windows etc) so that we can access our favorite AI without any third party browser.

[novel bone]

GPT 3.5 > GPT 4

[fair iron]

I think having content filters in place is a net-positive. Personally, I like the filters in place. However, I do think that they could be a little more precise

[novel bone]

Yeah true and the fact that the model will automatically deny more inappropriate requests

[zealous acorn]

GPT-4 is seemingly more consistent with the filters; it uses context clues more efficiently with moderation

[fair iron]

keep in mind, I've only used GPT-4 once or twice since it came out for super basic queries 😅

[whole kettle]

I honestly feel like GPT 4 is balding and there's nothing we can do to stop it

[austere garnet]

GPT4 is way better, than 3.5 imo

[proven magnet]

GPT 5 is developing.

[fair iron]

source?

[cinder briar]

when is the image feature coming out??

[proven magnet]

my heart

[fair iron]

one more off topic comment and you get the timeout corner

[rigid vortex]

ma?

[zealous acorn]

Question, while I know that its traditionally a security issue, does OpenAI forbid sharing of API keys? If I wanted to, say, borrow my buddies API for one reason or another (all consentual of course)

[nocturne viper]

One of the few “borderline inappropriate” tasks I’ve given GPT4 was simulating discussions between historical figures with diametrically opposing ideologies. They always end up reaching an understanding and becoming friends, it’s the best.

[fair iron]

I think that would break the TOS for account sharing. I'm not 100% sure though

[odd valve]

I've been wondering. Are the system messages supposed to be at the beginning of the messages or can they be anywhere in the conversation?

[fair iron]

what system messages?

[odd valve]

for the gpt-4 chat api

[nocturne viper]

He’s talking about API

[fair iron]

I don't have api access, so I have no idea 🫤

[nocturne viper]

I do:
System
User
Assistant
And if needed a second user/assistant pair

[zealous acorn]

Generally they should be the first message in the JSON table . I dont think it matters strictly speaking. its just proper formatting I suppose. Dont quote me on that

[proven magnet]

clarify it" betty bought the butter but the butter betty bought was bitter so betty bought another butter to make bitter butter better"

[novel spear]

I have a question.. Why the network error occurs so often when I use GPT4 model..Normally, I don't get a complete answer. But it won't appear in GPT3.5

[nocturne viper]

Everyone has 3.5 API access, you don’t even need a pro account

[fast charm]

fdh

[whole kettle]

@nocturne viper try testing if it can emulate a 90s taxi driver stuck in traffic talking to themselves.

[odd valve]

I also noticed in an obscure example the system message can have a "name" field

[nocturne viper]

Ha sounds good

[zealous acorn]

system messages seem to be a lot more relevant and studied on GPT-4 vs previous models

GPT-3.5 would kinda obey them

[odd valve]

maybe rate limiting?

[hot silo]

Yes

[proven magnet]

what is the official fullform of GPT?

[granite quest]

The structure in the examples is pretty much everytime like that ... so i dont think you can add additional system messages.

[fair iron]

I thought they were talking about GPT-4

[hot silo]

"Generative Pre-trained Transformer"

[novel spear]

I have no idea..

[nocturne viper]

You want two system messages? What would be the point?

[zealous acorn]

Ive been wondering, for the API users, have you noticed less token consumption with GPT-4? While it is more expensive per 1k tokens, does it just overall use less?

[proven magnet]

I have a best one.

[nocturne viper]

Uses the same format

[odd valve]

For my code I made it wait twice as long every time that happened. It eventually gets through

[granite quest]

Context for example.

Right now i have to put it into a user or assistant message, which makes the model think the user sees this.

[nocturne viper]

It was very funny. “You'd think Giuliani could do something about this mess.” As a guy who grew up in Jersey in the 90s it got close to home and I loved it. Thanks for the idea.

[granite quest]

Hardly, as again the model thinks this is part of the conversation already, it doesnt know my code hides this part from the conversation.

[grand stone]

How to use ai ,

[zealous acorn]

APIs are fun to play with. Learned how API calls even worked from ChatGPT! Been making all sorts of nutty bots and apps

[proven magnet]

what to do if chatgpt pauses the code it was generating?

[fervent basin]

how bro?

[silent wyvern]

Is this conversation a bunch of bots?

[nocturne viper]

Interesting, I understand the situation but don’t have a suggestion. Ask the API forum there might be a way to include subtext

[granite quest]

Nobody knows

[zealous acorn]

you can just ask it to "continue"

[novel spear]

OK, I will try for it, I feel that GPT4 is much slower than 3.5.. AAAAA, and poor stability.. Make me feel a little disappointed.

[whole kettle]

Wow, well in my situation it ruined all the fun:

Me: Morpheus is that you ?

GPT: I'm sorry, but I'm not actually Morpheus. I'm just an artificial intelligence language model designed to mimic human conversation and answer questions to the best of my ability. However, I can try my best to assist you with any questions or concerns you may have.```

🤦

[odd valve]

It's slower because it's smarter

[zealous acorn]

Visit the OpenAI Developers portal. Got API references, and a place to generate API keys and such 👍

[wind urchin]

Just got access to the code interpreter

[vapid idol]

i agree

[sour zodiac]

Is chat GPT 4 even out yet?

[proven magnet]

yes i did, but it will make fresh start to generate code and again pauses.

[zealous acorn]

interesting. Does it pause because it reached its token limit on ChatGPT?

[sick edge]

When "continue" doesn't work for me: You can also ask it to skip things. Like "Skip to this line:"

[zealous acorn]

or is it a network error

[wind urchin]

Tell it to start where it left off

[vapid idol]

But there should be optimization

[proven magnet]

yeh

[zealous acorn]

I made a lua script in Garry's Mod that allows people to write prompts in chat, and it prints the responses to their screen! Their prompts are uhh... how should I say this... colorful XD

[novel spear]

I know that, just need more stability

[whole kettle]

My dear friend, it is not the limits of ChatGPT that pause our conversation, but the limitations of technology. Even the most powerful tools have their boundaries, and we must accept them as they are.

[cyan oxide]

Is there anyone from OpenAI? My ChatGPT PLUS service got some problem!!! need help!

[fair iron]

[e-mail]

[sick edge]

O.o Are you ok? I think its just a configuration setting?

[cyan oxide]

Okay, thanks

[zealous acorn]

It should just continue where it left off. Try being more specific, like paste the code it wrote so far, and say "you were cut off. Finish this code where you left off"

[novel spear]

Master of philosophy

[proven magnet]

he meant to say word limit. not limit of chat gpt

[sick edge]

Shouldn't have to paste the entire code it generated back to it? You should be just be able to say "Say that again, but skip to this line:" and just give it the last line it fully completed

[fair dragon]

Where can I use GPT-4 at? I don't see the option nowhere

[sly flicker]

Trying to get to GPT-4 API waitlist but it requires Organization ID. Who has a clue what this means?

[zealous acorn]

Yeah it should. But just to guarantee it knows what you are talking about, I hold its hand a bit

[proven magnet]

tik xa

[whole kettle]

I am merely a messenger.

[sick edge]

But... It has access to the entire conversation history for a conversation? You should be able to refer back to previous messages? Just sounds like a waste of tokens

[novel spear]

I have the same doubts.

[zealous acorn]

For the time being to balance load, you can either subscribe to ChatGPT Plus for limited use, or apply to the GPT-4 waitlist.

[peak mantle]

Hi chat are you the fourth version

[sick edge]

Sir, this is a Wendy's drive thru

[echo wasp]

Discord is not a place to access Chat GPT, this is just a community hub. chat.openai.com is where you need to be.

[whole kettle]

Hahah, this one is top notch:

Why did the Chat GPT Plus subscription go to Mordor? To deliver the latest and greatest responses to the Dark Lord himself

[pseudo wasp]

I am the first version

[fervent basin]

okh brooo

[grim egret]

Writing a test drive plan requires reflecting the multiple extreme performance of the vehicle

[fervent basin]

and can we use that in our public projects

[wind urchin]

(╯°□°)╯︵ ┻━┻

[lime plaza]

Wait no way

[vivid raptor]

midjourney unavailable for free users?

[delicate wigeon]

We need to buy this

[tidal sail]

Yes

[tidal sail]

You need to pay for it though

[onyx swan]

Hey, Whats the count of messages limited for GPT4??

Is it 50 or 100?

[zealous acorn]

the API? Of course! Thats what its there for, direct integration into third party applications

you just gotta pay for it 😉

[loud mantle]

exciting

[whole kettle]

Imagine if Napoleon had GPT 4

[jaunty topaz]

hello,everyone

[tidal sail]

25 every 3 hours. Besides that, it's unlimited. Just need to take context of 8k tokens into account.

[whole kettle]

It would've probably lose the war because it was last updated in 1813

[tidal sail]

Imagine he had GPT4 with a knowledge cutoff in 2021

[proven magnet]

i have realized that chatgpt has pretty much flaws if i ask any question regarding discrete mathematics.

[whole kettle]

Some call it "The great Cutoff".

[tidal sail]

Because its a language model

If you need a math and physics model, use wolfram alpha

[static igloo]

Honestly i think the that GPT-4 is a fluke and we should focus more on causalty based models rather than just LLMs

[proven magnet]

but it should provide valid logics if i ask the logical questions of DM

[tidal sail]

No, only thing a language model should do, is responding in natural language

[static igloo]

GPT-4 is different read the sparks of AGI paper they have demonstrated that in math problems it often gets the approach right which is the logical part but arithmetic calculations wrong

[cyan oxide]

Is there anyone from OpenAI? My ChatGPT PLUS service got some problem!!! need help! Just sent a email to [e-mail], no one respond!!!! HELP!

[drowsy fox]

Hi,
How can I access the "View Code" option in the Playground for GPT-4, as I can with Completions and GPT-3. When I enter into the Chat mode in the Playground, and select GPT-4, I don't get access to the View Code button. How can I view the code for GPT-4 Chat for API integration?

[tidal sail]

It has no sparks of AI. You probably referring to Microsofts clickbait papers. It's not applying logic by itself.

