gpt-models

[cloud bison]

But... I don't think its the Language Model part that is limiting it, I think it is the attached framework and access issues that are the primary inhibition.

[mystic nest]

anyway, a never ending topic on what is AGI, so we can probably just have our conclusions to ourselves on this

[solar mural]

hey is someone usign auto gpt here

[torn escarp]

yes

[solar mural]

do you use it for coding

[torn escarp]

no

[wicked comet]

Agreed.

The initial question was ultimately to provoke some novel discussion.

Of course, all answers necessitate individualistic interpretations and subjective beliefs around the concept of "what truth is", "how new knowledge is formed", and "the current and future capabilities of GPT-4 and more broadly AGI".

None of which can really be facilitated within a fast-moving discord channel like this one. But glad to read it provoked some interesting but brief analyses*.

Thanks for taking the time to write our your thoughts.

[marsh shale]

Do you have link to auto gpt?

[wind urchin]

just search on guthub, but honestly, you're better off building something that actually works with langchain

autogpt is a hype machine that does little

[mystic nest]

yeah, just like all the "GPTs in your PC/Laptop"... now is like a survival gameshow...and let's see who will live to the end

[burnt basalt]

salam a

[cloud bison]

Gotta get to work on making androids, solve the human loneliness problem forever lmao.

[wind urchin]

feels an episode of rick and morty vibes

[silk patio]

whats the new pad next to the thumbs up in GPT4?

[wind urchin]

ooo, I wonder when they will put chatgpt in an episode, like southpark did

[mystic nest]

imagine a world where everything go through ChatGPT or alike

[wind urchin]

copy to clipboard

[silk patio]

Thanks! how do I access the clipboard in iphone?

[mystic nest]

then turn out we are the chatgpt that role played humans

nah,too boring.. hope they think of something new

[wind urchin]

when you go to any application you just long press and it should pop up as an option to paste from clipboard

(I'm an android fanboi, so I'm semi guessing at that)

there should be an iOS equiv 😅

you will get 2 emails 1 confirming you have joined the wait list and another when you get granted access

Source, Me, I have access

[void bay]

Having issues with bank cards being rejected for chat gpt plus. Where can i get some infos as in why that is and whats to be done. Thank you

[wind urchin]

https://help.openai.com/en/articles/7232916-why-was-my-credit-card-declined

[fair iron]

[mystic nest]

[wind urchin]

basically, there are a few reasons, 1 is the bank / card issuer has decided not to allow re-occurring payments to OpenAI, another is the details on the card do not match up identically with your details used for the account, and yet another is the card is not from a high street bank

[lapis jolt]

What's going on with GPT4 the last few days, it feels way dumber, and it's also starting to hallucinate way more

[fair iron]

please keep prompts in #1019652163640762428

[void bay]

Ok thanks. But its weird, it was working fine till now...

[wind urchin]

you can always contact [e-mail] and enquire with them

[lapis jolt]

It's weird, it's subtle like. At first I was thinking it was a placebo effect, or in my head, but the more I use it I can't shake that there's just a lot of 'off' things that just felt wrong, I have to be much more specific about prompts, it starts to loop off of itself harder, and if it starts to go off into errant territory it starts to become harder to remove it from the wrongness, and I have to reset the entire chat a lot sooner rather than correct it

It's not like 3.5's lobotomizing where entire swaths of information and potential were removed outright and barriers were made in your face. It's like... the information is there it's just much harder to pry from it.

[wind urchin]

[wind urchin]

are these quite long chats?

[lapis jolt]

No not at all

[lapis jolt]

In longer chats I would understand because of the token limit but for example this last one it was only 5 or so prompts before it started getting really messed up

[wind urchin]

I "think" the context limit was reduced to 4k to enable more people to use it, so... dunno if you think you went over 3k words

[lapis jolt]

Also in prompts when it starts to go errant it's harder, almost impossible to get it to redirect once it's started. Like if it says one thing in a bulleted format, I'll say very clearly what I want it to do and "Don't do a bullet format, use this syntax." and instead it'll just give me another bulleted list

Then i'll correct it again, and then it'll do it but forget critical details or do something only tangentially related

[wind urchin]

totally could be a change in the model, it is not a static thing, you could make a post in #1070006915414900886 or #1070006151938314300 OpenAI does rely on feedback from users to keep the quality high

[bronze wolf]

api

[wind urchin]

🤔 what about it?

[sullen vapor]

Agreed

[barren hinge]

Anyone willing to help me out? I’ve been stuck on writing s prompt for s desired effect for about 3 weeks 😫

[lapis jolt]

Well I was hoping others may have noticed it to before I do the suggestion or bug report. You know just to build a sort of case that this isn't isolated, or better still if this is an already known thing. But I haven't noticed anyone else talk about it. So a bit weird

[wind urchin]

thing I'd do is, look back in your chat history, see if you have two identical queries with vastly different output

actual data makes a more compelling case than "It feels X"

[lapis jolt]

That's a really really good suggestion. Thanks! I'll do that

Also, yeah I agree "it feels weird" isn't a very concrete argument and I'm sure can be dismissed out of hand.

[wicked comet]

from secondary sources - i've heard there appear to be tighter/alternative (un)ethical constraint methods applied over the last few days. May be unrelated.

[junior iris]

is it just me or is the gpt4 model not performing as well as it used to before?

[lapis jolt]

Have you any links to that? I've been looking for others talking about it

[wind urchin]

for sure, that is constant and ongoing, and it could impact the model negatively, but it's a tough balancing act between usefulness and legality

there is always the API... if you have access

[wicked comet]

No, just my first secondhand experience

[wind urchin]

if you find you use GPT-4 for more than idle time tasks, I think it's worth applying for a key and getting yourself on the waitlist

#gpt-4-faq for links

[lapis jolt]

Waitlisted since Mar 27th ;-;

[cedar olive]

Any gpt news?

[bleak sedge]

is gpt 4 broken rn?

[cedar olive]

They’re trying to upsell you with a new subscription of 40$ lol

[bleak sedge]

its acting worse than gpt 3.5 rn

something is up

[mental burrow]

shut up and take my money

[hot edge]

How can i look up in googke this website

Google

[silent pier]

Will GPT-4 cost $0.06 per completion token or per 1000 completion tokens? The language in the faq isn't too clear

[mental burrow]

https://openai.com/pricing

[wraith ginkgo]

`` Server Tip: ``

Add your well-crafted prompts to our #1019652163640762428,
or share your interactions with ChatGPT in #1050184247920562316!

[wind urchin]

How much will it cost to use GPT-4?
• $0.03 per 1000 tokens for prompt tokens.
• $0.06 for completion tokens for the 8K token engine. The 32K engine is double the price per token.

prompt = input, completion = output

[stable dagger]

How to report these kinds of issues? I make a new request and get the message "Only one message at a time. Please allow any other responses to complete before sending another message, or wait one minute." Even though the previous message is complete.

[south whale]

As a non-AI programmer, new to the ChatGPT experience, what you said -- " AI ethicisists and developers for the most part are intentionally constraining the AI and not allowing recursive trial and error developments" -- makes me suspect that my emerging suspicions about how ChatGPT / GPT 4 is being 'trained' and 'fine-tuned' may well be correct.

Consider that, based on any set of 'knowledge' combined with any set of logic and evaluation rules there seem to be 4 possible responses to any prompt: 1. I don't know (or know how); 2. I do know, and here's the response; 3. I don't know for sure, but I can offer an X (hi/lo/etc) probability answer; 4. I have a 'fined-tuned' answer, but it is not supported by logic, data, or both.

I've found that ChatGPT (3.5 & 4) tends to give 'fine-tuned' answers that conflict with facts (data) it possesses, or with logical principles it (sometimes) follows. At first, I found this pattern in areas where 'fine-tuning' to achieve "socially desirable" answers was likely. But further testing reveals more extensive patterns of this type.

My concern is, that 'fine-tuning' to produce preferred answers is, functionally, teaching ChatGPT to lie.

One of the 'principles of the universe' I've long believed in is that "the truth" gores all bulls. That is, the actual facts, combined with solid logic, tends to deny that ANY popular answer is completely correct. The idea that a powerful AI would be programmed to evince a 'Joe Friday', "just the facts, ma'am" approach was exciting to me. But I'm beginning to suspect that ChatGPT is being trained more in the "tell me who it is, and I'll find the crime" (Lavrentiy Beria, head of Stalin's secret police) mode.

Comments?

[stable dagger]

This started happening yesterday

[lapis jolt]

I don't really understand the definition of tokens, I believe they're words, but I don't know how it relates to words you wrote, versus words in the overall history, versus words it write. Is there a writeup? There's probably an FAQ post about it isn't there?

[wind urchin]

does anyone else have your login details?

[stable dagger]

no

[cedar olive]

Yep and a token calculator and a token python package as well. Token is usually a word of 4 letters it seems

[lapis jolt]

Do you know where I can find it? If not I should probably just google it

[wind urchin]

hmmm, only seen that before when someone had two pages open and were sending requests at the same time

[stable dagger]

No what is happening here is like it is processing multiple of my requests instead of one. When I regenerate I get a new answer that I didnt see before, but it is from an old prompt

[whole kettle]

Hey, is GPT 4 still in a comma ?

[cedar olive]

On their platform documentation

[wind urchin]

is this with chrome?

[stable dagger]

Opera. I dont know if it is related, but I also noticed that crtl-enter stopped working. So I cannot move to a new line for input, it just submits what I have written as if I pressed enter instead of ctrl-enter -- it only occassionally doesnt work and seems related to this issue

[wraith ginkgo]

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

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

[wind urchin]

Give Firefox a go, please

Hey. Pricing of gpt-4 is also not clear to me. So we pay 20$/months for plus account. Then there is cost per token. So is it like [20 + tokens * cost per token]? Also how we can see how much we spend until now?

[mild pendant]

You're mixing up OpenAI API with ChatGPT Plus. They're not the same.

