gpt-models

[somber dagger]

is there info about adding more messagages to gpt4?

[rustic wagon]

👀 My bad.

[edgy latch]

i learned how to drive by playing driving games

[fleet dove]

the topic is far gone at this point

[wind urchin]

well, lets fix that

[edgy latch]

With a controller..

[fleet dove]

You can still get access to the API even if you dont have the plus subscription

Still gotta pay up tho :(

[edgy latch]

the topic is on point cus all my answers come from GPT.

😈

[whole kettle]

Oh Gosh

[fleet dove]

clearly multiple parallel dimensions ahead

[naive lynx]

if you go to notionsmith dot com you can see what ChatGPT thinks about its own idea

[edgy latch]

when mind uploading

[edgy latch]

u tryna send me to some site thatll hack me

[naive lynx]

okay.

[peak hamlet]

Hello everyone, is ChatGPT (3.5 or 4) is able to write me a motivation letter and a CV for an IT enterprise ? with maybe adding some IT picture etc..

[edgy latch]

it can do more then that

[whole kettle]

And that’s where we come in, with GPT Protect we got you covered 😆

[edgy latch]

will GPT be able to decipher hierglyphs?

hieroglyphs

[naive lynx]

Sometimes I forget this is discord, not slack, so the average age is probably 10

[peak hamlet]

Like adding picture to the motivation letter ? or do a cool one ?

[edgy latch]

thats why your here right

[naive lynx]

they think someone needs to hack all $20 they saved up from christmas

[naive lynx]

yes exactly, I forgot

[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.

[fleet dove]

first time?

[naive lynx]

ha not the first and sadly may not be the last

between "your hear" and "peasent" I think I'll save the click.

[edgy latch]

🙈 good boy

lmfao aboutme: check out notionsmith Fool was tryna self promo

fko outta here clown

[mystic girder]

How does Infinite Seinfeld keep running 24/7 GPT-4 on twitch then? (NothingForever)

[wind urchin]

if you really want to push it, ask for it in klingon, in the style of a musical, with a sub plot about bacon

[blissful crater]

Whats good guys

[wind urchin]

bank holiday monday and coding AI

[earnest flame]

Is there any benifit of using gpt4-0314 instead of the normal gpt4?

[wind urchin]

as an endpoint?

[earnest flame]

As the model

[round rain]

gpt4-0314 has the power of pi

[wind urchin]

well, if you just use the generic, then when the model updates you get auto updated, but if you have some legacy application that you don't want anything to change on then you'd use the versioned one

[earnest flame]

What do you mean?

[earnest flame]

I see. So for the most accurate information and results, i'd just use the normal one?

[round rain]

pi = 3.14...

[wind urchin]

if your application is not tied to a particular result style... you can use the generic, if you have promised a client that there will be no updates for an agreed period, or you need to keep things the same ... you use the versioned

[round rain]

make dinner with it

[supple violet]

who says it runs on gpt4, the most expensive model? many of the twitch ai stuff have been running open sources gpt projects and not 3.5/4

[wind urchin]

wait... .. do you mean to say that a streamer might have just used the current buzz words in their title and not actually used the thing... shocked, I am shocked

it's very hard to tell 3.5 from 4 if you just use natural language and are in "story mode" I doubt that stream goes into complex coding tasks

[copper dock]

hi, how long does it take to get plugins access?

[wind urchin]

better tell ferrari they can't make faster cars, then, or dewalt they can't make a better drill.. someone might use them for harm

[slim swift]

thats true for almost everything in the world

[zealous acorn]

Man the "AI" clickbait is getting worse and worse on YouTube.

Dyno timed me out for using a few capital letters 😦

[mystic girder]

As another example, AIPD runs another AI show, this one I know for sure runs on GPT-4, but I feel like the API limit is high enough you will never hit it

[dusky iron]

adding pictures heck no, text, Totally.
You Can add pictures using hugging face with your own api.

[zealous acorn]

Does a request rate limit even exist on 3.5 anymore?

It seems I can request as fast as I can type

[mystic girder]

There is always a limit but its as high as a human needs

[dusky iron]

if there is a limit its most likely on non + users.

[mystic girder]

Now if you run 20 scripts making API calls thats another story

[zealous acorn]

Ye tru

[mental burrow]

👀👀

[dusky iron]

not that difficult tbh u can get a guide through gpt xD, it's neato

[zealous acorn]

Imma just run "list 10 fruits" on a think hook and see what it does :3

[mystic girder]

It will likely list fruits, I reckon

[zealous acorn]

A fair assumption

But it will list 10 fruits... once per rendered frame 👀

[mystic girder]

I just want to make a plugin to plug GPT-4 into Roll20.

Works manually but would be cool to have the process automated, hmm.

[zealous acorn]

I want GPT-4 so I can incorporate it into VRChat or Garry's Mod

Because why tf not 😄

[delicate island]

Hello there

[zealous acorn]

👋

[mystic girder]

You can already do that with text to speech GPT-4 apps haha

[icy delta]

Where do i go for that

[mystic girder]

No idea gotta check your local app store, varies by country

I've only seen youtubers demonstrating it

[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.