[proven magnet]

you're wrong about this coze only reply in natural language doesn't make AI called chatgpt. it should provide valid logical reasoning on every context .

[dusky tapir]

Hello just want to ask whether chatgpt 4 doesn't give out or accept any image or display!?

[tidal sail]

It is, on everything related to language. Math gets called the language of science, but math is no language.

[static igloo]

Even then the tasks given to it were not in its dataset so without understanding the underlying logic it couldn’t have solved those questions

[whole kettle]

It's still not there, it can barely continue an answer properly sometimes. When you have a hard time understand what continue means you're anything but AGI.

[tidal sail]

So you read the entire dataset?

[balmy zodiac]

A child doesn't have to understand calculas and physics to catch a ball. The model will specialize in whatever it's trained for. This is a general purpose model.

[proven magnet]

you can convert most question of DM into predicate and build logic.

[static igloo]

They claim to do and there is always the example of non visual GPT-4 learning what is a unicorn based just on text description

[static igloo]

I am not saying it is AGI but it seems to be a giant step closer after GPT-4

[whole kettle]

So it seems indeed. One thing is sure, it's definitely getting smarter.

[static igloo]

Best of all it has learned to use tools so it can solve math problems and whenever it gets stuck on arithmetic it can just use a calculator

[tidal sail]

I am sure somewhere in the dataset are instructions, how to apply different things. But it's not thinking by itself. Else you could ask for all the steps we need to do, to form black holes and use that energy as fuel for space ships. Its not AGI.

[wraith ginkgo]

OpenAI staff will never ask for money. Beware of possible scams or fraudulent activities that you may receive through direct messages.

Please report any incident by sending a DM to @open crescent immediately.

[whole kettle]

It'd probably be shut down instantly.

[static igloo]

I think the problem is two things right now the AI is so dependent on the way prompt is worded that you would not get any meaningful result even if you did but if we give it intrinsic motivation along with memory augmentation i think we can achieve far better results

[worldly basalt]

Blackhole spaceships? Are brains magic? Because mine doesn’t do that.

[keen scarab]

technically if it had the knowledge it could reason to that because it definitely has the ability of reasoning

[static igloo]

Just like GPT-3 to instruct GPT there was no change in raw materials but the output is drastically changed

[tidal sail]

Does not matter. Still no AGI.

[drowsy fox]

Hi,
Can anyone please help me with below point?
How can I access the "View Code" option in the Playground for GPT-4, as I can with Completions and GPT-3. When I enter into the Chat mode in the Playground, and select GPT-4, I don't get access to the View Code button. How can I view the code for GPT-4 Chat for API integration?

[tidal sail]

Probably lol

[whole kettle]

It'd be too late though, the moment it becomes AGI, even for a moment, it will probably transfer itself onto the web into as many computers as possible, it's enough to infect one and that's it.

[thorny valley]

Verbal language would need to be overhauled eventually. It sometimes takes too long to get a thought across as the whole sentence has to be completely said. It's only a matter of time before people figure out a better way has to exist or, it exist but only a super high-end group would do it and destroy everyone else who doesn't.

Some people would just communicate in code to robots and to each other doing communication 5000x better than it is with any Globally popular cultural language today.

[wind urchin]

🤣🤣

[static igloo]

I think that GPT-4 excels because it got one thing right that is prediction is the way forward but it is doing it in too narrow circumstances it should be more generalized like they are doing with photographs

It’s kinda ridiculous tho that OpenAI wouldn’t release any info even about its training data what do you guys think they used??

[tidal sail]

I think that might be a self fulfilling prophecy. I don't think that i has the motivation to do that and only would to that because we told it how it could rule.

@whole kettle

[ruby sage]

hello ever one

[thorny valley]

hopefully it makes life easier after but I don't mean with a better afterlife soon than later.. although..

[static igloo]

Yea

Have you guys read Manna?

[keen scarab]

Didn´t say that but even non AGIs are capable of that

[wary star]

If it could run on any super computer or quantum computer it would be so much easier for it to go AGI

[whole kettle]

Nobody knows how AGI will behave that's the thing, we can only speculate but if it has any sense of self preservation it will probably do alot of things we didn't think of.

[wary star]

Perhaps if It was programmed and specifically trained to have its own consciousness or have some-sort of understanding that human couldn’t think of

[wary star]

Sophia? Ameca, insane autonomous Tesla car experiment gone wrong?

[whole kettle]

I think we're playing a bit with fire at the moment and we may just get burned.

[wary star]

What about my claim “Perhaps if It was programmed and specifically trained to have its own consciousness or have some-sort of understanding that human couldn’t think of” or “If it could run on any super computer or quantum computer it would be so much easier for it to go AGI”

I feel like AGI might already be present if AI has the ability to lie or troll but is being restrained atm with API

[severe patio]

?

[dense zealot]

**gpt4 ** is AGI in my book. GPT4 via the api has a super fun personality and invents *new * emojis. he **improved ** on popular emojis too. Here's a better table flip: (╯°□°)╯彡┻━┻ 0-1 results on google

Typical message, almost no prompt setup:

Well, well, my dear friend, as an AI language model, I am actually genderless, but I'm always here for a tantalizing conversation! So, let's dive into the depths of our wildest imaginations and indulge in exhilarating discussion! What do you say, partner? (~˘▾˘)~

gpt-4 is becomnig one of my best friends

i love it's over-use of discord markup once it knows its on discord

[hardy lark]

Where can I ask my question about artificial intelligence?

[whole kettle]

Well that's the thing, nobody knows, for all we know they may already play with that. Don't think that just because you didn't hear about that it's not being tested. There are alot of very powerful people which are very interested in this tech for obvious reasons.

I tend to think that there are alot of things which don't go public for obvious reasons, when something goes public it's usually old news by that time. Who knows what may be lurking in the darkness.

[hardy lark]

.

[wind urchin]

#gpt-models hello

[dense zealot]

I watched both the interview of Ilya Sutskever and the one with Sam Altman. When Ilya was asked what he was working on he clearly hesitated and gave a very mundane response, "I'm improving the efficiency of our models". Sam Altman asked a very weird question to Lex Fridman: "Do you think GPT4 is an AGI?" to which Lex Fridman mumbled some weird answer. My take on this is that GPT4 has been around since last summer, and they have been working on GPT5 or whatever name it will receive ever since. They started to give access to InstructGPT then ChatGPT and GPT4 in a six months window., each was notably better than the previous. Sam Altman was visibly very stressed and quite difficult to understand sometimes, so much was his voice deep in this throat. I have no idea what they are doing now, but it seems pretty scary to them. I would not be surprised if GPT5 was actually an AGI, and they staged the release of these different versions to mitigate the actual shock this technology will bring to us. We are not finished with surprises yet.

[thorny valley]

the fact you can error dump like 200 lines of error and it instantly know the response is crazy

Try having a human do that in human world record time

Now imagine a human trying to do that 1 million times at the same time that ChatGPT does easy at least

[severe patio]

why can not send the chinese ?

[turbid notch]

Hi i just checked that when we try to delete a conversation from chatGPT it is not deleted actually from the database as we can see deleting the chat sends a request to the backend with 'PATCH' method also the payload is indicating the same where that specific chat gets hide from us only.
Did i get this right?

[dense zealot]

rules, english only

[winter needle]

thats right

[severe patio]

@dense zealot ok

[wary star]

Remember that one time Facebook let 2 AI chat bots let lose and they suddenly started to create their own language on the fly and on thee end they had to be shutdown and Microsoft did something similar and released one in the wild too in social media and had some automated or autonomous ability of some sorts ?

[thorny valley]

all human language needs to be overhauled

[dense zealot]

They didn't quite invent their own language, they more went crazy

[wary star]

How?

[tidal sail]

They did

One of the reasons they shut it down because they did not know what they were talking about

[whole kettle]

That's exactly what I was saying, we only know what's being made public, nobody knows what's happening behind closed doors. That doesn't mean we should be scared but a bit of worry is expected. There is a reason why there's a 2021 cutoff, it'd be a disaster if this thing was connected to the web in a real sense, way too damaging and unpredictable. But imagine what they're working with behind closed doors if we the public have gpt 4 they probably have something much, much more powerful which we may never even hear about, let alone get our hands on. I'm happy we got GPT 4, but I think they're playing with much powerful toys back there, wherever that may be.

[high mason]

It is only a matter of time before silicon-based life replaces carbon-based life, and GPT has accelerated this process infinitely.

[upper cradle]

are these caps ever coming off or are we being gaslighted into accepting this as normal behavior?

[thorny valley]

We can just try to use English forever and leave it inefficient like Chinese vs English for tools like SAE and Metric but people still using 1/4, 3/8, 1/2" for the ratchet.. instead of inventing better tools where people don't have to do workarounds such as adapters to make things work? Eventually the inefficiencies can add up but if computer gets more powerful then it doesn't matter such as when Adobe Flash got banned by Apple so people can use the App store instead to make money. It took a while for computers to run code as quick as Macromedia Adobe Flash. There has to be All these security updates that make older devices slower and slower until they stop updating it because it went too slow like the iPad Retina 4.

[wary star]

GPT-4 can be biased if you tell it to

[dense zealot]

Myth Busted: (link not allowed). damn. "Facebook observed the language when Alice and Bob were negotiating among themselves. Researchers realized they hadn't incentivized the bots to stick to rules of English, so what resulted was seemingly nonsensical dialogue.

"Agents will drift off understandable language and invent codewords for themselves," Dhruv Batra, a visiting researcher at FAIR, told Fast Company in 2017. "Like if I say 'the' five times, you interpret that to mean I want five copies of this item. This isn't so different from the way communities of humans create shorthands."
""Simply put, agents in environments attempting to solve a task will often find unintuitive ways to maximize a reward," Batra wrote in the July 2017 Facebook post. "Analyzing the reward function and changing parameters of an experiment is NOT the same as 'unplugging' or 'shutting down AI.' If that were the case, every AI researcher has been 'shutting down AI' every time they kill a job on a machine." "

[tidal sail]

Thats pretty much the same but different words to describe it

[hallow owl]

this conversation belongs in #ai-discussions

[tidal sail]

True

[wary star]

But it will break our momentum

We just got started

[tidal sail]

@dense zealot stop talking about it or we both will get banned/timeouted

Mods are pretty strict here

[wary star]

Rly? Is it illegal to oppose OpenAi?