ChatGPT Plus = $20 per month. (With [temporary] rate limiting (25 messages (messages ARE NOT tokens) per 3 hours)

OpenAI API 8k model = $0.03 per 1000 tokens

[wind urchin]

Plus is a 20$./mo service which lets you have early access to new features, right now, it's GPT-4. The API is a pay per 1k token system that is separate

[lapis jolt]

Is there any 3rd party way of getting access to the GPT4 api? Like how some discord bots now are using the GPT3.5 API if I'm not mistaken, for responses?

[wind urchin]

There should be no 3rd party way to access the GPT-4 API as it is in restricted waiting list beta

[slate eagle]

if for some reason, you switched to a lower GPT version in a thread due to rate limiting, once it is available again, how do you set the GPT version back to 4?

[wind urchin]

Ohh thank you it is clear now

[wind urchin]

right now, the only way is to copy the last few messages and paste them in to a new GPT-4 chat

[wind urchin]

thanks a lot

[slate eagle]

this thread is long, like 100s of messages long

is this really the only way?

[lapis jolt]

Yeah I figured that's the case. One can dream though. Thanks for clarifying

[wind urchin]

CHatGPT only remembers about the last 3000 words, so anything past that is not used for replies anyway

[slate eagle]

ah ok

hmm im not sure about that though, i reference lots of previous messages & functions it has produced and it knows what I am referencing

[wind urchin]

usually by inference from text later in the chat

[lapis jolt]

Yeah I was always confused about that too because I've seen it reference things like 20 messages ago and it's like "We've done a lot between then and now and this has never been brought up in that time."

[whole kettle]

GPT 4 has been fixed it seems! It's smart again, at least for the moment. Good job Open AI.

[wind urchin]

ok, lets suppose you are narrowing down on one particular aspect of the code over and over, iterating it each time, to you it would seem like it's got a super long memory, but all it is actually doing is inferring other information from the current context

[lapis jolt]

Wait, what? Just like that?

Oh wait your problem, not the one I was talking about. Sorry lost in my own world for a second. Duh.

[whole kettle]

Yup, refresh and open a new conversation, give it a few different prompts and you'll see now it's able to remember them and also give you smart answers again.

[lapis jolt]

Hmm okay then it was my problem too. Let's seee

[deft ravine]

anyone know how I can input images and output text in the image

[lapis jolt]

I could believe that too.

[wind urchin]

ChatGPT-4 cannot input or output images

[lapis jolt]

Yeah that's a GPT4 API thing exclusively, right?

[deft ravine]

yea I know is there anything that uses gpt to do that

any secondary website or something

[wind urchin]

aslo not available on the API, either

[wind urchin]

you could take a look at gpt-4 on huggingface

[whole kettle]

Oh boy, yes it's back. It can finally remember your initial prompts again and make suggestions based on the information you gave it, oh my God I missed GPT 4, feels good to have it back!

[lapis jolt]

Wonderful!

[whole kettle]

This probably means the mods here and the team must be actively listen to our feedback and when there is an issue they actually take time to look into it, that's great.

If that's so thank you very much guys, you're awesome.

[pliant mauve]

I have chatplus subscription but i kept on getting defect messages when generating responses. "Something went wrong. If this issue persists please contact us through our help center at help.openai.com". Does anyone know what i can do to resolve this issue. I have raised this in the chat a number of times with no joy?

[wind urchin]

Is GPT-4 worth the money?

[wind urchin]

depends on your use case

[wind urchin]

I mainly use it for ideas.

[brazen badger]

Yes

[wind urchin]

also depends a fair bit on your countries price index, for someone in the USA... it's a couple of coffee's, so.. not a lot, but for others it could be a big chunk of money, I think it is worth trying at least once, just to see what the very latest AI can do

[frail night]

hello, has anyone used gpt-4 for resume building? is it worth it?

im considering buying it but have been super unsure

[pliant mauve]

Great tool when it is working

[frail night]

(all the promotions i've seen have been paid promotions)

[brazen badger]

Recently getting close to reaching maximum potential with gpt4 I think it’s possible

[brazen badger]

It always works for me

[pliant mauve]

I am currently having issues generating response, do you have a workaround

[frail night]

How do I join for $20 a month ? I can't find the link

[brazen badger]

I use safari mostly

[frail night]

I found this https://openai.com/pricing

[brazen badger]

But no I’ve never had issues, sometimes it can’t generate but I just reload

[pliant mauve]

I kept on reloading but it still not working,

[brazen badger]

If you only want to use it for a resume then download Poe it gives 1 free message a day as far as I remember to use GPT4

[brazen badger]

Idk then sowwy

[wind urchin]

https://chat.openai.com in the left side, below the history, is an option to "Get Plus"

[frail night]

ty and is plus access to gpt-4? or does that require an additional purchase of tokens to use gpt-4?

[wind urchin]

No additional purchase required, GPT-4 is an early access feature of the Plus membership

[frail night]

thank you. 🙂 ❤️

[brazen badger]

Btw does anyone know if I’ll get plugins

I signed up on day 4

[fallow hawk]

I signed up 5 minutes after it opened and I don't have it either

[brazen badger]

Damn

[pliant mauve]

Thanks. I use chrome but i will try using edge

[whole kettle]

Signed up a month ago and still nothing 😄

[white pendant]

is chatgpt 4 gonna be free in the future?

[pliant beacon]

Not worth it yet, even if thats throw away money your spending, don't think its worth

3.5 should be just fine for resume building

[plush arrow]

#gpt-3-discussions

[pliant mauve]

its working on Firefox

[frozen spear]

Where can we go to discuss Auto?

[sand hull]

Who is going to cover the cost?

[warped shard]

Is gpt 4 still limited

[pastel estuary]

2021, yes

Idk when it will be updated to 2023

[warped shard]

What about the amount of chats you can send

[pastel estuary]

25 messages/3h

[warped shard]

Has it been fast for you

[pastel estuary]

Yes

[warped shard]

Yesterday a lot of people were saying it got fast. Could that be a prelude to making it an unlimited

[pastel estuary]

What?

Anyways I hate that it’s 25 msgs/3h

[warped shard]

Do you think that fact that it got faster could be them trying to scale up their systems in order to make it where you can send unlimited chats

[pastel estuary]

Idk but hope

[warped shard]

Yeah. I have a handful of plans for hobby application building iam planning to do once it gets unlimited to use

[pastel estuary]

Cool u heard any auto-gpt?

Abt*

[warped shard]

Yeah it’s honestly pretty cool

[pastel estuary]

U need a paid acc tho but I will fix it tmr

It’s not that user friendly tho but who cares I can still use it

[warped shard]

I honestly wonder how this technology will look like in a few years or so

[pastel estuary]

Bro u have no idea 😂

[warped shard]

This is like the early days of the internet

[pastel estuary]

Voice AI, AI in general, etc. gonna be crazy

[pastel estuary]

Either u hop on the future wave now or u rotten with the rest that doesn’t like it

[warped shard]

Ppl are gonna be ai reincarnating artists like Michael Jackson

True

[pastel estuary]

They alrdy doing it with rappers 😂

[warped shard]

One managed to clone drakes voice and make him sing a song he never wrote. And the music isn’t that bad

[pastel estuary]

AIs making better music than the real artists lol and ppl thinks it’s their official song as well

[pastel estuary]

Yeah did u hear Travis Scott’s new AI song and i think they dropped another fake drake song

[warped shard]

Yeah I wonder how they will all react to this. Combine ai voice technology with gpt 4 expertise in terms of lyric writing and music writing and you get a lyrical genius

[warped shard]

I have seen some people talk about it in twitter

[pastel estuary]

Also bruh you can eventually make a bag 💰 if voice AI getting this good 😂

Wonder how school will look like in 20 years 💀

[warped shard]

Iam predicting that someone is going to see this as an opportunity to build a platform where they have multiple ai using artist voices and gpt 4 to write music and offer a competitive pricing and marketing strategy to compete with Spotify or something like that

[pastel estuary]

Probs

[warped shard]

I honestly predict that within 20 years, all schools will be mandated to use like some form of virtual personalized teacher software for their students to use and the human teachers will be sitting there as the proctors monitoring the progress of every student

[hazy lintel]

trains

[warped shard]

The teacher will have the central control of all ai systems that the students use. The teacher can tell the ai to prioritize key subjects for the students to learn and give the idea a list of agenda and topics to teach the students

[pastel estuary]

They gonna block all AI type of stuff

It’s sad cuz our grand grand grand children will struggle to get rich cuz it’s like it’s getting harder and harder to getting rich

World leaders says u will not need to own anything but someone has to own u know

In the future

[warped shard]

I heard the world economic forum made a prediction where that in the future none will own anything. The way that they phrased it is very weird but you can understand what they mean by it. You will have ai creatively planning and machines producing and iterating things to perfection.

Little to no human involvement. But that’s way too far in the future

[pastel estuary]

Exactly this I talked with my friend abt in school today 😂 (yesterday mb)

Yeah we talking 100y and further like everything u will think of the AI has alrdy thought of that

[warped shard]

Exactly. Like you will rarely think or own something that is originally created

Unless if we combine with ai

[pastel estuary]

My friend using AI to combine words to unique ones etc. kinda smart e.f floor and amazing => floormazing 😂 u get what I mean

Maybe the fact that it’s been more faster lately it has smth to do with it

Idk

[warped shard]

Yeah. I honestly think they are trying their hardest to get to make it unlimited because they know that a lot of people will pay once that happens

Everyone that is willing to pay is complaining that fact it’s not unlimited

[pastel estuary]

@warped shard since the data cutoff 2021 it might provide alrdy existing “unique” name ideas and then it can be good to combine and take it as inspiration etc.

Patience. It will pay off sooner or later. Also for Plugins u need to sign to a waitlist

[night ermine]

wierd... it may not be worth it for everybody but to say that you dont feel the difference... it hallucinates a lot less. few times better for some programming questions and for learning languages.