[icy delta]

Whats VCR chat???

[zealous acorn]

I saw an AI vtuber, pretty neato

[mystic girder]

VRchat is a steam game

[edgy barn]

when plugins ? 😦

[icy delta]

Ok honey hanks

[floral egret]

can u guys plz stop using chatgpt i need it to lawn my grass

[zealous acorn]

👀

[covert rose]

i just started using chagpt today but i keep getting the "something is wrong " message

[hot blaze]

the image recognition feature , has anyone used it? when is it available?

[covert rose]

what can i do to fix this?

[zealous acorn]

It won't be available for regular consumers for some time

[thorny valley]

So I looked up Connor Sturgeon on Bing Chat and Bard, 2 different answers. In a way Bing is more recent as Bard says nothing about the recent event.

[hot blaze]

and for developer?

[rustic wagon]

It just means that chatGPT is becoming self aware. That’s it externalising its existential dread

[thorny valley]

basically if you are doing some advanced research, just prompt the same prompt on Bing Chat and Bard

[mystic girder]

Check out Claude/Claude Instant too, its my main model for day to day use

[thorny valley]

Claude?

[mystic girder]

Its available on the poe website

[thorny valley]

link

[mystic girder]

Interesting to compare with gpt-4

[copper dock]

hi, how long does it take to get access to plugins?

[zealous acorn]

At my convention this weekend, someone used GPT-4 to make an anime "Who's Line Is It Anyway" skit. I don't think I've laughed so hard in a while

[supple violet]

there is no queue or guaranteed access

[thorny valley]

anthropic early access?

[copper dock]

there is a waitlist

[mystic girder]

Yeah

[supple violet]

yeah but its not first in first served, they look at your use case, past history, location etc., only small number of plus users are invited

[thorny valley]

nice

[loud dove]

Are any of y’all having api connection issues today?

[zealous acorn]

Well I mean, it's a brand new product, gotta prioritize those that have a beneficial use case

[crude dew]

Question — is it normal for people on the GPT 4 API waitlist to still not have access? I signed up for the waitlist on the first day it was announced and still haven't gotten an email.

[supple violet]

yes its normal

[mystic girder]

Your use case might not be what OpenAI is looking for

[crude dew]

Got it thanks

[mystic girder]

Can you imagine how much GPT processing power is wasted on memes and stuff?

[crude dew]

I'm a developer, so I wouldn't be wasting on memes, but yeah I hear your point

[zealous acorn]

In time, we all will get it. But just give them a month or two. We still are even figuring out 4's capabilities

[wind urchin]

how use midjourney

[hot blaze]

@dusty geyser How do I ask the bot?

[wind urchin]

?

[supple violet]

go to midjourney server

[mystic girder]

As an aside, give stable diffusion a try if you have a gaming gpu, or dalle2 if you wanna go the OpenAI route

[zealous acorn]

SD is fantastic on a 3090

[wind urchin]

hey what's the model name for gpt-4

in the api

[mystic girder]

Thats the model name

Or is it a trick question

[left ocean]

I asked gpt-4 to generate responses to the waitlist questions that would maximize my chances of gaining access off the waitlist, lol. We will see how that goes.

[wind urchin]

nah like e.g. "model": "text-davinci-002", is the model name for gpt 3

[loud dove]

stable diffusion?

[zealous acorn]

gpt-4

[pseudo topaz]

its just gpt-4

[loud dove]

it is literally just "gpt-4"

[pseudo topaz]

cant wait to try it on a 4080 😆

[zealous acorn]

Ye

[pseudo topaz]

3090 does have more vram, though

[loud dove]

have you used x formers with it? it makes it so much more efficient

[zealous acorn]

Image finished in 3 secs lel

[wind urchin]

oh weird cause gpt-3's model name is text-davinci-002

[pseudo topaz]

i remember in the early stages of text to image, like 3 or so years ago, it would take HOURS for a single prompt to render, even with the best card of its time (1080ti lol)

[zealous acorn]

Should be gpt-3.5-turbo in API calls

[wind urchin]

that's confusing

[pseudo topaz]

absolutely amazing how far its come, being able to be ran in the browser and complete in 30 seconds

[loud dove]

Gpt3 is a lot more than text davinci, it’s a large amount of modes with 4 base text models from Ada to davinci, while gpt4 has only 1 base model

[vocal pumice]

And then you got a 128x128 image of a deformed dog/sea creature hybrid?

[zealous acorn]

You can have the API list all of the model names if that helps ^^

[wind urchin]

Ahh

[wind urchin]

oh yeah that would be cool!

[pseudo topaz]

pretty much lmaoooo. was pretty cool for its time though ¯_(ツ)_/¯

[loud dove]

Y’all remember the og chat bots? Like eve and stuff?