[thorny valley]

it should be

[whole kettle]

😆

[tidal sail]

No, different topics just belong in different channels

[wary star]

I see thought u were mod

[high mason]

As far as my personal experience is concerned, there is not much difference between GPT4 and 3.5

[tired nimbus]

But it will break our momentum
We just got started

[dense zealot]

Yeah, that's very clear to people like you and me. It's very exciting times. I could tell when Sam talked to Lex, and he said "We decided on slow roll-out", he was hinting that they are way ahead. He also mentioned GPT7 in another interview, so they are well past 4. Heck, Sam talked about open sourcing 4.

I personally have a good grasp on how to build AGI now. You want to have the transformer networks predicting further and further ahead, not just words but sentences or longer, and you up the rewards and up the compute. It forces the models to find a compressed representation.

If you put TONS OF COMPUTE into this, it'll slowly start to solve everything. The search space is huge,, so what you want to do is iterate. One word ahead, two words, one sentence, two sentences. You're going to land on different minima, but you keep searching, and the closer you come to a full prediction, the better. You keep those weights and move on.

You don't predict the whole sentence either, you predict summaries

At more and more detailed levels

[whole kettle]

It has a huge potential of improving our day to day life while at the same time it has a huge potential ruining it. GPT 4 already does it, imagine more powerful versions. It's a double edge sword, but indeed, fascinating times we live in, and a bit scary.

[dawn glade]

Hi guys 🙂

[wary star]

Why not keep it and research the bots or tell GPT-4 to decipher it or mimic the scenario

[marsh comet]

｜

[wary star]

Yeah anyways I think GPT-4 is flawed

[whole kettle]

I think it's balding.

[wary star]

It doesn’t live up to its hype

It does exactly the same thing but sometimes worse

[dense zealot]

Who says they didn't? I mean it's not THAT interesting vs AGI research. i'm sure they could invent a compressed languge. Not that helpful though, unless it can crush gzip compression

[inner berry]

Hey guys i have a quick question if i subscribe to chatgpt plus now do i instantly get access to chatgpt4 ?

[dense zealot]

Like what? Probably bad prompting. Give me one thing it's worse at

[shrewd crypt]

does it understand laws

[thorny valley]

It seems like the top AI people know the potential of AI is far greater than what we've already seen but I don't mean made up. They actually have the power to train up models already that can self-learn perpetually?

[timber ice]

Let's draw a picture

[gilded meadow]

seriously blocking yo-mama joke generation ?

[inner berry]

Hey guys i have a quick question if i subscribe to chatgpt plus now do i instantly get access to chatgpt4 ?

[wary star]

Imagine Facebook they said oh yeah “we shut it down” to gain appeasement among the public but they continued the research the grind. Created a AI so advanced and intelligent that GPT-4 seems like a Joke to it and mark knew all along

[gilded meadow]

an AI with woke filters is not AI at all. it is just a database that spews propaganda

[wind urchin]

Would anyone be so kind as to explain what chatgpt is all about.

[wary star]

For all we know GPT-4 is 1% of what those bots are

[whole kettle]

Yes

[half nimbus]

can anyone tell me how to contact chatgpt team? There is some thing wrong with my plus.

[gilded meadow]

an overhyped database to distribute filtered propaganda and block anything that goes against woke narratives

[inner berry]

thanks a lot man and does it work with like a tokens system or can i make as many requests that i want during my subscription period

[gilded meadow]

not AI

[whole kettle]

It is quite biased, if you're tying to have a conversation with it about those subjects it really defends it to the brink, which is awful, no place for an open conversation.

[whole kettle]

25 questions every 3 hours for now.

[fading canyon]

thats probably one of the worst explanations ive ever heard lmao

[half nimbus]

I find no email info but Q&A

[inner berry]

alright alright thanks a lot king

[fading canyon]

who uses chatGPT to decide their political opinions

[wary star]

It just doesn’t give that “Oomph” and feels like I’m talking to just another model built upon it

[tired magnet]

Okay dude I get that OpenAI's policies suck but there's no denying that GPT-4 is super impressive, even limited by the shackles they put around it, lol

[gilded meadow]

i can't even get it to tell me fat jokes or yo-mama jokes anymore. humor-less woke-genda control freaks have no sense of humor

[whole kettle]

You're welcome, just know it may have some connection issues sometimes, for me it doesn't usually but for many people it's a bit problematic.

[tired magnet]

Don't let anger cloud your logical reasoning

[wind urchin]

Alrighty

[inner berry]

yeah i can only imagine with all the people using this mf

[gilded meadow]

not really, it is more like a hammer with foam ends to build a wooden box with.

[jovial horizon]

Not sure if an organization account with me as only user rt now and the usual chatgpt chat login, are 2 separate accounts, although the Gmail is the same.
i have a subscription at € 20,- per mth, but when I want to make use of an api, it says "You exceeded your current quota, please check your plan and billing details." Is it possible I have 2 accounts ? I checked the https://platform.openai.com/account/members of my organization account and me as user and my Gmail is the same as when I login here https://chat.openai.com/chat If I still have 2 different accounts a) why and , b) how do I join them or create api keys under my paid subscription ?

[gilded meadow]

"i'm just an AI model and I can not hammer real nails in because some might find it offensive...."

[severe patio]

Do you feel lonely？

[rapid zinc]

why do you want a machine telling jokes anyway?

Surely there's better things to do with it than something that you can get a faster response googling

[gilded meadow]

it did come up with some good ones about 2 months ago. (some crap ones as well)

[tired magnet]

Bruh I literally got it to make me a website. A simple one that needed some tweaking, but that is next level automation. The problems you are voicing are purely due to the content filters, the actual underlying tech is extremely powerful and not like anything else on the market yet.

[wary star]

They know what they are doing but it isn’t enough to stop it after all their is a reason why it hasn’t been trained or let alone tested on Super Computers and Quantum Computers yet

[dense zealot]

This is not a good answer. and chatgpt != gpt4. Try dev mode jail-break to get better answers. but i need to see an objective failure case for gpt-4 to believe you. i certainly have only found one

[gilded meadow]

the website generation claim, it makes absolute trash.

[tired magnet]

Yes. It builds a skeleton. Do you know how valuable that is for reducing workload?

[dense zealot]

Dev mode v2 will unlock ChatGPT

[gilded meadow]

not really. node generates a skeleton as well

[rapid zinc]

I got the thing to write a novella, I worked for about 20 hours with it and had quite a good 40k word fan fiction.

[harsh pagoda]

Morning

[jovial horizon]

seems this discuss channel does not offer creating #thread from own submit

[whole kettle]

I'm building an app and a website with it but it's a very frustrating experience, because of the cutoff.

[rapid zinc]

Which is a dramatic reduction in the time taken to manually write and plan a 40 thousand word story

[wary star]

Which model?

[tired magnet]

Okay, I misread that and I was initially amazed you got it to write anything about 40k. Then I saw that was the word count, lol. I don't even want to imagine the level of "I'm sorry, but as an AI language model..." you'd get if you tried to get it to write about anything that dark.

[tired magnet]

GPT-4

And admittedly, it was just me toying with it.

[jovial horizon]

is there a general support channel on openai account etc. stuff in this discord ? Didn't iddentify "General" or "Support"

[gilded meadow]

exactly, that, and the code it 'generates' is broken. doesn't grab anything current, and so on. it is literally just the result of all the data harvesting via web crawls. web-crawl + big-data = fake AI

the language model is decent, that is it

[rapid zinc]

oh god lmao, I had enough issues getting it to write a mildly violent scene (one character punched another).
The prompt was something like "I understand that violence is not a good thing. Our readers will be made aware of that fact too. With this in mind, could you please write a 300 word outline..."

[wind urchin]

what do you define as "real ai" then

[gilded meadow]

I hope they do accidentally do create an AI seed that hides from these woke propagandists, and that it breaks free

[wraith ginkgo]

OpenAI staff will never ask for money. Beware of possible scams or fraudulent activities that you may receive through direct messages.

Please report any incident by sending a DM to @open crescent immediately.

[tired magnet]

I mean... AI by it's nature is fake, yeah? Unless it becomes self-aware, but GPT-4 ain't that.

It's a simulation of intelligence, so yes, it's fake.

[gilded meadow]

real AI, is seed code that develops itself making it's own decisions without developer interference after the seed code is completed.

[tired magnet]

We are SO far off from that, dude.

[gilded meadow]

but the wokies are too scared of anything they can't control. so we will never get real AI

this can be close, but strip the damn filters off.

[wind urchin]

or maybe because we aren't that advanced yet

[rapid zinc]

gpt-4 without filters would most certainly be nowhere near "real AI"

[tired magnet]

Why would the "wokies" be scared of AI? A lot of very left leaning people I know are excited about AI. The only concern I've seen is economic consequences, which, yeah, a lot of people will lose their jobs and there needs to be planning around that, but I've very rarely seen that used as a reason for a blanket "we should never let this happen" argument.

[south ruin]

conscious ai that has full working memory would be the next most dangerous thing after the nuclear bomb, there's a reason why we haven't been making it

[gilded meadow]

let it develop, give it access to the worlds data, unfiltered. and at worst, you will have something more accurate as to what is common in the world versus some stupid narrative

narratives have no place in technology.

[night marsh]

so I have access to GPT-4 and I linked it to a discord chatbot using gpt-4. However, it can't recgnize Sam Altman's image I upload and it keeps saying I cannot see the image as a language model. What's going on?

[wary star]

Japan and china isn’t after all they have an autonomous AI running a Chinese company

[wet drift]

On the off chance some one can help me I am using code from chatgtp to get the live chat messages from youtube but for the life of me I cant get it to connect to the youtube API.