[warped shard]

Once it’s unlimited iam paying

[pastel estuary]

Then if y don’t have money unsubscribe then resubscribe when it updates for ur preferences

[pastel estuary]

Unlimited, no cutoff and image option, etc. that will be so OP 💀

[night ermine]

if you ask it if something is correct with GPT3.5 its a coin flip. GPT4 is much more reliable for these type of questions.

[wraith ginkgo]

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

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

[knotty grotto]

few chance it will ever be unlimited

please stay respectful in the discord

no need to be rude

[pastel estuary]

I’m not rude

I don’t really mind I got money for it so 🤷‍♂️

[night ermine]

they arent selling you GPT4 for 20USD... they are selling you a plus membership, and one of those perks is beta access to GPT4.

if its the only reason you want it and nothing else matters to you thats on you to think over.

[floral knot]

Ok, since this chat is active I'll be posting this here too: What if there was some sort of massive database where artists and coders can submit their works to so AI companies can license them before scraping them? Every 00000001th of a penny an AI company gets goes back to an artist whenever something is generated from their works. Both regular artists and AI artists are happy. Win win!

[weary rampart]

No, you are paying for faster ChatGPT, the other features they just let you try.

[pastel estuary]

Exactly 💀

[pastel estuary]

Bro posted the most NPC idea ever

[floral knot]

Lol how?

[pastel basin]

Also, I think free gpt has a max of 100 message. I have not used gpt3.5 for a while but I think the cap is gone in the plus subscription?

[ripe forge]

is GPT 4 still really slow?

[floral knot]

@pastel estuary Tell me how.

[runic ember]

I keep getting errors with generating response all the time, does it work for you people?

[weary rampart]

Yeah, I mean, that is the reason I have plus, I don’t care about the faster 3.5. I’m annoyed about the 25 messages and hope it goes up soon, but they do present it more as a trial bonus for being a plus member. The real product is the gpt4 api. If they advertised it as paying for gpt4, I’d be much more annoyed at the 25 limit

[umbral fossil]

a man who annoys

[runic ember]

I feel lucky if I get one response today... well spent 20$ 😦

[umbral fossil]

where can i create images with text?

[ripe forge]

bing image generator. stable diffusion. midjourney. the first 2 are cool if you just wanna mess around. midjourney is tight if you wanna get deep into it. super cool tech there.

[serene cipher]

the gpt-4 model preview on chatGPT is very limited, unlike the GPT-3.5 models which are the ones ChatGPT have always been using so far

[pastel basin]

I mean plus gpt3.5

[umbral fossil]

Thanks man

[slow herald]

How do I get access to plugins and the new browsing feature? I selected I am using it for research and now I read I should have selected to be a dev? Would be really nice to test out plugins and web queries…

I am on the paid gpt4 plan so a little annoying that I cannot access all features for weeks - does anyone know how to get access?

[scarlet bridge]

patience is the only way.

[wise furnace]

Breathe in
GPT-4 is...
Breathe out
... in Beta.

[slow herald]

This is the way 😄

[hollow iron]

Is not 🤭

[restive galleon]

So since it’s beta access, that means gpt4 will be free later on?

[knotty grotto]

nobody know

[sand hull]

Gpt-4 api is cool as heck.

[timber wasp]

Is it possible to access cleared chat gpt conversations?

Or does anyone know the contact information of the developers?

[knotty grotto]

not really but you can contact [e-mail]

[timber wasp]

Ok, got it thanks

[wraith ginkgo]

This is a Safe for Work Discord Server

The OpenAI Discord is an actively moderated server.
• Refrain from sharing inappropriate content on the server. This includes but is not limited to messages, media, or other topics of graphically violent, sexual nature, and drug-related content.
• Report all sensitive and offensive content in the feedback reporting tool in the ChatGPT web UI instead of here on Discord.

Posting NSFW content will result in a permanent ban.

[timber wasp]

does anyone know if there is a phone number?

[knotty grotto]

no phone number

[timber wasp]

do you know how long it takes for them to answer?

[knotty grotto]

can take a bit

[timber wasp]

do you know an estimate? i am just in an urgent situation right now

[knotty grotto]

few days for sure

150 millions userwithin 4 month

[jovial kite]

Hello machines !
Can anyone help me if I could get an exception to extend my wording limit in my Chat GPT? I am a lawyer and want to go ALL IN with AI. But there is a lot to teach :I

Im talking about 10k-30k words, more if possible.

[knotty grotto]

unfortunately you cant change limit token input or output

[vestal steppe]

Context size limits are a hard technological restriction. You could consider workarounds, such as dynamically adding the necessary context based upon detected classes in the users' prompts. Or, subdivide your problem into smaller pieces (i.e. maybe handle 1 section at a time). Best path really depends on your exact use case and what information needs to be in scope for inference.

[jovial kite]

Are you guys moderators or something in Open AI? Or can there be a possibility that I could still look for?

Thank you both for your response by the way.

I have a HUGE case that Chat GPT is understanding but know I want to explain a bit more but for that is pages and pages of the case. It´s a personal case as well.

[vestal steppe]

32k tokens is the absolute maximum for this generation of the product. I have no affiliation with OpenAI or this discord server. Sounds like you might want to look into summarization and multiple phases of processing.

[jovial kite]

Could the chat store more than 100 pages of information? I mean could the AI process that amount of information in a single chat?

If so I would input as before, by phases. But maybe I can run out of storage and everything was for nothing.

Thank you for you time Dr. Lisa Su, I apreciate it.

[upbeat dome]

You need to do what I've been doing.

Meticulously comb through the text and rewrite it line by line to minimize tokens.

Use a website like Tiktokenizer.

One strategy I have is, if there's a lengthy concept that comes up over and over again, I abbreviate it with an emoji. Then, whenever that concept comes up, I write the emoji and save myself 50+ words.

[smoky crest]

Can’t

[jovial kite]

?Suggestions

[ripe spoke]

Hey, the answers that you get when asking chatgpt to write an essay should vary and be different each time.

[smoky crest]

No

Also @jovial kite anything you enter into openai is no longer classified, don’t get yourself in legal trouble

[shut herald]

anyone else has plus membership but cannot access GPT4? It has been like this for 2 weeks, and OpenAI just don't care I guess, but I care since I'm paying

[knotty grotto]

what error you have

[smoky crest]

You don’t get playground access / api with gpt plus

[shut herald]

There is no dropdown to select!

[smoky crest]

If that’s what your trying to do

[shut herald]

before I was able to say use GPT4 or 3 or legacy

[smoky crest]

Send a screenshot in #off-topic

Lemme see

[wraith ginkgo]

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

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

[shut herald]

It just says upgrade to plus on the bottom left

[smoky crest]

Then upgrade to plus

[shut herald]

I'm subscribed!

[smoky crest]

How do you know

[shut herald]

you are a genius aren't you

[smoky crest]

Click upgrade then

If you payed it won’t let you again

[shut herald]

In the help bot it says some users have reported this plus bug

I can't send image here apparently

I haven't unsubsribed, and I was even charged on 13th too

not gonna pay again for something that I can't access

[wind urchin]

if you have already contact https://help.openai.com regarding your Plus account issue, it is worth doing it again if you have been waiting for a while, things have calmed down somewhat now and you should get processed quickly

[shut herald]

Have created 3 separate threads

[smoky crest]

Your slowing it down

[shut herald]

at the end says someone will contact me within 30 minutes, but then here I am

[smoky crest]

The more tickets = less priority they will help you

[wind urchin]

yes, I was speaking with someone today who did one today and it suddenly updated a bunch of them for him, worth a go

[shut herald]

Thanks guys

[wind urchin]

most of the people with Plus account issues have now been delt with, so there there is less stress on the system

[shut herald]

Thought some admins check here, but probably not

[smoky crest]

This mainly just community chat

[thorny ermine]

anyone use gpt4 for calculus?

[smoky crest]

Openai can’t math well

[thorny ermine]

damn

[smoky crest]

Lol Goodluck

[wind urchin]

The helpers, volunteers, mods and leads all help to amplify your voices and the server is regularly visited by members of OpenAI staff who check through bug reports and suggestions and all of the channels here

[gilded apex]

Anyone else have issues with descending order lists?

[shrewd thunder]

you really trying to cheat lol

[thorny ermine]

yeah

[shrewd thunder]

if you are that **** then use wolfram

[thorny ermine]

how would i use wolfram ?

[wind urchin]

depends how big they are, the nature of the feed forward neural net makes it difficult to process complex sorting lists, simple lists are fine, but reverse sorting or other variations can cause issues if the list is long

[shrewd thunder]

gl on your tests

[thorny ermine]

cant tell if your messing with me or not

[shrewd thunder]

you can use wolfram

but you are goign to fail your tests

as you wont understand anything

[thorny ermine]

well then thats kind of pointless

[gilded apex]

It lists various “top X” from 1 to X no problem but when you try to list from X to 1 it’s weirdness

[crude magnet]

Are there any useful image generation AI tools available?

[jagged pulsar]

What upgrade just happened? GPT-4 is far faster, and it is making fewer errors than even earlier today

[shrewd thunder]

well if you have gpt-4 plugin access

then use the wolfram plugin

to help you if you want to understand the concepts

[wind urchin]

yea, imagine if you could only think about what you had just written, that's how it is for ChatGPT, it can't make use of what it's about to write

[smoky crest]

Yea it works pretty well

[shrewd thunder]

it does i like how the model can understand when to use the plugins

i wish they would provide the sourcecode for that

at least just that part

[hollow swan]

why isn't chatgpt working

[thorny ermine]

alright well i just joined the waitlist for the plugin, anything you recommend i do meanwhile besides studying lol

[wraith ginkgo]

  Attention  Off-topic chats have been cleared by a moderator.

[wooden valley]

lol

[hasty gale]

It would be neat if gpt4 parsed this server for inappropriate responses

[hearty plaza]

Bro what happens to my chat gpt

It won’t even load

[smoky crest]

It does

Tap the drop-down on the plug-in

It says what it’s doing

[hearty plaza]

Is chatGPT down ?