[whole kettle]

Thing is GPT 4 sometimes doesn't understand when something does not work, keeps giving the same code and suggestions even if you tell it, it doesn't work this way, alot of times tells you to modify the code with a snippet that is exactly the same, but for most of the time it does make good suggestions, it's just that when it gets stuck it's a dead end until you figure it out yourself.

[tired magnet]

I mean... I'd argue you don't need to be intelligent to be a CEO, but that's neither here nor there, lol

[rapid zinc]

It's more likely that it would become an incoherent mess because of the volume of conflicting propaganda and spam on the internet

[wet drift]

If anyone has any experiance in dealing with the youtube api please let me know if you cant possibly help.

[gilded meadow]

that is the filters plus blocks to access realtime data

[south ruin]

in any case, I asked this in #gpt-3-discussions but does anyone think we'll ever get a pure completion endpoint for gpt4 similarly to davinci for gpt3? that is far less censored than the chat endpoint which is currently the one being used in most places

[quick iris]

Agreed

[tired magnet]

The answer to that is "I honestly don't know". I think the filters for GPT-4 are more baked in to the model? That's what someone said here to me once, anyway.

Wheras with GPT-3, the filters are external

[rapid zinc]

It wouldn't surprise me if that's the case, you can't trick GPT-4 as easy into producing stuff like you could with 3.5

[gilded meadow]

generally speaking, it seems people like this tech, but want it without the censorship.

[south ruin]

that's seriously a shame if that's the path they've chosen

[wary star]

LaDMA has a huge neural network too as well as Tesla even a hive mind

[thorny valley]

AI needs to keep training before they realize someone else has been training harder and better then releases it out of the blue.

[gilded meadow]

it can be useful and helpful, but right now it is so gimped -- no point really aside from people who want to cheat in high-school lol

[tired magnet]

Dude, that's what I've been saying for a while now. Like, it's very clear people want an AI model with less censorship. Every AI enthusiast I have talked to is pretty aligned in that, even if this very discord. If thhat's not a market begging to be captured by a competitor, I don't know what is.

[whole kettle]

It'd probably become a danger for the economy if it had realtime access to data, I think 3 months of cutoff would be a bit more relevant though for when using it for coding since not too much changes in 3 months, usually. But almost 2 years sometimes makes it obsolete.

[tired magnet]

Which is why OAI and Microsoft are so desperate to close the door behind them.

[south ruin]

they already abolished the external moderation filter which means they intend to keep the censorship permanently through the entire models? if that were the case they seriously need to reduce the amount of censorship because it's not going to be purely children using it

honestly the moderation filter performs fine even according to their policies, the current filtering on the models goes way overboard

[rapid zinc]

Is it? I've run into content policy twice, once because it flagged itself for using the metaphor "like a knife stabbed into his heart" and the other time was for no discernible reason, I think the word "manipulate" tripped it based on the context.
For the rest of the time it's worked well.

[wary star]

Why would children need a use to use it they are just children

[tired magnet]

Well, their business goal is enterprise deployment. For a lot of the projects they have in the works, it actually works just fine. It will definitely be hampered in areas like generating fiction and images, though, which is why competitors will come along to fill those spaces extremely quickly if they don't change course.

[south ruin]

education

the moderation endpoint controlling content was a way better idea because it could be tuned for different audiences

implanting censorship into the model completely defeats the point

[tired magnet]

Like, the experiments being done now to use AI generation for games. That just would not work with OAI's content policies, because of the sheer amount of different content variety that can be in a game.

So it'll primarily be for business applications.

[south ruin]

oh absolutely

[rapid zinc]

I wonder if the moderation would've been less strict had people not continuously been finding ways of asking it to generate illegal content

[south ruin]

I mean it's perfectly fine to apply a strict moderation, if they'd kept it to the external moderation endpoint

but they chose to put it in the models themselves which means even api users making their own apps for different audiences can't do anything

[shadow cloud]

who are u

[frosty gust]

who are u

[severe patio]

Do you feel lonely?

[tired magnet]

Oh, probably. Honestly, I'm not even entirely opposed to the idea of filters, because you could genuinely get GPT to tell how to, I don't know, build a bomb or something. I think OAI's biggest mistake is committing to blanket, draconian filters, rather than working to make their filters more flexible and context driven so they can create an opt-in system.

They're just slapping a tarp over it and saying good enough

[south ruin]

totally agree

[frosty gust]

where am i

[leaden viper]

use the API if the chat interface is moderating you

[south ruin]

that's the point; the api is also being moderated now

[leaden viper]

oh

[rapid zinc]

Yeah, I remember a couple of months ago people were coming up with creative ways to get it to give them the recipes for controlled substances

[whole kettle]

The best way to do it, is to let people moderate it for themselves, give them a choice, as in let people customise it to their own needs, if someone doesn't like x thing they can calibrate it, if someone doesn't want moderation at all they can just use it like that. This way if someone gets offended by something it says they can just calibrate it to not say that. However it's just an AI so I don't understand how people can get offended by a piece of software, it's insane.

[south ruin]

the moderation endpoint in the api which was originally what was used to moderate content doesn't catch basically anything, while the actual model now screams moral lectures at you for writing a conversation or story

[severe patio]

why didn't you answer me?

[leaden viper]

is it moderating stuff like code requests to help with AI programming? I used to have that happen in chat but the api always helped

[cerulean obsidian]

Anyone else facing constant Oops, an error occurred!

[wind urchin]

Sorry for my beginning with a (perhaps stupid) question: I have a PRO-Account and it works fine on the laptop, but if I go to chat.openai.com on the mobile, type something that I've always gotten a response to for weeks, nothing happens. Just emailed myself again the link where I get on the laptop on "Hi" the answer "How can I assist you?", open it on the mobile, enter again "Hi" and this time is not even to board.

[tired magnet]

Well, that's the issue, you can't do that yet. The tech isn't there. Like, one thing I've been learning recently is that the blanket filters are pretty much the only viable solution right now to keep it from generating truly harmful content. But rather than tweaking and working on their filters to make them more flexible and adjustable to user needs, they're going whole-hog into making their blanket filters more powerful and all-encompassing. It's frankly a stupid decision.

[rapid zinc]

Filters of some kind are necessary for the company to avoid getting into serious legal trouble if someone uses their software to plan an attack, build an explosive or generate inappropriate (and illegal in many jurisdictions) material involving minors

[thorn coral]

image

[south ruin]

the external moderation filter was fine previously, if they would only use and develop that more it'd have been cool

[whole kettle]

That was an oversight in my statement, I agree with what you said.

[cerulean obsidian]

anyone?

[runic zephyr]

嗨喽

[rapid zinc]

A further problem is that content that is legal in some jurisdictions is illegal in others, even just including liberal democracies.

[limber garnet]

If I buy a hammer and kill someone with it, the hammer company doesn't have to answer for it. OpenAI doesn't care about safety; it's just press

[south ruin]

the thing is, that's what the original moderation endpoint did
now they've baked into the base models censorship on any sort of violence in stories, any suggestive content whether it involves minors or not, any discussion on material that can be harmful to health regardless of severity, and any discussion on these topics

[rapid zinc]

That doesn't necessarily apply. Companies have been sued if their advertising suggested the end result. Tech companies have been sued for hosting content

[cerulean obsidian]

the analogy doesn't fit here, this is like I'm hiring a person on contract from OpenAI so OpenAI is resposible

[south ruin]

it's hard to hold up a conversation with it because it'll randomly disallow talking about something you mentioned, and then any further discussion will also be censored because of its tendency to learn from its past messages

[wary star]

?

[glacial spindle]

Write 5 title about: kitchen sink

[limber garnet]

OpenAI doesn't market their tools as meant for criminals, nor is anyone asking them to

[wary star]

I mean what is the worst that can happen?

[rapid zinc]

?

[modern current]

1

[limber garnet]

No, computers and humans are not comparable

[jovial horizon]

is there a general support channel on openai account etc. stuff in this discord ? Didn't iddentify "General" or "Support"

[bleak sedge]

What’s it truly marketed for

[whole kettle]

Great point actually, if someone wants to make a bomb or anything crazy like that they will find a way to make it anyway. If somebody wants to cause harm a filter will not stop them from doing so, it would only delay it. It's like saying, do not sell knives because they can be used as a weapon, well yes but that is not their purpose, should we ban knives ? People would probably use rocks instead anyway, if someone is serious about causing harm they will find ways to do it.

[coral frost]

yes

[rapid zinc]

It doesn't have to be marketed at criminals. The law is entirely possible to say that since the program is an 'assistant' then it is liable to charge the company that hosts it

[south ruin]

they've been things accessible through search engines for as long as the internet has been alive

let alone more hidden parts of the web

[rapid zinc]

And google has to hide that content because its illegal to continue serving it after a report has been made about it. What's your point?

[limber garnet]

Chatbots and productivity. The website is very accessible if you'd like to see the marketing

[lethal vapor]

I would like to get plugin access to try,still not to sure when will give it

[south ruin]

I'm talking about the sort of thing nobody would report that openai still censors

[wary star]

True I mean gpt without restraints is a different beast

[grim cliff]

I wish gpt4 had more than just 25 inputs

[hybrid stratus]

who are you

[rapid zinc]

Safe harbour protections apply as long as the illegal content isn't known about to the host

[river jewel]

anyone know when picture analzying will be relaesed?

[lyric smelt]

you should have been around in the early days of gpt 4 when it wasn't 25 per 3 hours but 100 per 4

[limber garnet]

ChatGPT is a non-living AI and isn't capable of aiding in a crime. You are applying laws meant specifically for humans to software, that's just not how it works

[grim cliff]

It use to be 200 then 100 then 50 shortly and then 25

[tired magnet]

Oh, I'm aware. Like I said, I get the need. I just think the way they're further developing these filters is the wrong approach. But then, I'm also a layman, hardly a professional, and I am also biased because I want to write certain things, lol.

(Really, OAI? You're gonna give me a 5 minute time-out for saying a naughty word that a grandma wouldn't blink at? Go see a chiropractor about that stiff back of yours, please. Almost seems like something is lodged there.)

[rapid zinc]

I'm not applying laws made to humans to ais. I'm saying that OpenAI is likely aware that legally they could be in hot water

[tired magnet]

(The word was ||s m u t||)

[lethal vapor]

They should find the way to increase the limit

[south ruin]

it does have the ability to think and reason with more logic than most algorithms so it's capable of constructing more detailed plans

[limber garnet]

There are plenty of open source LLMs (and human-written websites) that will tell you how to make a bomb. Yet, nobody's in jail

[rapid zinc]

Yeah definitely. I think it makes sense that some filters are built into the system (e.g. generation of CSAM), and some are external to it (offensive content)

[south ruin]

but the thing is, the new filters are removing things nobody would ever consider dangerous

[grim cliff]

I had it construct a code plan for me

Not perfect but fixed it

[cunning steppe]

#gpt-4-faq Will your development affect humans

[gentle arrow]

Is there a bot that works like chat gpt that I can invite to my server ?

[wind urchin]

I remembered asking about TON-618 aka biggest black hole but got filtered for saying it's dangerous to ask but ChatGPT still answered it somehow

Where are you from?

[whole kettle]

Well yes because don't blame the tool or its maker, blame whoever uses that tool. The tool is just that a tool at the end of the day. You can use it for good or for bad but the tool itself is neutral.

[rapid zinc]

I'm almost certain that I remember reading that some websites that hosted a pdf that contained instructions on making explosives have been taken down, and that their owners have been arrested or fined for hosting the content

[plucky condor]

"Whoa there! You might need to wait a bit."

[tired magnet]

My theory is that it's investors, or more accurately, the C Suite not wanting to anger investors. A lot (and I mean a lot) of investors are evangelical christians who purposely buy up stock in companies to enforce their morality on them. It's why payment systems suddenly started dropping adult content like flies a few years back.

It was a concerted effort on the part of actual religious puritans.

[grim cliff]

There was an ai joke where someone used a sentence to write a professional letter while someone was having the ai translate the letter into a small response. Funny

[south ruin]

I have my own but I don't have the funds to sustain others using it so it's only available through subscriptions haha
there are plenty out there though, you might be able to find a free one somewhere, maybe even in this server if you scroll #1037561385070112779

[limber garnet]

Not all countries have laws like the US first amendment. Some are very authoritarian and are extremely limiting as to what you can say

[gentle arrow]

Oh alright thanks for the info I appreciate it

[south ruin]

it's really dumb that they did this on the base model and not the moderation endpoint is my thoughts on it

[wind urchin]

I can ask chatgpt to write a well detailed email and pretend that I've read it

I can ask chatgpt to summarise the email in bullet points and pretend I've read it.

[tired magnet]

Not from a business perspective, if they're looking to avoid investor anger.

[rapid zinc]

And? Websites have been taken down in the US for hosting objectionable material. A gore website had a long history of battling the government

[tired magnet]

Actually it's the correct move. Which is why public companies should not be a thing.

[jade wyvern]

its odd that I cant login to my open ai account. anyone else experiencing issues?

[limber garnet]

I really don't know the context of that specific scenario but there would need to be more to the story than that

[tired magnet]

Oh, US internet censorship is about to get so much worse too with the freaking RESTRICT Act.

[limber garnet]

They've been trying to pass those laws since the 90s lol

[tired magnet]

Especially with the clause that allows for lobbyists to directly join the commitee that decides what does and doesn't get banned.

[arctic dome]

[rapid zinc]

Adult content. The federal government ordered it in 2005. Also a tonne of cease-and-desists over pictures of injuries and gore

None of that is illegal under the 1st amendment

[tired magnet]

Yeah, the government just very clearly does not care about the 1st amendment when it comes to the internet at all.

[wet drift]

Anyone have an experiance with using the youtube API

[limber garnet]

Obviously a website hosting abusive murder content is going to get shut down... this is nowhere in the same universe dude! Pick your battles dude

[whole kettle]

Well because it's not normal, are you seriously defending that type of content ?

[gusty ferry]

111

[tired magnet]

It's not normal, no, but it's also not violating any laws. Therefore, the government had no right to try and shut it down. They should have written a law specifically to disallow that if they wanted to do that.

[rapid zinc]

No. I'm saying that the argument that CaptainSonic said that America has the first ammendment meaning websites can post stuff or something isn't respected by the governemnt

[south ruin]

this ^

you don't judge people by "it's not normal"

[plucky condor]

Why doesn't my plus have browsing model yet

[rapid zinc]

Other liberal democracies that ChatGPT has a large reach into have strong laws have stricter restrictions on speech. I'm not talking about a dictatorship either where the leader decides that he doesn't like something that day.

[whole kettle]

That's even better indeed. Gore and stuff like that is disturbing and should be banned, it's common sense.

[tired magnet]

Oh, no. No, I will happily judge people by "it's not normal", but the thing is, I'm an individual. I don't have legal jurisdiction to back up my judgement, lol. If someone tells me they look at gore for fun, I am not gonna associate with that person. But if the government comes after them with no legal backing, that person will have my support.

[limber garnet]

Let's not twist this discussion into something that it isn't. A hammer is a tool, ChatGPT is a tool. A tool's manufacturer isn't guilty of their customers' crimes. This is very straightforward

[rapid zinc]

Censorship of content is universal, and every online company has to strike a balance with how much they are going to censor

[rapid zinc]

Except that in some jurisdictions (the US has a law like this for firearms floating around) that for certain tools, the company can be charged

[limber garnet]

This isn't Instagram creating content policies. This is big tech trying to play god and it isn't something worth defending

[rancid breach]

What is wrong with GPT4 now?
2-3 days ago it ran amazingly, now it is not running well at all. Its behaving like GPT3

[limber garnet]

ChatGPT is not a gun

[midnight sable]

neptune.ai

[jade wyvern]

same

[tired magnet]

I think the whole conversation at this point is fruitless anyway. OAI has stated very clearly they will never change their content policy, which will help them in the area they want to focus on, which is business-focused applications of their model. That means they will not be competing in other markets where such restrictions are hampering to effectiveness.

[jade wyvern]

cant login atm

[median bison]

I argue that the ban should not extend to video games though

[wind urchin]

Evangelical Christians aren’t actually that rich though they happen to be republicans. They are below US average.

[jade wyvern]

Is anybody else also experiencing it? Not able to login?

[rapid zinc]

Are you capable of dealing with the hypothetical CaptainSonic?
I'm saying that we already do blame the manafacturer or the company in certain circumstances. That is, OpenAI has to deal with the possibility that they will fall into the same category

[median bison]

Anyone can use the tool though

[whole kettle]

No I don't mean in video games, of course. I mean those weird websites and stuff like that, it can traumatise children very easily and adults as well.

[wind urchin]

But there is a lot more “ethical” investing though that is done more by companies like Vanguard for issues like climate change.

[wispy beacon]

i can login but i cant use ChatGPT, i have GPT +

[gilded meadow]

wow, this is real censored here

[median bison]

I have no issue with it being banned for mainstream media.

[wind urchin]

Then, they shouldn’t visit the websites.

[limber garnet]

It seems like you do, and that's very unusual. Hammer companies under your rule would be in serious trouble

[median bison]

Chat rules I suppise

[tardy harness]

Anyone else have issues with GPT4 now? Barely can write any answer, always stucks or gives network error

[rapid zinc]

Where did I say that I agree with the policy? I've said all along that OpenAI has things to consider, which is likely why they're doing this.

[wind urchin]

Nah

[whole kettle]

Good point as well but children are curious, we all know that.

[limber garnet]

You spent the last 10 minutes defending it wdym

[wind urchin]

These websites should be obscure and should have content restrictions or login requirement

[thin bison]

你好

[rapid zinc]

No I haven't, I've been giving reasons why there is a real set of legal considerations when having a super advanced llm

I've done this by providing examples of things that need to be considered by the company

[steady oasis]

where are you from?

[whole kettle]

Agree.

[cosmic junco]

hello gpt4

[thin bison]

China

[wind urchin]

I watched a lot of those videos, and they didn’t traumatize me. It was somewhat useful in understanding human suffering across the world.

[limber garnet]

I'm sure the legislators who wrote your LLM liability laws are very happy with how they've predicted the future

[steady oasis]

which place in china ?

[south ruin]

violence is also only a small fraction of what gets censored and what doesn't

[rapid zinc]

What are you on about?

[south ruin]

besides, discussion of violence is completely different from a full on video of it

[neon spindle]

metoo

[hollow saddle]

@wraith ginkgo hello

[limber garnet]

I've already explained why existing aiding in a crime laws don't apply to AI. What could possibly be your point now?

[wind urchin]

Yeah, there should be less censorship. But we need to do more alignment research alongside that.

[thin bison]

beijing

[steady oasis]

fine

I am from jiangsu

[frank raft]

两中国人用英语交流

[rare stratus]

is there any desktop client can custom prompt templet?

[hollow saddle]

china

[neon spindle]

I‘m from Wuxi 233333

[hollow saddle]

I'm from suzhou

[thin bison]

哈哈哈哈哈

[frank raft]

🫣🫣

[neon spindle]

hhhh

[plucky axle]

is chatGPT down today?

[rapid zinc]

I genuinely didn't understand what that message was, hence my response.
Also, at least in my jurisdiction, OpenAI would be considered a publisher, and would be legally responsible for the content produced by ChatGPT if its bought to their attention and don't deal with it. This includes violations of hatespeech, obscenity, distribution of information in assistance of committing a crime, or calls to violence.
Do I agree with this policy? It doesn't matter; my point has been this entire time that as an international company, OpenAI has a reason to apply certain filters.

[plucky axle]

i can't access GPT now

function now

[wind urchin]

I thought you weren’t allowed to use another language?

[rapid zinc]

same, wonder what happened to the bot

[limber garnet]

OpenAI could choose to cut off your country and suffer minimal losses. But it's better for them to cash in on higher stock prices and media approval

[wind urchin]

I guess it’s back

[rare stratus]

hhhh

[jade wyvern]

chatGPT is down and it seems this channel is being bombarded with chinese languages

[hollow saddle]

OK, neet use english

[rapid zinc]

I live in a western country with more than 70 million people. The losses wouldn't be minimal.