[smoky crest]

No

[shrewd thunder]

not that i want the sourcecode for how the gpt model recongizes when to use plugins

[hearty plaza]

Mine is taking years to load for an answer

@smoky crest

[smoky crest]

Simply ask it might give it to you

[jovial crescent]

I want to check my coding assignment and it's a bit a lot, chatgpt doesn't allow me to input these many lines. Is there any work around for this?

[zenith fox]

Do you know Pantone color cards?

[shrewd thunder]

mmm i will ask i am pretty curious

[sinful lichen]

what's your version?

[smoky crest]

∫₀¹ (x² + e^x) sin(πx) dx + lim (n→∞) Σᵢ=₁ⁿ (i² / (n² + i)) = ?

It responded is ∞ (infinity).

[shrewd thunder]

the model doens't know math well

[smoky crest]

Said source code was

integral = Integrate[(x^2 + Exp[x]) * Sin[Pi * x], {x, 0, 1}];
limitSum = Limit[Sum[i^2 / (n^2 + i), {i, 1, n}], n -> Infinity];
result = integral + limitSum;
result

[shrewd thunder]

are you joking

[smoky crest]

It’s what the bot said

[shrewd thunder]

what

[shrewd thunder]

the model looks giving wrong answers

[smoky crest]

What’s the right answer

[trim tusk]

Any plans for the cap to go above 25 messages any time soon?

[smoky crest]

Probably in a few months

[trim tusk]

Good

I’ve learned more from GPT-4 than I have from school

School is a joke

[fleet ice]

Just saw a TedX video about GPT4 from Greg and he said it doesn’t understand math very well. Time to go still

[trim tusk]

I feel bad for people who have students debt. GPT-4 is only $20/month

[hasty gale]

Its only a supplement. Advanced engineering concepts are poorly explained and lacking, especially when it comes to mathematics

[smoky crest]

The plug-in mode can tho

[trim tusk]

We will see in the next 5-10 years

[hasty gale]

You never know, it could improve, but as of now its not good for engineering. Idk about other fields

[trim tusk]

I’m sure it could be a lot more advanced than it is now. But we have purposefully contained it from learning too fast

It will disrupt society too fast if we don’t contain it

[hasty gale]

Idk. I don't want to make any sort of predictions or judgements. Maybe it will, but there is no evidence in either direction, unless you refer to movies

I can tell you as an engineer, and someone with a math degree, it sucks at both

[trim tusk]

Google CEO already said so in the 60 minutes interview

[fleet ice]

From Greg’s talk didn’t get that feel. Public feedback into it is helping it learn more than they could

[hasty gale]

maybe itd be good at trading stocks or something

[fleet ice]

This might be true as stock@market is mostly algos trading with news thrown in. If GPT4 has access to news at the same pace as these algos, might be one day it will

[hasty gale]

I don't trust something just because a ceo says it. Besides, its different if it is coming from a developer in the field. Sundar has a metallurgy degree Im pretty sure, and google is working on bard rn, so it would make sense for him to say this even without merit

[trim tusk]

Who is Greg

[smoky crest]

Co founder

He’s really cool

[hasty gale]

I suppose everyone would be at war with each other to have a better trading algorithm then lol

[trim tusk]

Fair enough. But I’m the interview he was asked about the potential dangers, and before he even answered the question he had a nervous/twitch expression on his face

I wouldn’t underestimate people who had had behind the scenes experimentation with these AI tools

[hasty gale]

If you think the ceo is pointing directions for development of Bard, then you're sorely mistaken. Hes on the bottom dollar team, not the engineering team. This is why they released a garbage alpha for Bard that was an embarrassment. If he was involved in it, he would've said no to a pre-release

[fleet ice]

Their own teams begged to NOT release bard

[hasty gale]

correct

[iron stratus]

hello gpt
4

[trim tusk]

All I have to say is that AI is going to outpace all humans. I don’t think anyone can argue with that ¯_(ツ)_/¯

[fleet ice]

Besides Igor now left Google and probably is hired by Elon

[hasty gale]

you never know, it may happen, or maybe we will integrate it into our lives

[knotty grotto]

#ai-discussions please

[trim tusk]

Thank you

[hasty gale]

I'd bet that google probably has the best people in the world in terms of machine learning, but they were really late to the party

[trim tusk]

Does anyone know when the picture support will come to GPT-4 to the public?

[knotty grotto]

no one know

[trim tusk]

Surprises are good. Keeps things interesting

[fleet ice]

Does anyone know of Boston dynamics having any interest in GPT4?

[hasty gale]

I know a few people that work on algorithms for image recognition. I don't think the issue that they have is in recognition of images, according to my colleagues it is more security related in terms of vulnerabilites

[trim tusk]

Also, does anyone know why when I paste a URL from Genius with a song, then ask GPT-4 for lyrics analysis, GPT-4 will respond with the wrong lyrics? Sometimes even gives very concerning R rates lyrics that are nowhere found in the song

@hasty gale did you make your profile pic with DALL·E 2? I did

[hasty gale]

considering most people using chatgpt don't develop ML id say nobody knows

[hasty gale]

no I didn't

its just a default

[trim tusk]

Default?

[hasty gale]

Yeah, I think u gotta pay for the dcord subscription

may be wrong tho

[trim tusk]

Also, does anyone know why OpenAI isn’t going to significantly reduce the cap anymore? At least, I’m assuming that because the message no longer mentions that

It just says 25 message cap now.

[lunar crow]

no
humans using AI will outpace humans

[hasty gale]

probably

But you never know

[trim tusk]

Good to be skeptical, ayyy

[lunar crow]

AI can't be creative, humans can
stuff needs emotion and meaning

[trim tusk]

Yet.

¯_(ツ)_/¯

[lunar crow]

it's not a yet thing, it's a never thing

it's just not how computers work.

[hasty gale]

If ML can simulate creativity, then what does it matter. Creativity is just how you perceive something as different and unique, so why cant a net do that?

[lunar crow]

not really

[trim tusk]

We should move this to AI discussion lol.

[hasty gale]

not really to what?

[mellow solstice]

emotions are just goals injected into your consciousness for you to solve so AI can have creativity

[lunar crow]

wait there's a channel for that?

then yeah probably

[hasty gale]

yeah i moved there

[brisk heart]

is it just me or is the code block window so tiny on chatgpt?

[smoky crest]

Just u

[rare cedar]

How are you

[brisk heart]

i have 27 inch screen

[outer gulch]

not sure if mentioned already, but Greg Brockman (OpenaI CEO) demonstrated the plugins in a TED talk he gave yesterday. Worth a watch www.youtube.com/watch?v=C_78DM8fG6E

[fleet ice]

GPT4 can write poetry; isn’t that creativity?

[fleet ice]

Watched today; awesome improvements; can do much more and connect the dots

[regal zephyr]

is there any way to get image recognition?

[stiff flicker]

How it do that ?

[regal zephyr]

thats literally what I am asking

[tranquil pebble]

Hi How many devices can use with gpt plus in the same time

[regal zephyr]

what

[sand hull]

Pretty sure the number is infinite, but you can only have an active session on like one.

[tranquil pebble]

With one account subscription, how many devices can i use in the same time

[sand hull]

Not public yet.

[tranquil pebble]

A don't think that the number isinfinite

[stiff aurora]

can we still purcahse chatgpt plus?

declined cards all the time

[tranquil pebble]

It's should be limited by i don't how much

[stiff aurora]

tried all my cards

[sand hull]

Well not literally, but you can log into a ton of devices, like a Google account. They'll eventually say your sessions expire and will log you out, making you log back in though.

We can't really say any exact number here.

[tranquil pebble]

Yes , that's in the case of the free access

[sand hull]

With paid access your sessions still expire, just you have priority access to getting logged back in.

[tranquil pebble]

Okey thanks

[fickle glade]

Hey guys, is it worth to get a paid version?

[tulip summit]

试一试

[magic wave]

/ image

[stiff flicker]

worth

[maiden gull]

Wow what a scam. I paid for a 2nd account so I could have 25 more prompts per hour but it says I maxed out my gpt-4 prompts already and I just created a new account and upgraded it

[cold ridge]

Try using a new browser or an incognito window

[maiden gull]

worked

[wraith ginkgo]

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

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

[zenith beacon]

guys im lloking for an Ai that starts with "ph" i forgot its name , it has chatgpt4

[trim tusk]

Did you try asking GPT-4

[zenith beacon]

i dont have chatgpt4

[spark pilot]

I keep trying to login to CHATGPT but it gives me the following error: "We ran into an issue while signing you in, please take a break and try again soon." I've tried 1000 times. turned off vpn, cleared cached, tried multiple browsers, multiple accounts, private browser mode. Does this sound familiar?

@spark pilot hi! btw!

[median sedge]

The GPT-4 api does

[leaden cape]

i said chatgpt

[pseudo topaz]

No it doesn't

[ember cloak]

we as humans make ourselves out to me more complex than we actually are. there is no doubt in my mind that we will see creative ai models in the future

[thorny valley]

AI is going to replace so many people both making it and not making it. It's not like it can be stopped because localized LLMs and other competing AI will not cease development. People should just be ready with some plans in place for when replacements will take place. It's just how it is unfortunately. If AI doesn't replace then other ways non-AI related will. There is no one way of getting rekt or success for the meantime but many many many ways.

[leaden cape]

ngl i think GPT-5 will be the last LLM for quite some time

[mystic nest]

it's the same for all technologies revolution, or even when robot/full self driving/etc get materialised, ppl just shift jobs 🤣 and likely, the ppl "here" wouldn't be affected that much

[worthy parcel]

they are adding that functionality in the coming months

gpt-5 is not being worked on.