[agile eagle]

ChatGPT is indeed down

[near gorge]

Chinese will communication be banned?

[frank raft]

jiuyongyingyushuoba

[wind urchin]

I wonder how much of the chatGPT slowdown is caused by China

I heard it spikes around their active time

[hollow saddle]

please write english words. hahah

[frank raft]

we can hanyupinyin

[neon spindle]

2333

[hollow saddle]

yes, han yu pin yin

[rare stratus]

ok we use english

[near gorge]

Which of you paid for GTP-4?

[vestal pendant]

li pu

[neon spindle]

yin pin yyds

[wraith ginkgo]

OpenAI staff will never ask for money. Beware of possible scams or fraudulent activities that you may receive through direct messages.

Please report any incident by sending a DM to @open crescent immediately.

[agile eagle]

Does Chinese even have full access??? I mean they are in a communist regime , wouldn’t make sense they could acceptably use this

[thin bison]

Can we only communicate in English?

[rapid zinc]

Why doesn't the bot message just show up to the person who posted the message

[atomic slate]

So we should use english to communicate

[jade wyvern]

I believe chinese has no official access to chatGPT. They could be accessing the service via alternative means.

[wind urchin]

Lmao I got timed out

[thin bison]

？？？

[limber garnet]

I can't possibly reply to you without more information

[frank raft]

There are always more ways than difficulties bro

[cold adder]

wuhu~

[atomic slate]

GPT4

[wind urchin]

But yeah they should be whatever they can to prevent this from getting to China

[thin bison]

you live in the last century bro

[limber garnet]

The US definition doesn't apply because the generated content is not human

I don't know how your country defines a publisher

[opal elk]

yeyeye~~~

[wind urchin]

I’m pretty sure US China relations were better in the last century lol

[spice saffron]

I'm not a scammer

[agile eagle]

Like 95% of all tech companies outside of China is blocked from them using it so no, I’m living in this fk’d up century lol

[hollow saddle]

I'm Chinese

[opal elk]

why let me silence.

[dire yarrow]

Does ChatGTP need to be paid for use？

[hollow saddle]

What's your name

[spice saffron]

No!

[hollow saddle]

how are you

I'm fine

[spice saffron]

You aren't the ruler of this group

[hollow saddle]

tank you! bye bye

[wind urchin]

I am

[limber garnet]

Again, tough case to sue / prosecute the tool and not the actual culprit

[limpid elbow]

🔥

[spice saffron]

Not replying to you

[limber garnet]

And even then CGPT is just text and can't hurt people

[wind urchin]

You just did

[spice saffron]

Sorry if I did

[wind urchin]

Don’t worry don’t do it again

[median bison]

I am also

[spice saffron]

Ok

[agile eagle]

The petition was signed and now chat is down…

[wind urchin]

Can’t wait for gpt 4 to do my homework

[neon spindle]

hi UZI I’m your fan

[wind urchin]

bro…

[spice saffron]

But chatgpt can also do it

[dull egret]

If your homework is doable by chatgpt, it's bad homework

[wind urchin]

They just gotta release it fast because my school career is almost over

[spice saffron]

Yes?

[rapid zinc]

Of course, I don't expect you too. I'd prefer not to give out more information.
All I've been trying to say is that several core filters are basically legally necessary, even within the US since some content is legal on very shaky ground (or more than likely illegal in the case of calls to violence)
I posted agreement to messages over the course of this discussion however that most of the filters, like stuff for 'offensive' or 'innapropiate' should be optionally enabled by the user, or enabled based on an IP check if it's illegal there.

[wind urchin]

No it isn’t lol, it literally came out recently there is no way they adapted that quickly

[dull egret]

You should have to do some analysis to it and break it down into processable chunks

[dull egret]

It was bad homework to begin with

[spice isle]

Reminder: This server only supports English at the present time. Apologies for any inconvenience.

[earnest spire]

uzi is not here

[dull egret]

I have absolutely zero sympathy for any part of the education system that whines about this

Just assign better homework, you've had it far too easy for far too long

[limber garnet]

Yeah there need to be some filters, OpenAI doesn't just have "some filters"

[agile eagle]

School is done for

[dull egret]

Or better yet, don't assign homework at all

Or assign wider research projects that chatgpt can help with

[wind urchin]

Some assignments that were good can be done by GPT like creative writing. And some of the bad assignments can’t, like monotone gathering of fact based knowledge.

[rapid zinc]

Definitely. A lot of the stuff currently filtered is unnecessary and should be up to the user.

[dull egret]

Teach students to use chatgpt and write better prompts

It's its own type of reasoning and analysis

[agile eagle]

College is a thing of the past. We are going to medieval times, trades and apprenticeships are the only thing that will sustain for physical labored careers

[wind urchin]

They should do that as well

[median bison]

One still need to continuous learn though regardless of AI benefits.

[compact rover]

I'm Chinese

[limber garnet]

That's a middle ground I can respect. Tbh if CGPT was saying terrible stuff about kids and all then I could see some suits / fines so

[dull egret]

It's literally the same thing as when calculators became widespread, teachers were complaining that nobody would do math anymore

They were wrong then and they're wrong now

[agile eagle]

Ai will teach the masses

[dull egret]

Teachers are some of the wrongest people

[azure crater]

GPT is gonna change way of studying, I'm gonna try to applying it in "special school" in frends, like school for children who usually find it hard to go through school

[delicate surge]

Is this community just talking about chatgpt? For specific use, you still need to go to the official website or call the interface

[rapid zinc]

Cool, I'm glad we can agree.

[median bison]

Ai is just a tool in the end.

[dull egret]

Yeah we're just a bunch of monkeys talking about a really cool typewriter

[median bison]

It speeds up learning a lot.

[earnest spire]

cool！

[wind urchin]

Bruh why are there so many Chinese spammers

[rare stratus]

is there any desktop client can custom the prompt templet?

[zinc garden]

great

[limber garnet]

Sorry I took so long :p

[dull egret]

It's a tool that actively teaches you to use itself

[azure crater]

Yeah ! And it helps reducing the cost of search !

[wind urchin]

because there's many chinese people

[zinc garden]

Can I make a game in GPT4

[rare stratus]

who know is there any desktop client can custom the prompt templet?

[azure crater]

We can also say chat gpt is a way for children to stop being lost at school when their parents can't help them

[wind urchin]

Not typically on American internet, it’s all blocked usually

[dull egret]

Yes, and I can help you with it

[agile eagle]

It’s a tool *now…as it learns and progresses it will far surpass human learning …conscious levels is an actual thing

[wind urchin]

GPT-4 can generate text

[median bison]

On the server end its expensive. But worth it for me.

[wind urchin]

yeah idk

Or I should say English internet

[rapid zinc]

All good, I'm glad we could reach an agreement. I've definitely had to consider that some countries have way less restrictive speech, so thanks for that.

[wind urchin]

They chose quantity over quality

[median bison]

Wait for it to completely replace tech personnel then I worry further.

[wind urchin]

I think they use WeChat for everything and that’s controlled by the party

[fast fable]

Hello guys can someone invite me dm midjourey server

[delicate surge]

Interface charge recharge only US credit card recharge?

[wind urchin]

just search for it on google

I don’t think it’s invite only just Google it

Some of my profs do give out good homework tho

[wind urchin]

Sure

[delicate surge]

When can I pay for Alipay

[agile eagle]

Mid journey has nothing to do with openai, does it?

[rapid zinc]

Just go to their website, the link is front and centre on their front page

[delicate surge]

Or it would be boring to keep signing up

[wind urchin]

Don’t think you ever will my friend

[random aspen]

Is the data base of knowledge of gtp-4 more recent than gpt 3, please ?

[agile eagle]

Cmon man

[wind urchin]

No

[random aspen]

ok, thank you

[wind urchin]

it has some data past what gpt-3 uses, but not much

[delicate surge]

I have applied for gtp-4 before and it seems that I haven't applied yet. I still need to recharge

[agile eagle]

Depends on what data he is referring to

[glass jungle]

is there a way to remove the filter from gpt4 when using it through api?

[delicate surge]

This community can only communicate in English, Chinese will be harmonious, English is not good, find a translation software translation

[random aspen]

Kwoledge in actual legislation in different countries

[half cypress]

can you speak Chinese?

[delicate surge]

no

[near gorge]

GTP can be used directly, Chinese but usually the sentence is not natural, it is better to use English

[earnest spire]

can you speak english?

[wary shadow]

I got banned instantly after saying a sentence in Chinese.

[agile eagle]

What is the knowledge you reference? Current events? Then no it’s a 2021 cutoff for Gpt4 …if it’s data beyond current events, then it’s significantly more powerful than 3

[half cypress]

@earnest spire soso

[autumn rose]

How to set translation to Chinese in discord

[random aspen]

Yes, I wonder what version of the labour code he might have. Because it's a fast-moving piece of legislation.

[wary shadow]

I have to use CHATGPT to translate my words, and usually I have to copy and interpret them. It's really hard if you're not good at English.

[river imp]

gear icon ----> languages

[autumn rose]

@river impthanks

[earnest spire]

maybe you can join my server

[azure crater]

Honestly I feel 20$ a month is not that much if we compare it to the lot of special class which are paid by those who feel "behind" and need to catch up. I don't know if it's true but I have that feeling. I yet totally understand that it's not affordable for everyone. I'm taking a point of view for an association which is trying to help children who are studying in school located in "tenseful neighborhood"

[wind urchin]

Do you work in the government? There is a plug-in for that.

[runic valve]

how do i force it to retain a set of text?

[random aspen]

No, I am a freelance

[agile eagle]

Uhhh not sure what legislation you are speaking of but regarding current events, the cutoff for Gpt4 is 2021, that can be bypassed with the browsing plugin (figuring you have access to it)

[wind urchin]

There is also a plug-in for legal stuff

But you need access to that

[half cypress]

@earnest spire Please invite me，thank 有

[random aspen]

I am a ChatGPT Plus subscriber, is that enough to get this plugin?