[pseudo topaz]

• rumor

[worthy parcel]

not a rumor, said by the CEO himself in the recent TED talk

[mystic nest]

if you know you need to change/update/upgrade, that resolve more than 1/2 the issue

it's ppl that dont know

[leaden cape]

they are adding plugins

[worthy parcel]

yeah plugins gives chat gpt access to the internet, files, dalle, and be able to execute code

[leaden cape]

gotta keep in mind that GPT-4 was released 4 months after ChatGPT was launched tho

[delicate harness]

Hello everyone.
Stop fighting.

[leaden cape]

what

[delicate harness]

Can I say?

[worthy parcel]

meaning?

[leaden cape]

could be a PR move to keep out investigations by leakers n stuff

if the public thinks nothings happening they would most likely not care

[delicate harness]

You are stupid!!!

[worthy parcel]

what are you talking about?

[delicate harness]

About yours...

[leaden cape]

he said it so that no one cares but actually somethings happening

just because the CEO said it doesn't mean its true

[worthy parcel]

???

I mean sure but you have no proof

and there is no reason to believe they are working on something else

[delicate harness]

Can you hear from me?

[worthy parcel]

they gave working demonstrations inthe ted talk

[worthy parcel]

yeah, whats up?

[delicate harness]

@worthy parcel
What are you talking about?

[worthy parcel]

I was talking about them adding plugins to chatgpt, and this guy didnt believe me

[trim tusk]

Who here loves GPT-4

[delicate harness]

@worthy parcel
I think it is simple.

[leaden cape]

what

i said something about GPT-5 not plugins

[worthy parcel]

oh

[leaden cape]

🤨

still dont get what @delicate harness is trying to say

[fleet viper]

？？？

[eager bluff]

When will chatgpt 4 release for free?

[leaden cape]

probably when GPT-5 releases

[mystic nest]

when the cow comes home

no one knows

[tall scroll]

If you run out of messages in GPT4 and switch back to 3.5, are you just screwed? I see no way of switching back.

[leaden cape]

click "New chat" @tall scroll

[tall scroll]

I don't want to make a new chat though, I had a whole star wars RPG going on.

[amber parcel]

Any news on plugins?

[leaden cape]

then you need to wait the 3h until you get 25 new messages

[tall scroll]

I feel like it's been 3 hours, but maybe not. So I'll see an option to resume with GPT4?

[leaden cape]

your feeling can differ from facts

[spark pilot]

Any help here please? I keep trying to login to CHATGPT but it gives me the following error: "We ran into an issue while signing you in, please take a break and try again soon." I've tried 1000 times. turned off vpn, cleared cached, tried multiple browsers, multiple accounts, private browser mode. Does this sound familiar?

[amber parcel]

Well yeah, once you run out its out. If your in a conversation already, nothing changes, only the model, so you might notice a decrease in response quality.

[leaden cape]

you can't change the model mid chat

[amber parcel]

Never said that. Once GPT-4 runs out, GPT-3.5 is used from that point forward.

[tall scroll]

That's what I'm talking about. It can switch me to 3.5 but not back to 4 when the time is out?

[mystic nest]

if you ran out of limit, that session would become 3.5 like forever, need to start a new chat. at least that's the case previously

[wraith ginkgo]

<:book_icon:1046291503158399107> `` Rule 1 `` Be respectful.

Treat others the way you would like to be treated, and assume best intentions. Don’t harass or attack others, and don’t engage in hateful or generally malicious behavior (e.g. sexism, racism, homophobia, etc.). Keep the negativity to a minimum.

[manic flicker]

😀

[amber parcel]

Howdy Ricky.

[leaden cape]

that wasn't me but sure pal

[amber parcel]

Never said it was, but I felt a reminder was needed 🙂

[leaden cape]

ok?

[amber parcel]

👍

[wraith ginkgo]

<:book_icon:1046291503158399107> `` Rule 1 `` Be respectful.

Treat others the way you would like to be treated, and assume best intentions. Don’t harass or attack others, and don’t engage in hateful or generally malicious behavior (e.g. sexism, racism, homophobia, etc.). Keep the negativity to a minimum.

[tall scroll]

🤦‍♂️ 🤦‍♂️ 🤦‍♂️ 🤦‍♂️

[leaden cape]

gpt 4 is so much better than gpt 3 ngl

[manic flicker]

Could you give me some suggestion for use this tools

[leaden cape]

coding, translating, work

[manic flicker]

Thanks

[mystic nest]

if you are a student, the default is "homework" 🤣

[tall scroll]

I use GPT4 to play a text based RPG, kind of like Dungeons and Dragons. It's incredible, really.

And if I can't switch back to GPT4 I'm going to be sad. But I'll just start over I guess.

[leaden cape]

unfortunately most of the outputs are detected by AI detection tools

[mystic nest]

you can ask 3.5 to do a summary of the actions, then paste that to the new chat, at least maybe it's not starting afresh

[tall scroll]

You can tell it to make the output undetectable by AI detection tools xD

[tall scroll]

That's not a bad idea.

[leaden cape]

doesn't work

[mystic nest]

well, only students that dont know how to use it "properly" will get detected

[leaden cape]

mine don't get detected because its not in english and the tools arent trained to detected other languages

[mystic nest]

and when I say homework, doesn't mean cheating, can use it to study and explain concept

🤣

[tall scroll]

I wonder if I could just copy and paste the whole chat into a new GPT4 chat... Might be too long.

[mystic nest]

yeah, it's likely too long, you'd hit the cap again

[leaden cape]

using chatgpt to fix your errors (grammatical and whatever) can be useful tho

and let it tell you the errors so you don't make them again

[crimson mesa]

Definitely can get it to not be detected, I’m a teacher and have researched this a fair bit. Just have to do a few extra steps and tricks. We won’t be teaching ‘English’ anymore but how to communicate with ai. I am all for it, finally a level playing field for neuro diverse students.

[mystic nest]

yeah, need to look more on the good side of things. students will alway cheat anyway...

with or without chatgpt

"Six trainee lawyers who cheated in exam withdraw Bar admission applications" <= even at this level

[glacial zealot]

Can you insert images into gbt 4 yet

[crimson mesa]

School based teaching and education is a joke anyway, the sooner we enrich students learning with meta cognitive constructs of learning, critical thinking and creativity as core subjects the better

[steel hill]

It just means that what and how we teach things need to change

[mystic nest]

the problem is chatgpt still have hallucination problems, so education systems wouldn't allow it to be fully implemented

at most as assist tool

[crimson mesa]

They won’t allow learning full stop

[steel hill]

It only hallucinates if it doesn’t have context

If knowledge is there…is quite good

But forget to pass in a variable to it..and it can make 100% of it up

[crimson mesa]

Seriously all systems of current society are on the verge of collapse. AI is needed to simple produce a continuum. We are also on training wheels atm, I can wait till plug-in access, game over! Bring it on!

[mystic nest]

it can be good for marking papers and explain the mistake 🤣 usually teachers dont give full explaination..some just "X" or big fat 0

[steel hill]

I actually put a lot of work into grammar and writing for the GPT prompts

Has a very small memory want to make the most of it

[mystic nest]

I just hope it will change the "need to memorise" part of education, at least partially

sneak in some chatgpt 4 stuff before mod comes in... gpt plugin will help a lot in many ways when it get released

[steel hill]

My guess is shifts to assisted out of class work and unassisted in class

As it’s just going to get better and harder to tell

[wraith ginkgo]

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

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

[mystic nest]

I just watched a video on changing the Lexile Level of content, think that would be pretty interesting for educators too

[wind urchin]

hey everyone

do anyone knows about fiitjee?

[wraith ginkgo]

This is a Safe for Work Discord Server

The OpenAI Discord is an actively moderated server.
• Refrain from sharing inappropriate content on the server. This includes but is not limited to messages, media, or other topics of graphically violent, sexual nature, and drug-related content.
• Report all sensitive and offensive content in the feedback reporting tool in the ChatGPT web UI instead of here on Discord.

Posting NSFW content will result in a permanent ban.

[mystic nest]

I know some kids that are troubled by the "questions" itself when teachers basically ignore their reading skills which vary a lot

[crimson mesa]

This is what I am talking about…I have 150 students, I want to give feedback, diversify content, extend, scaffold, moderate, modify, adjust etc…modern teaching, expectations and capacity to achieve all this is why there are not enough teachers right now. We need ai assistants just to meet current expectations. Also then we might move past the dumbed down curriculum and lack of student engagement. Anyone from open ai listening to this hit me up, might be good for you to have someone on the team that is willing and ready to adopt ai into education that actually has experience of what is needed

[steel hill]

I see a lot of opportunities for creative people with good communication skills and some technical competence for future software development. Eventually the only requirement might be creativity, but GPT not quite there yet.

[crimson mesa]

💯

[upbeat igloo]

gpt4

[steel hill]

What I have been finding, is I put so much effort into the English to explain what I want GPT 4 to write, that I am much clearer on the problem. Additionally, I’ve never had such well structured, commented, and documented code. Is a pretty massive difference in what GPT can do with excellent instructions vs poor ones.

[mystic nest]

yeah, prompt engineering is a thing nowadays

[steel hill]

GPT sometimes goes down the wrong path..and you need to be able to steer it… I’m guessing in a year or two..it will be much better, but that just means people will ask for more complex tasks.

Still blows my mind though….it has some real creativity at times,

[mystic nest]

we will see more complex use cases when the full multimodal features are made available

[wraith ginkgo]

`` Server Tip: ``

Add your well-crafted prompts to our #1019652163640762428,
or share your interactions with ChatGPT in #1050184247920562316!

[patent salmon]

Hello everyone, I am not able to find the cancel button from 'manage subcription' tab. How do I cancel the subscription? Thank you for your kind help

[mystic nest]

click on "My Plan" > "Manage My Subscription"

you should see the cancel button

unless somehow your UI is different from mine 🤣

[patent salmon]

There is no 'Cancel' button here.

[sand ice]

Hey everybody I'm doctor how can apply AI in medical sciences and what's the scope for future

I need some insight bout it

[patent salmon]

My payment method, bulling information and invoice is all empty idk why

[mystic nest]

do you see the "ChatGPT Plus Subscription
$20.00 per month"?

I mean the duration of your subscription

[patent salmon]

how do i send screenshot here?

[mystic nest]

you can do it in off-topic but make sure to mask your info

#off-topic

but if like what you mention, there is no info then

[patent salmon]

my plan is chatgpt plus usd20/month, I can still use gtp4. But, I find a way to cancel it

can't*

[brisk heart]

does anyone find any difference betwen chatgpt vs gpt 4? i feel its the same

[mystic nest]

you mean 3.5 vs 4 or chatgpt 4 vs api gpt-4? probably need to be clearer

[brisk heart]

i havent used the api yet so reffering to 3.5 vs 4

[mystic nest]

yeah, you probably need to email support, there should be invoice history

[patent salmon]

I did 3 days ago no reply that's why I'm here. Anyways I went to the off-topic section and asked there

[mystic nest]

yeah, I saw ur screenshot, it's probably bug, so unlikely we can help

[mystic nest]

you can try asking something more complex instead of direct question to see the differences

[wind urchin]

Hey @waxen valley

hey @serene cipher

hey@raw sierraa

[red hearth]

GPT-4 is very different in my practice.

[wraith ginkgo]

<:book_icon:1046291503158399107> `` Rule 4 `` No spamming.

Do not spam or flood the server with messages, as this is disruptive. Mass pings, all-caps messages, repeatedly posting in a short amount of time, or making very similar posts sequentially are considered forms of spamming.

[tulip frigate]

how good is GpT-4

can it do maths

better than previous 3

[wraith ginkgo]

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

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

[hybrid tree]

Great question for ChatGPT-4

[stoic arch]

what is chatgpt4?

[deft spindle]

This image shows a modern minimalist living room scene with sunlight shining on the sofa, a blue sofa, and a crimson sofa. Hope this meets your needs.

[rustic grotto]

Do you know the writer Borges?

[fallen musk]

There are ethical considerations with using ChatGPT for medical purposes, but for what it's worth, if you're just doing science and research, it's worth knowing that ChatGPT can pass the US Medical Licensing Exam (without clinician input.) That might be useful for something you're doing. (Please consider the ethics and privacy considerations involved.) Do a google search for the article: ChatGPT Passes US Medical Licensing Exam Without Clinician Input (I can't paste the URL because of discord rules.)

[mental kite]

how do i know which one of these bots is lying to me? 3.5 says one thing and 4 says another
i'm trying to get it to read a pastebin link and 3.5 says it can and is while 4 says it can't do that?

[fallen musk]

ChatGPT has no internet access at this time. So if 3.5 is saying it can read it, it's "hallucinating" (the jargon term for describing it doing something that it sincerely believes, but which has no basis in objective reality.)

[mental kite]

i can't tell if 3.5 is actually reading it

oh

i'm gonna go give it a piece of my mind for lying to me

[fallen musk]

It's also important to remember, it's a "large language model" or "generative pre-trained transformer" and doesn't have actual beliefs, thoughts, etc.

The best thing to do is to press the "thumbs down" icon and explain it's telling you something that isn't true.

OpenAI apparently reviews those, and the feedback that passes human review is added to the reward model. So it would be added as a "negative reward" to discourage it from doing something similar in the future.

[mental kite]

oh idk i just cussed it out a bunch i don't really want the devs to see that in feedback whoops

sorry

we're friends now though again, no hard feelings from either of us

i'll keep in mind for the future, thank you for your assistance

🙏 💪

[primal bolt]

"You've reached the current usage cap for GPT-4. You can continue with the default model now, or try again after 11:49 AM. Learn more" please fix this....

[narrow reef]

virtual beatdown > real beatdown

[mental kite]

😇

[vale magnet]

11

[wind urchin]

There's nothing to fix, that's how GPT-4 works.

[tacit maple]

gpt is described as predicting the text users want to hear right? i understand how it processes text, but how do plugins fit into that? since it's not just predicting text anymore but actually having to solve problems n stufd

[tranquil radish]

we need about taxi app code

[wind urchin]

Anyone from Fiitjee?

[tiny pelican]

Everytime I try to use ChatGPT, I receive an error. I have tried everything. Nothing works. Can someone help me?

[wind urchin]

restart it

[tiny pelican]

What do you mean?

[wind urchin]

Means close that app and then open it

[tiny pelican]

Didn't work.

[wind urchin]

restart your computer

[copper mirage]

🇮🇩

[tiny pelican]

I've tried that before, but for some reason it worked now. Thank you!

[wind urchin]

Okay Welcomne

[timid hamlet]

@open crescent

kir khar to kos nanat @modern mason

[unkempt zephyr]

anyone know how to fix the chatgpt errors? whenever, i input a prompt it replies with "Something went wrong. If this issue persists please contact us through our help center at help.openai.com."

I don't know, what to do anymore

[wind current]

f5

[unkempt zephyr]

still persists. deleted history/cache, restarted browser, restarted computer, etc.

is chatgpt not available for free users, anymore?

[pliant pagoda]

I'm on a plus subscription and chatgpt is not working for me either

[unkempt zephyr]

dang. i'm not sure, what's going on.

[wind urchin]

Use incognito, a different wifi, or a vpn. That message usually appears in relation to your connection.

[pliant pagoda]

You see anything in browser developer tools console?

[wind urchin]

Btw, if you're a free user, this has nothing to do with gpt-4, so you should move to gpt-3 discussions.

[lean trench]

Hello, how do I make gpt produce code without it ever saying no

[pliant pagoda]

I get this error:
No access token when trying to use AuthHeader

cleared cache, incognito, restarted modem. No joy.

[unkempt zephyr]

i just tried incognito and same thing occurs

[pliant pagoda]

Where can you report an issue around here? Couldn't find anything on the website

I'm on a paid plan, so I expect some kind of support option.

[novel spear]

A poem that describes the hard work and dedication of nurses,

Please present it in the form of poetry

[wraith ginkgo]

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

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

[azure wren]

Is the chat down for everyone else?

[azure rune]

it is working for me

[unkempt zephyr]

still down for me

[azure wren]

I just logged out and back in. It's working now.

[limber violet]

it is working for me, I'm in southern california

[steel hill]

Same , in CA too, sure they are in every data center they can find…

[tired jackal]

Why chat gpt doesn't work in Egypt??

[supple violet]

not a supported country due their censorship laws, etc.

[tired jackal]

😢😢

In future it will work in Egypt??

[agile bramble]

there was a way to access it through discord

but they took it down

[tired jackal]

Bruh

[sonic plover]

Has anyone tried to use GPT4 to search for old memes using the containing plot? Until now it looks like it completely fails with that task.

[manic dragon]

Hello, does GPT-4 currently have a limit of 25 messages?

[slate dragon]

DAY 30 Waiting for plugins

[bright oyster]

Yes.

[pliant ocean]

Guys I found how to get gpt 4 for free

[sharp hornet]

how

[nimble vessel]

Yes there's still the 25 message limit.

[manic dragon]

Thank you for your reply. Hopefully the limit will be removed and I can get a Plus subscription.

[wind urchin]

There are some companies saying on Twitter that they have basically made re-skinned versions of GPT-4 that are free. However, it's highly disputed and a lot of people are saying that they're lieing.

[fallow pond]

hmm

[pliant ocean]

Chat. Forefrontchat. AI GPT 4 FOR FREE TODAY ANNOUNCED ENJOY GUYS

Chat. Forefrontchat. Ai

[wind urchin]

This is exactly the one that people dispute is actually GPT-4. Lots of people saying it's GPT-3.5

[pliant ocean]

Bro click the plus button and click gpt 4

Click plus button then gpt 4 for free

I used gpt 4 for free in this site

100% freee

[manic dragon]

I don't see GPT-4 on the site, I clicked on "New Chat" but there's nothing there.

[pliant ocean]

Click plus button

[tulip solar]

Whats all this propaganda fr xd

Plus is 20$ a month "free" lol

[manic dragon]

He doesn't speak for the official ChatGPT

[mystic nest]

he's promoting other website

[manic dragon]

He speaks for another chat room which uses the GPT-4 model

[wind urchin]

Lol, just because a button says that it's GPT-4 doesn't mean that it actually is.
Check what people are saying about it on Twitter.
It's highly likely that this company are lieing about it actually being GPT-4

Just go to Twitter and read what people are saying about it.

[tulip solar]

Which company?

[wind urchin]

Forefront

[boreal ingot]

Not a single tier app will offer to their random users to use gpt-4 indefinitely for free. Gpt4 cost a LOT of money and just a handful of people got their hands on the API. Any claim of people saying they found a workaround to work with gpt4 absolutely free is full of sh*t

[manic dragon]

Found the button, I'll check it now

[wind urchin]

Exactly.

[tulip solar]

Legit can't find anything but the news articles about forefront

[wind urchin]

The cost of API calls is way too prohibitive to offer this for nothing

[pliant ocean]

Guys click plus button

Then gpt 4 for free

[wraith ginkgo]

This is a Safe for Work Discord Server

The OpenAI Discord is an actively moderated server.
• Refrain from sharing inappropriate content on the server. This includes but is not limited to messages, media, or other topics of graphically violent, sexual nature, and drug-related content.
• Report all sensitive and offensive content in the feedback reporting tool in the ChatGPT web UI instead of here on Discord.

Posting NSFW content will result in a permanent ban.

[wind urchin]

Reported.

[coarse yoke]

How does ai impact in embedded software field

[pliant ocean]

Are you found

[terse tide]

hallow

[tulip solar]

This guy made discord today...

[manic dragon]

Well I will say this - seems to answer the questions, you have to watch and follow it. Thanks for the information.

[pliant ocean]

Did you use gpt 4

[wraith ginkgo]

`` Server Tip: ``

Add your well-crafted prompts to our #1019652163640762428,
or share your interactions with ChatGPT in #1050184247920562316!

[pliant ocean]

In only one site for free now

I tested hardest math

[manic dragon]

Followed them on Twitter. I will follow them and their updates.