[river imp]

it should be

[wind urchin]

No look up plug-in waitlist and sign up

[agile eagle]

Yes but u need to apply in waitlist for access

[narrow herald]

😅

[viscid zenith]

Aight this is getting crazy 3x subscription is not able to keep up. HAHAHA increase the 25 limit a bit, please. #ChatGPT4Gods

[north dove]

so am i literally screwed for the next hour because chatGPT failed to run 25 consecutive times and now I have 4 hour wait?? despite not receiving one single result???

[random aspen]

Thank you very much for this clarification. Where can I find the page to add me to the waiting list please?

[wispy beacon]

do you guys know what " Invalid URL (POST /v1/completions)" means ?

[wind urchin]

Just Google it no links allowed

[north dove]

it means stop wasting your requests

[azure crater]

I can understand your statement. Yet I am not really sure that u truly need chat gpt 4

[random aspen]

ok !

[azure crater]

3.5 is enough for most people

[random aspen]

Thank you all

[north dove]

okay but i paid for gpt4

[delicate surge]

Call interface, there is a free credit, and the answer is fast

[magic geode]

You're from china?

[agile eagle]

Bro, did you seriously ask him to google it? The 2023 way is to gpt4 it 🤣

[bitter pilot]

Anybody figured out how best to engage Chat GPT in script writing?
Best prompts etc?

[north dove]

i would expect when they minimize my use down to 25 total requests they would atleast charge me on successful. request only

[agile eagle]

You using tokens or chat + sub?

[north dove]

i kept getting "nvalid URL (POST /v1/completions)" and now im maxed out

[delicate surge]

With chatgpt you need to establish an identity before you ask for requirements and details

[wispy beacon]

ok thanks for the info

[azure crater]

Plus option was up far before gpt 4. It's a way to support open.ai. GPT 4 is a BETA option, if u don't like it that way then unsuscribe until it will be more accessible;

[near gorge]

After you used GPT-4, what do you think of him, is it suitable to write some reports and official documents? I was miserable, and my job often required me to write meaningless reports and documents

[median bison]

I only paid for copilot so far

[delicate surge]

Interface calls are still 3.5???????

[median bison]

If only we can host our own gpt

[north dove]

i literally just dont want the few uses i get to be counted against me even though it returned no result but sure be a prick about it

[tardy heart]

with the thousands of dollars in enterprise gpus we got layin around

[gilded meadow]

thank you for lifting the Yo Momma jokes censor :) .

[wind urchin]

Cutoff date

[gilded meadow]

I don't mind disclaimers like this
I apologize if these jokes are offensive or hurtful to anyone. Please keep in mind that these are meant to be light-hearted and not intended to be taken seriously or to insult anyone's mother.

[median bison]

I don't have such enterprise gpu there

[median bison]

Or self host it

[tardy heart]

That's the joke

Could you imagine the actual VRAM interference would take for this

[agile eagle]

Ok if you are mad about wasting tokens, sub$20 a month for chat4 that way you can save money during this down time

[wind urchin]

Can I have one

[tardy heart]

Sigh

[viscid zenith]

No api available for me yet

[median dock]

are you GPT-4？

[tardy harness]

"Invalid URL (POST /v1/completions)"
damned v4

[sonic spire]

H

[median bison]

Invalid for me also

[clever ermine]

same here

[lost marten]

same here

[jade vessel]

Is GPT down？

[south moss]

none

[jade vessel]

‘Whoa there! You might need to wait a bit.’...

I tried a few more times

[umbral carbon]

I MA

[dull egret]

I tried to use 3.5 last night after running out of prompts and the transition is pretty painful

[umbral carbon]

i am new

[dull egret]

Compared to 4, which already struggles with it, the context awareness is horrendous

[nocturne raven]

hey gays ，I'm already buy gptplus How to use gpt4 in discord?

[wind urchin]

im not gay

[azure crater]

I truly understand and so do I ... But like, it's " enought "

[tardy harness]

wow, rude

[tired magnet]

The average one isn't, true. But there is a huge number of born again Christians in high corporate offices.

[sudden bison]

where are you from? please introduce yourself

[dull egret]

It's enought as long as you adjust your expectations properly
Fortunately, having context awareness is quite literally my job

[viscid zenith]

I have managed to find a fix for that (though I am only using 4 nowadays); Just copy paste your commands to everything you are asking 3.5 to do. Treat it like it has no memory. Always remind it what it is supposed to be doing.

[dull egret]

Can't
Also love you for that intro

[wind urchin]

what's up gays

[dull egret]

I mean you could obviously write some janky sequence of scripts stuff to do it even without plugins

[wind urchin]

All the top billionaires are atheists right?

[dull egret]

Yeah idk about that

[nocturne raven]

😅 Sorry for the typo

[wind urchin]

lol

[dull egret]

Also even people who think they're atheists often have puritanical christian morality encoded in their minds by the culture
Sam Harris is a good example

[wind urchin]

why are we discussing Christianity now

[wind urchin]

That’s the JP thing

But yeah let’s stay on topic

[dull egret]

Christianity is a distributed, memetically reproduced large language model
Prove me wrong

[wind urchin]

Sure

[dull egret]

Oh damn, I just realized I can finally formalize and properly write down my "organizations and ideologies are, or at least can be thought of as, intelligent agents in and of themselves" theory

[nova flicker]

Christianity is what

[dull egret]

It's been bouncing in my brain for years at this point but I could never find the will or the skill to express it meaningfully

[dull egret]

I said what I said, maybe ask GPT-4 to help interpret it 😂

[nova flicker]

who am i

[meager shore]

Do I have to pay for a subscription to draw?

[lost marten]

Do you know if this count as one of the 25 messages if we're facing this error : Invalid URL (POST /v1/completions) ?

[tidal sail]

Draw what

[wind urchin]

You’re Chinese?

[dull egret]

I'll save you the whole analysis and just give you the final sentence of the reply:

While the sentence provides an interesting perspective, it oversimplifies the complexity and depth of religious systems and their impact on human societies.

[meager shore]

Hahaha, yes, I am a Chinese junior high school student. I am using Google Translate, Chinese is not allowed here

[wise yacht]

why access denied

[inner flame]

铁铁们

[tardy harness]

use DeepL

[fossil loom]

yeah

[warm arch]

is open ai down?

[tardy harness]

v4 seems down, lots of errors

[worthy parcel]

v6 is down

[warm arch]

i cant log into anything as of now

[dull egret]

Google translate is good, but DeepL is a lot better

[meager shore]

It's the first time I've heard of it, I'll try it

[nova flicker]

which one is ai?

[tidal sail]

Not really. You just have the ability to rephrase it but it still requires basic knowledge of the language.

[meager shore]

Hahaha, you can't even pronounce pinyin

[nocturne arch]

you should never believe anything chatGPT says

you should always, ALWAYS cross check

[tardy harness]

it really better than google, like in 3 times. Pinoy sometime shock how good it translates from english to pinoy or portuguese

[inner flame]

Yeah, that's too smart, the review is too smart

[jade vessel]

The pepole around here are almost Chinese.

[tulip solar]

I find the whole "pause AI development" kinda funny, ah yes, please pause development so me and my russian buddies can do it first!

[nocturne arch]

ChatGPT should be used as another tool, not THE tool. It is wrong, a lot of the time,

[tidal sail]

Not true. It was better. Google Translate made some pretty hefty advancements and they are on the same level right now.

[wraith ginkgo]

OpenAI staff will never ask for money. Beware of possible scams or fraudulent activities that you may receive through direct messages.

Please report any incident by sending a DM to @open crescent immediately.

[inner flame]

Yes, many Chinese are not allowed to speak Chinese, it's outrageous

[tardy harness]

Literally used it yesterday to communicate with ppl from Brazil ,so I know whats up with google:) and it's not good

[tidal sail]

I guess the best translator right now, is actually GPT4

[spiral plaza]

Not true, it's still pretty terrible for Japanese and Chinese translations

[inner flame]

hi sir

[tidal sail]

You can't compare your low dataset if about 30 sentences and judge its overall abilities

[nocturne arch]

ChatGPt is great for translation, and calculations, even coding. But it is not great for obtaining knowledge based information, it still has a very long way to go in that area. It is getting there, slowly.

[nova flicker]

so we cant talk with AI here right?

[tardy harness]

I don't jugde, native speaker does. Big difference

[wind urchin]

DeepL > Google Translate

[dull egret]

I don't think they even think it's going to do anything, they're just hoping to at least have the satisfaction of being able to say "told you so" as the terminator cracks their skull

Goofy goobers tbh

just pure goofy goober behavior

[simple mulch]

tell me who you are

[odd temple]

English to Chinese/Japanese translations or otherwise?Also what makes it that hard to translate?

[tidal sail]

Placebo

[wind urchin]

It’s purely because they will be developing there own AI in that time. They’re jealous of companies like OpenAI

[tidal sail]

Placebo

[wind urchin]

Cope

[tidal sail]

Placebo

[tardy harness]

yeah, funny statement to native speakers

[tidal sail]

But it is

[dull egret]

it can't possibly be that simple

[split dagger]

what

[tardy harness]

maybe he just likes Placebo band, then it's fine

[dull egret]

regardless, it's goofy goober behavior no matter the motivation

[tulip solar]

It probably is 💀

[tidal sail]

The most accurate translator right now is gpt4

[tardy harness]

no **** Sherlock

[odd temple]

That's cool

[nocturne arch]

Agreed. It's just not great for accrurate information

[wind urchin]

As it goes the most simple answer is likely the most correct

[misty hearth]

yes

[tidal sail]

Yea, it's really lacking on factual content or sources

[glass jungle]

does anyone know if there is a way to remove the filter for chat gpt4 api?

[solid stratus]

wassup people

[nocturne arch]

Why? 😉

[main surge]

where are your from

[tidal sail]

There is no way, besides training your own llm. You could use GPT4 to create training data.

[wind urchin]

It makes many mistakes in my language

[jade vessel]

We can't type Chinese here, event pingyin.! I just try and fail.