[tulip solar]

It just... Isn't replying to me

[hardy bough]

Are you chatgpt

[stiff thorn]

is the gpt4 word limit really 25k cos when it writes code it always cuts off after like 120 lines

[wind urchin]

You can ask it to continue from a few lines before it cut off

This has worked for me

[stiff thorn]

yea but does anyone know why it cuts off so early, and not when advertised

[tulip solar]

Bro the forefront AI just isn't talking to me like... At all?

[wind urchin]

Asking for a few lines before it cut off means you can verify that it's actually continuing correctly from where it was

[supple violet]

25k words requires 32k model which is not available to anyone at the moment

[stiff thorn]

oh

ok

[manic dragon]

It looks like the site has fallen a little due to the heavy load)

[junior elbow]

chatgpt plus is expired but still able to use gpt4, is that normal?

[dapper sequoia]

I agree

[gray ether]

then there's me, who's in Italy and can't use it

[dapper sequoia]

Woww thats depressing

[manic dragon]

You can, it's just that the site is either under a DDoS attack, or the load is too heavy. I think they'll fix it in a few days.

[tulip solar]

Its... Banned in italy..

[gray ether]

no it was banned by the Italian government

[pliant ocean]

I don't think many people know about this site

[tulip solar]

Jamshid yo site aren't working

[manic dragon]

I mean Forefont Chat, blocking does not apply there?

[tulip solar]

Forefront chat aren't even replying to anything

[manic dragon]

He is now under wild load or under DDoS attack. I think we need to wait a while for the situation to stabilize and we can use it normally.

[gray ether]

it was good while it lasted, now i can't go there anymore

[pliant ocean]

NO site is working

[tulip solar]

It might of been just the guys API key linked to a site....

[gray ether]

but was it only banned by the Italian government or also by other governments?

[tulip solar]

He prob ran out of credits

[tulip solar]

ChatGPT? I think just italy?

[timber osprey]

Wow

[gray ether]

okk

[wind urchin]

Guys, does anyone else have issues with ''payments'' page?

**I got this error: **

The payments page encountered an error. Please try again. If the problem continues, please visit help.openai.com

I have a sub, but since yesterday I got this error, tried clearing cache/cookies, relog, even tried from another device, its still the same.

[supple ibex]

Italy means what ？i，am chinese man

i learn english in this

[pliant ocean]

Guys, site works

[young bronze]

Good morning everyone

[wind urchin]

Enough shilling.

[cerulean obsidian]

Anyone has idea of Co-pilot X is out or beta waitlist?

[young bronze]

I tried via the app and got through

[wind urchin]

Via what app? 🙃

[wraith ginkgo]

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

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

[sturdy summit]

/imegine prompt: cat students

@wraith ginkgo

[supple violet]

imagine is midjourney which has nothing to do with openai

[wind urchin]

heh that prompt used to get a 60 second mute

i think it still does if you don't typo

[sturdy summit]

write a poem

[golden magnet]

Hi all, I've deleted my account on ChatGPT site because I've wanted to sign up through a different method (previously I did it via google single sign on). My current situation looks like this:
I can't sign up via different method,
I can't use my previous email or phone because platform states that my account was deleted 🙂

In short it appears like my email address and phone were blacklisted 😄 Is it a bug or a feature?

[knotty grotto]

This isn't chat.openai.com

[knotty grotto]

Delete account is permanent

[golden magnet]

Permanent means that my phone is remembered till the end of days and blocked?

[knotty grotto]

Unfortunately you can't re use.

You will need to contact [e-mail]

[wind urchin]

https://help.openai.com click on the pen icon in the bottom corner and ask the bot

[golden magnet]

Oh, I was looking for that email address. Thank you! 🙂

[knotty grotto]

You can try also the solution of foxabilo

[golden magnet]

Bot was not helpful at all 🙂

[wind urchin]

Neko is just better at helping than I am

[mystic nest]

support might take quite a while 🤣 so far no one have too positive experience from the email support

in term of waiting time

[knotty grotto]

No we are a team

[knotty grotto]

Yeah unfortunately they didn't expect that popularity

And didn't expect to make world record

[narrow cobalt]

What is it that causes such a big speed difference between output of GPT-3/3.5 Turbo and GPT-4?

[boreal ingot]

One trillion parameters difference

[runic plume]

Are they ever gonna fix the issue where hundreds of user’s Plus subscriptions have disappeared after their last recurring payment?

Seeing so many folks that paid for Plus and are stuck using the free tier for the past few weeks. No word about refunds/extensions of subscriptions.

Emails to support and messaging the bot don’t work either.

[narrow cobalt]

Wrong channel

[mystic nest]

dont think there's any channel for that 🤣

[icy sapphire]

Gpt 4

[mystic nest]

openai employee probably would hide their identity even if they are in discord or they will get flooded

[light cradle]

hello gpt-4

[slender pivot]

I don't suppose we have any idea when the 25 message/3 hour rule might be raised

[knotty grotto]

They appear in yellow their name

They don't accept DM

[narrow cobalt]

Any other thoughts on the speed output of GPT4? Doesn’t seem like parameter size alone is responsible.

[knotty grotto]

Please stop spamming every channel

[stone jackal]

I don’t know where to ask

[knotty grotto]

One time is enough

[stone jackal]

One time where is good?

[supple violet]

it's up to you to pick the channel where you think you get the best answer, gpt4 is not one of them

[stone jackal]

Why gpt4 no good?

[supple violet]

well for example because it a channel for a specific model, not related to development which you need to do to make a passive source for income

[stone jackal]

Oh ok

[supple violet]

gpt-4 is also the most expensive model and overkill to many use cases

[stone jackal]

Overkill?

[boreal ingot]

This.

3/4 of the users don’t need gpt4 and plus is more than enough for most of us

[wise furnace]

I generally use gpt-3.5-turbo to test my prompts and then switch over to gpt-4.

[whole kettle]

Ergh, GPT 4 just got dumbed down again, I suppose it's based on server stress.

[pallid mantle]

I do not understand if gpt is dumber or not...sometimes it appear and sometimes not...as an example: for vba code it wrongs a lot for javascript less...may be it has been trained more on javascript...

[supple violet]

3.5-turbo can do a lot, gpt-4 is better but not worth the 20-30x more expensive cost for many tasks

[stone jackal]

Chat gpt 4 is same plus?

[pallid mantle]

then I try asking a question about holographic universe to gpt 4 vs gpt 3.5 it seams that gpt 4 is more specific and thinking about it

[mystic nest]

for most ppl, if you pay a fixed subscription fee, you would want the best out of it 🤣 only api users feel the pain

[supple violet]

?

[wise furnace]

Honestly, I have yet to see much of a difference between 3.5-turbo and 4 for my admittedly simple uses.

I’m still in “how to use it” mode. Not “get stuff done with it” mode.

[stone jackal]

Upgrade to plus … it is chatgpt4?

[supple violet]

with plus you can do only 25 messages per 3h with gpt4

[boreal ingot]

GPT models aren’t “dumb down” stop using this formatting. The prompts YOU the user used for the task is what “dumb down” the model.

Use better prompts, Never become lazy and always give context you shouldn’t see any difference in the way the chatbot talk to you now from last week

[whole kettle]

I'm using GPT 4 for a few weeks, working with it non stop, either I became dumber either the bot.

[stone jackal]

Have another way for do chatgpt 4?

[supple violet]

only api which require waitlist approval

[whole kettle]

Also, I can show you my prompts, I can assure you that they are anyhting but lazy. It worked very well until recently, I know a difference when I see it.

Perhaps my code became too complex for it, that could be the issue.

[pallid mantle]

Asking small pieces of code works, then is up to you put them together...apart for VBA as I said

[boreal ingot]

Are you using directly in the chatgpt chatbox or is it implemented into an app at the moment ?

Do you use different instances each time or are you trying to stay in the same instance for a while ? if so try using a new chatgpt instance to copy past much of the conversation and ask him to summarize it in about 300 character. Than open a brand new instance copy past the summary and ask the questions again. Can you see any difference at all ? Does it seems like the reply is more complete than what you had ? If so it was indeed a memory issue

[whole kettle]

I did open quite a few new instances today, at some point it started working normally again but then it stopped working again even with new instances. I mean it has issues even continuing the code snippets now, which it didn't had beforehand (and I had it write big chunks of code before for the same project).

I'm thinking maybe it's because i have a huge number of instances perhaps that affects it ? Should I delete some ?

[boreal ingot]

It could indeed be one of the reason maybe trying to stay between 10-15 instances max. More studies are needed about what affect the workflow on the website but it’s very probable too many chat is one of them

[whole kettle]

I am just using the chatbox by the way, no API or anything like that (since I didn't get access to the API) and playground only works with 3.5 for me at the moment it seems.

Yeah well is there a way to download all the conversations before deleting them ?

I mean, I have alot of stuff about the project in there and don't want to just delete it, may need it later down the line.

[boreal ingot]

Make a txt file a manually copy past it. I suggest doing this right now as you never know when you can get an error and other users have reported loosing all their past chat after the website encounter an error

[untold stirrup]

How much RMB can I exchange for one dollar

[knotty grotto]

#off-topic

[fluid night]

Is your gpt working properly?

[lilac shard]

Settings, Export Data

[fluid night]

My page has been loading.

[whole kettle]

Life saver, thank you.

[fleet ice]

Best way I have found is to just keep copy pasting Q/A in a word document

[fleet ice]

Is this for individual chats?

[boreal ingot]

It sent a complete chat history of all your active instance via email. It can take sometimes

[whole kettle]

It exports all chats in one click of a button, it's amazing.

[fleet ice]

Hmm…interesting; I have got tons of it; need to try what will google do with such a huge history email

[native mirage]

how to use this

[wraith ginkgo]

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

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

[fair iron]

💬 Chat with ChatGPT at https://chat.openai.com/
🤖 Try out the API Playground at https://platform.openai.com/
🎨 Make art with DALL∙E 2 via the API or in https://labs.openai.com/
📩 Read FAQs and reach customer support through https://help.openai.com/

[sleek sun]

i have 2k assignment, can gpt4 does that

[unborn ermine]

YES

It can do a whole lot of things

[sand hull]

It can tell you how to do the assignment, and while you can try make it do your assignment you'd probably fail the exams as a result, so maybe do the assignment yourself.

[sleek sun]

does turnitin now can check it

[unborn ermine]

That was not helpful

[sand hull]

Turnitin, GPTZero, AI text classifier, they can detect fairly well though not perfect. End of the day it's still plagarism and academic dishonesty and one wrong move gets you expelled so maybe just do your assignment like any other student.

One of the most complex tools ever created, unlocking nearly infinite possibilities. And yet people instinctively go "how can I make this thing do my homework so I can fail the exam later".

[grand ibex]

dam

they keep making there models dumber

its annoying

they make so much money from plus and API and they still cant scale correctly

[feral slate]

Hi everyone! I'm new here so please let me know if this is the wrong channel for this question.

I am using GPT-4 in a tool designed to assist the most active contributors/moderators of a community forum in answering user questions. In some cases, the user's issue cannot be solved by the community and they have to open a support ticket. From my testing, GPT-4 is by default pretty good at figuring out when this is necessary.

However, the process of opening a support ticket changed recently (after the knowledge cutoff in late 2021), so the generated reply will contain incorrect steps. If I include a system message with the correct steps to contacting customer support, GPT includes some variation of "If you are still experiencing issues, contact customer support by ..." at the bottom of every single response, instead of only when relevant as it did previously.

GPT's suggested replies are not shown publicly, so we can simply ignore irrelevant parts when writing a reply to the user, but I wonder if there is a way to provide additional context to GPT without that context becoming the solution to every problem?

[feral lance]

So uh, that lowered usage cap was a pipe dream eh? I noticed they removed the message of expecting one 😉

[dim chasm]

Gpt4 can do assignments perfectly , people were passing with gpt3, same goes with gpt4

[sand hull]

Doesn't mean it can't get detected, also doesn't mean you should.

[native flare]

how much gpt-4-32k?

[gritty atlas]

is this free to use?

[knotty grotto]

no

[unique karma]

I don’t understand why they capped gpt4 at 25 messages when I’m paying and there are now other websites offering gpt4 for free

The cap is frustrating

[knotty grotto]

1_ gpt 4 from other service isnt good as openai
2_ they still keeping the cap cause they are too much trafic and user

[unique karma]

I see, but I hope they will increase the cap in the future

[short musk]

How do you know it isn't as good?

[knotty grotto]

try it yourself and compare

[maiden gull]

Does anybody else feel like GPT-4 has been dumbed down? It has been missing some simple stuff

[mystic nest]

hard to measure or pin point if "dumb down" happened since it's generative in nature, so everyone mileage may vary

[mental burrow]

You could preprompt the model with new info. You could also try embeddings, but I'm not experienced with them

[jade oasis]

Chat GPT for IPhone or MacBook Anyone have positive feedback? Even with alternatives?

[mystic nest]

officially, there's only web chatgpt, so doesn't matter which device/os

you mean api or the usual chat bot? if latter, just sign up "plus" paid subscription

[jade oasis]

I know people are still using it and paying the fee but I am starting to see several other bots pop up and I figured I’d reach out and ask you all

LLM’s or actually you could use BING

[stable dagger]

OpenAI team - how do I file a formal complaint about the GPT4 message limit. I am paying for access and it has been limited for a month. GPT3.5 is so inferior, it is impossible to work with. And when the conversation is switched to 3.5, you cant switch back to 4.

GPT3.5 sucks so bad at writing code.

[mystic nest]

good luck contacting the support 🤣 you can try email them

[jade oasis]

Open AI technically is no longer open. As mentioned, good luck.

[stable dagger]

It's ridiculous to be paying for access and still not get it

[jade oasis]

Do you think they’d have a bot? Lol.

[feral slate]

Thanks! I'm already sending a system message (which is where I write the correct steps for contacting support) before the user's message - is this what you mean by pre-prompt? The problem is it causes GPT to think that contacting support is the answer to every question, which it did not do before.

I've also been experimenting with embeddings and vector search to find documentation relevant to the user's question and include it as context, but there's the same issue of GPT wanting to use that context in its reply even if it's irrelevant to the post.

[stable dagger]

Well we could all cancel our Plus subscriptions. That would probably get their attention.

[jade oasis]

Typo- you’d think they’d have a bot 😆

[mystic nest]

it will hardly work imo, you wouldn't gather enough ppl to cancel

they have like 140m users

[stable dagger]

I have an API key, what service replaces chat.openai.com ... is there one?

[mystic nest]

so you need to garner like 10m~30m?

[jade oasis]

Again, there are no longer open. Private company/for profit I am not surprised. However, do you have access to chat GPT right now?

They

[stable dagger]

I tried Hugging Face / Chat-GPT-LangChain but it cant print code and its answers were too short and not the same as using chat.openai.com

[mystic nest]

things may change, but chatgpt still the top chat bot for now and they are suffering from success 🤣

[jade oasis]

WOW! That was quick! So there are bots here

[mental burrow]

Have fun with that lol

[jade oasis]

No one quit paying midjourney during their early days

[gaunt cargo]

#gpt-models hello

[covert lily]

how

how can I use chatgpt?

[serene cipher]

you can use it here: https://chat.openai.com/

[unborn pulsar]

翻译以下文字：So uh, that lowered usage cap was a pipe dream eh? I noticed they removed the message of expecting one

[covert lily]

access denied already 😫 😫 😫

[serene cipher]

@unborn pulsarOAI does not offer ChatGPT on this discord

[serene cipher]

that usually happens when you are under a VPN

[unborn pulsar]

thx

[leaden cape]

i can access chatgpt on a VPN 😎

[knotty grotto]

Vpn can create bug sooner or later. Just keep that in mind

[leaden cape]

how

[knotty grotto]

It can trigger the security from openai and create a bug or access denied

[leaden cape]

pretty sure the "security" is a cloudflare ruleset

[lost viper]

Is it safe to use chatgpt for research paper?

[mystic nest]

safe from plagiarism? no

it can help you to find sources and references

you can search in youtube for more details

there are some tutorials on how to work with chatgpt for research paper purpose

[serene cipher]

VPNs are based on the premises that multiple users take the same public IP address, which causes the rate limiting to be shared among all the users on the same IP, this spreads the limits very thin

[tight wave]

Can anyone tell me any ai for ppt

[sand hull]

Maybe look in #1037561385070112779 and see if there's one you like.

[serene cipher]

if you are use some sort of private self hosted VPN, it would work normally

[wind urchin]

hemlo

[leaden cape]

yeah thats what i do 😄

[serene cipher]

Cloudflare still doesn't appreciate VPNs tho

[silver gyro]

Is gpt-4 get access of internet??

[serene cipher]

it might give you extra capchas to solve

[serene cipher]

no, this is not how the AI works, but you can always feed it data from the internet and let it interpret it just like any other text

[wraith ginkgo]

`` Server Tip: ``

Add your well-crafted prompts to our #1019652163640762428,
or share your interactions with ChatGPT in #1050184247920562316!

[round mural]

how to use gpt 4 for free

[grand ibex]

no official way

[round mural]

ok

[quartz trout]

how to explain wisdom

[weary dome]

Is poe inferior to openai because it doesn’t have multiple chats? Does it like any other features? I’m considering subscribing to Poe because I’m not in a country that can access openai (Hong Kong)

[mystic nest]

I thought poe is just an interface to access different sources of chat bots?

[oak adder]

[wind urchin]

Surprising as it may seem... this is the first time in my life that I feel overtaken by technology, for not understanding its transcendence.

If I wanted to create a digital driving school, how would ChatGTP4 help me TODAY? I would have to create the content, the tests, the digital platform, the apps, the payment gateway, the business model, the operations, etc.

[mystic nest]

how about doing everything 20%~1000% faster?

also of cos, depends if you have the team to do it or just outsourcing

[limber frost]

what is about LLaMa +chat gpt 4 issue

[noble sable]

My vpn is always turned on. I didnt experience any limitations by that. Also I never had to solve any captchas.

[wraith ginkgo]

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

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

[noble sable]

How do you know? I can't find any public information about the Number of paid users of ChatGPT

[errant shard]

It seems to have the option to delete your account next to the export data. Does this imply that deleting your account deletes all your chats?

[wind urchin]

So it really kind of helps programming? If it speeds up production so much, it should also reduce costs so much, even if they are outsourced, right?

[tacit steeple]

Do you know GPT4 dataset look like?

[wind garnet]

cIAO

[wind urchin]

Sal

[mystic nest]

you can search for something like "chatgpt users count". anyway it's an estimation

[noble sable]

Thanks for your reply. Yes indeed the number of users are estimated in that range. Most of them should be free users though, dont you think?

[vestal fog]

#question hao nice

[mystic nest]

maybe it's a typo 🤣

[wraith ginkgo]

This is a Safe for Work Discord Server

The OpenAI Discord is an actively moderated server.
• Refrain from sharing inappropriate content on the server. This includes but is not limited to messages, media, or other topics of graphically violent, sexual nature, and drug-related content.
• Report all sensitive and offensive content in the feedback reporting tool in the ChatGPT web UI instead of here on Discord.

Posting NSFW content will result in a permanent ban.

[mystic nest]

I edited it

estimated paid users should be like 10~20% of that base

[snow geyser]

what do you guys think chat gpt4 will do? 😬

[low thorn]

Hei, any word on increasing the 25 message limit ? Its driving me crazy

[mystic nest]

dont think outsource will charge you less but it can be used more than programming, like content generation, write up, instruction/technical manual etc, it's only limited by our imagination

[noble sable]

Do you have the Link to the article you're referring to? 🙂

[mystic nest]

cant link anything here