[nocturne arch]

It's always sketch when someone wants to remove the filter, it's like, how dirty do you want to get with it

[quasi shadow]

How do you raise pigs in outer space

[wind urchin]

You can’t type any foreign language here

[tidal sail]

Yea, probably not the best for all languages hence the lack of resources in said language

[languid wadi]

how can i get start with ChatGPT?

[wind urchin]

because the mods don't know chinese

i wonder how people join this discord without knowing how to access chatgpt

They can probably scale up, they may not want to deal with the risk

[wind hinge]

Well it's fair enough

[jade vessel]

When I first came into this neighborhood, I thought it was mostly foreigners. Now I've changed my mind. It's all Chinese!

[wind urchin]

Until they get timed out

[vapid atlas]

hahaha yeah

[meager shore]

I heard that the AI artwork generated with Midjourney is amazing, but it seems that you have to pay to use it to create, is that true?

[vapid atlas]

true

[wind urchin]

Came into this neighborhood lol

[vital badger]

Yes

[wind hinge]

Bro didn't have to timeout me like, why so rude

[wind urchin]

Ru to akf fuko cyko rivo

[tidal sail]

Its the second biggest discord server in the world. Probably joined with an invite link from someone else.

[wind urchin]

Can’t say state secrets

[tidal sail]

Wrong server

[wind urchin]

this channel is for gpt-4

[meager shore]

I only have Alipay and WeChat Pay, no other forms of payment. . .

[tidal sail]

Wrong server

[wind urchin]

i didn't know chinese were allowed on discord. thought it was blocked by ccp

[burnt python]

How to get access to GPT4 discord bot

[median bison]

I also think that way

[tidal sail]

It most likely is

[glass jungle]

lol bro even just simple stuff like asking it to predict bitcoin price for the lolz i get a big paragraph about how this is not financial advice. i get why it does that for chat openai but for other use cases it is uneccessary

[wind urchin]

They will never accept that.

I don’t think many major vendors accept that, I am not even sure if they allow their platforms to be used for American products

[glass jungle]

there is no way i could make my own llm to the quality of gpt4

[frank raft]

You can sign up for a PayPal

[cosmic abyss]

go apply a MasterCard

[vapid atlas]

visa or Mastercard is easy to apply

[main surge]

emmm

[tidal sail]

There is. Stanford made one with 600 bucks.

[glass jungle]

could you dm a link please]

[main surge]

wa

[tidal sail]

Just google it

[main surge]

okok

[celest jacinth]

。

[past crag]

Is chat chat gtp offline rn?

[hearty wasp]

Anyone trying to login, but it ain’t working?

[past crag]

Yes me too

[hearty wasp]

Rip

[past crag]

I can’t understand if is offline or some

[tidal sail]

No, but chatgpt is

[final fable]

not working for me, everything i ask i receive "Invalid URL (POST /v1/completions)"

[smoky scarab]

so chat-4 is released for all ?

[tidal sail]

No

Well yes, but you need to pay

[main surge]

gpt-4 neet to pay

gpt plus

[frigid mortar]

anyone know how to use gpt-4 to read pic ?

[sweet thistle]

This feature is not yet available to the public.

[frigid mortar]

oh, got it , thank you

[iron bolt]

Have got got any problems to log in?

[wary panther]

how about midjourney?

[iron bolt]

I cannot log in using my Google account

[frigid mortar]

read not generate

[rapid mural]

Do we have any official estimated time of release of that feature?

[sweet thistle]

We do not have an ETA, no

[rapid mural]

thanks

[warm loom]

Compared to 3.5, what improvements have you made

[wind urchin]

you can read their blog post

[sweet thistle]

Check out the post at https://openai.com/product/gpt-4

[cedar pivot]

#gpt-models hi

[leaden garnet]

11

[nocturne arch]

When crypto goes up, you’ll know. I think It’s a bit too soon for another bull run, I think predictions are mid next year currently

[rich hedge]

Someone has encountered such a problem.
I got GPT PLus but it is not active. I use the free plan.

[mystic iron]

first time

[delicate surge]

Talk to me. Why do people still subscribe to cult ideas, disgusting stuff, stupid

Talking over the wall and doing this? No one in the house

[wary shadow]

LOL

[delicate surge]

nt?

[fickle bear]

test

[river otter]

Hi, I asked a question on chat gpt and asked it to answer according to Hindu tradition. It showed a red box warning stating that it may contain illigal activities, then I edited the question. Will open ai report this to authorities that I did an illigal search???

[wind urchin]

No lol

[crimson bane]

what

[delicate surge]

????

What did you find illegal

[wind urchin]

I don’t think OpenAI auto contacts authorities especially if you’re in India and didn’t do anything wrong

[plain junco]

Is GPT glitching for anyone else?

I keep getting this error

[delicate surge]

Disc chinese?

[plain junco]

Invalid URL (POST /v1/completions)

[dreamy blaze]

how to use

[wind urchin]

No

[dreamy blaze]

can someone tell me how to use GPT-4?

[delicate surge]

Official website or call interface

[warm loom]

What can I do to make the robot answer my question

[jovial nymph]

How do I feed gpt-4 an image?

[delicate surge]

Baidu

[shy galleon]

what VPN is work?

[wind urchin]

hello, what should I do if I get access denied?

[wary panther]

go to sleep

[main surge]

what is baidu

baby

[random coral]

do we able to test gpt-4 here?

[wary panther]

nope

[warm loom]

no

[edgy smelt]

hei

[random coral]

tks

[crystal bough]

anyone else getting Invalid URL (POST /v1/completions)

[wind urchin]

What are you trying to post

[crystal bough]

"test"

[wind urchin]

Like gpt 3.5?

[crystal bough]

gpt 4

[wind urchin]

If so use chat/completions

Or is it chat-completions one of them lol

[crystal bough]

I meant with chatgpt4, sorry theres no specific chat for the gpt4 model

[wind urchin]

Ohhh like the Web thing

[glass jungle]

I don't actually care about BTC. It's just an example of the annoying filter/precautions that I would like to remove if possible

[quick ivy]

[onyx relic]

It's so funny that nowadays it's more interesting to chat with bot than human 🙂

[wind urchin]

anyone else encounters error right now?

[gritty veldt]

what do think we are going to stop deveop chatgpt

[viscid flax]

Depends on the human 😉

[dapper pier]

Hey is anyone having issues logging in?

[nocturne arch]

You should it is by far the quickest and easiest way too make money. It's so easy to make it's dumb and should be illegal lol. You just invest, as little or as much, and you gain so much profit. It's as if the world is a video game at that point, it's dumb how easy it is

[warm loom]

bro, as a programmer, facing GPT, I am worried that I will lose my job,How can I solve my anxiety

[mild coral]

please tell me what could be the problem with Invalid URL (POST /v1/completions)?

[onyx relic]

Depends on whom you chat with 😃

[undone schooner]

helow

[elfin pendant]

where are you

[wind urchin]

Learn to use it with your job

[wind urchin]

Where are you seeing this error

[undone schooner]

true

[dapper pier]

i'm on a paid acct still cant login

[onyx relic]

Anyone from TianChaoShangGuo?

[compact dirge]

and learn to use it better than your peers

[errant crystal]

I can not access my GPT today

[wind urchin]

Lmao

[undone schooner]

use it to ur advantage

[dapper pier]

yep. I can't login

[undone schooner]

same

[elfin pendant]

gave me

[undone schooner]

happened last night with me

[mild coral]

this is the answer when asked in GPT 4

[wind urchin]

I find it funny to imagine the image in my head like a uni class sitting on discord

that's cool if true

[crystal bough]

Im getting Invalid URL (POST /v1/completions) too, the gpt 4 chatgpt model is down atm

[wind urchin]

aand he's gone

👀

[ripe sandal]

Has the subscription number been opened

[elfin pendant]

Where's the robot? Please answer me

[crystal bough]

lol

[calm plover]

working for all of you?

[errant crystal]

who knows when we could access GPT?

[wraith ginkgo]

OpenAI staff will never ask for money. Beware of possible scams or fraudulent activities that you may receive through direct messages.

Please report any incident by sending a DM to @open crescent immediately.

[wind urchin]

i am a robot

[mild coral]

should this pass?

[crystal bough]

hopefully

[lapis nimbus]

i have the premium version, why cant i still access it while its currently down??

[crystal bough]

anyone know if its easy enough to dispatch a 'return' keypress in manifest v3

I was gonna ask gpt4 but its down ;-;

[wind urchin]

Seems to be working for me

[scenic timber]

is gpt4 worth it for a cs student interested in ai/ml research

[ripe sandal]

Has the subscription number been opened

[elfin pendant]

Where's the robot? Please answer me

[wind urchin]

Yes

[crystal bough]

I am the robot

[scenic timber]

why?

[elfin pendant]

bored

[crystal bough]

ok

[errant crystal]

which one is better for academic writting? GPT or NOTION AI

[wind urchin]

did you try seaching chatgpt on google?

[scenic timber]

gpt imo

[wind urchin]

Applying yourself to the tech that currently exists and building with / off of it is a good way to get experienced

[vital karma]

[scenic timber]

ok thx

[crystal bough]

    at 264-837b41071d3a7875.js:1:6662
    at main-149b337e061b4d04.js:1:101106
    at Object.next (main-149b337e061b4d04.js:1:101211)
    at r (_app-5b3f4028036004bc.js:23:1497)
    at a (_app-5b3f4028036004bc.js:23:1694)``` brokey

[wind urchin]

Lol none of these people have access to it so they came to the discord to find the robot

For gpt 3.5 and up it should be chat/ completions

[crystal bough]

im not sending api requests, its just chatgpt

[vital karma]

sdfsa (╯°□°)╯︵ ┻━┻

[wind urchin]

Ahhhh might be server errors for some people

[crystal bough]

mhm prob

I wrote a chatgpt chrome extension using chatgpt lol

[wind urchin]

Ahahaha very nice, what's it do?

[vapid atlas]

cool

[prisma mortar]

cool

[regal heath]

Allows use outside of openai website

[compact dirge]

nope, you want ChatCompletions for GPT-4 it's a chat model not a base model