Or clever

[mystic girder]

Cleverbot enters the chat

[pseudo topaz]

yeah

mathways chatbot was ahead of its time

[mystic girder]

Markov chains were a pretty cool stopgap solution

[pseudo topaz]

i still use simple markov chains today for data generation

[loud dove]

Crazy how much just the existence of transformers has changed the capabilities of text generation

[lost niche]

was this answerred?

[pseudo topaz]

that question doesnt really make sense, elaborate?

[soft marten]

sheesh

[mystic girder]

There is a button saying use default model, right where you have the timer

Or you can start a new thread using 3.5

[zealous acorn]

I think they mean switching from one model to another, in the same conversation

I dont think you can do that within the web UI no

[mystic girder]

Thats the default model button

But once you switch to 3.5 you cant go back to 4 (In the same conversation)

[zealous acorn]

😮

[lost niche]

thanks.

[icy obsidian]

[pseudo topaz]

vice versa also

whatever model you start with, will remain the entire thread

[zealous acorn]

I wonder, is Cleverbot a dead project?

[mystic girder]

Google it? Seems to still work last I checked.

Wonder how it has shaped up after learning from users for like 10 years

[loud dove]

Clever bot api is no longer free

[pseudo topaz]

rip

[loud dove]

I think there’s still a free website tho

[ember mauve]

hey guys, I have pro but still waiting 1 hour? how can it be possible?

[mystic girder]

Welcome to the club pal

We still have to wait

[zealous acorn]

If anything, I have learned a lot about software development and web APIs from OpenAI 😄

[mystic girder]

All I have learned is that the magic computer oracle will write stuff for me and thats enough.

[zealous acorn]

it writes some cleeeean code if you prompt it right.

[cursive heron]

t3nsor.tech

[pseudo topaz]

just looks like a chatgpt clone

[spice anvil]

[pseudo topaz]

👋

[cursive heron]

is gpt3.5

[hasty egret]

[fleet grove]

Hey, I noticed that starting today, OriginalityAI has started labeling articles that were previously identified as 90%+ human-written as 100% AI-written.

Has anyone else experienced this? What do you currently use for rewriting?

I used to rephrase texts with WordAI.

[pseudo topaz]

well most ai detectors are iffy at best...

[meager bloom]

I think it's kind of stupid to ask here but.... my dog asked to ask if it's okay to share the account with a friend to share the amount with him or it's a problem

[mystic girder]

Yeah dont trust AI detectors, plenty of legit essays getting caught in them

[fleet grove]

Yes, and yet, just yesterday when checking through different detectors, the result was 90%+, and now the same texts have dropped to 10%

[mystic girder]

As long as said friend is part of your organization there is no problem

OpenAI accounts are meant to be used by the entire organization

So the admin can gauge API usage costs

[meager bloom]

I'm premium, to access the visual part instead of the api, do I have to share the account or just add the member, does it also become a plus?

[remote ferry]

I suggested a format to ChatGPT and it's using that now. How can I make it highlight a certain line in this suggested format?

[mystic girder]

Plus users don't get access to the visual component, that is a separate waitlist afaik

[meager bloom]

thank you very much

[fleet dove]

Visual capabilities have a waitlist? Link?

[stark trail]

oops!

[jovial fractal]

can anyone give me an example of how to access the other "models" within chat gpt 4 to break it past the language setting so it gives legitimate business help

[mystic girder]

No idea never searched for it myself, I just know that plus users dont have it

[mystic girder]

This kind of thing is against the server rules and you'll get banned if you discuss it here.

[jovial fractal]

oh i apologize, i meant only educationally but thank you

[rugged dawn]

Anyone with gpt 4 wanna help me with an assignment plz lmao 🤲

[naive lynx]

check out my post in API Projects, I worked on that exact problem

[slate dragon]

hmu

[naive lynx]

(or at least, I think it's your exact problem, if you try it and it is, let me know and we can chat about how I did it. You can describe things like "a death ray" and get legitimate business insights)

[gentle portal]

The best ia

[floral vault]

Leave letter

[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.

[wispy trellis]

[balmy star]

finished gpt4 tokens but urgently need it, is there anything i could do? gpt3 is rather stupid lol

[lusty nebula]

buy more

or use unofficial api

[balmy star]

dont think thats possible

[balmy star]

how

[lusty nebula]

You mean the API or chatgpt?

What are you using

[balmy star]

gpt

[lean bloom]

finally bought plus yesterday xD

[lusty nebula]

If you're rate limited you can make another account and buy subscription but that's about it

as far as I'm concerned

[shrewd citrus]

That is against ToS from what I understand

[sly axle]

👏📌

[wind urchin]

when next update

[ashen scarab]

do people that do not do evals still get approved on gpt4 api Whitelist? If so can anyone Who got approved tell how much they have waited?

[limber garnet]

You just have to fill out a form

And wait a bit

[ashen scarab]

how much would that bit be? because I have a project I have to complete

[mystic girder]

Could be that you selected a use case OpenAI doesnt have as a priority

[ashen scarab]

it's not been that much time i Just Need to know around how much

[wind urchin]

I got mine after 3 days, some have waited 2 weeks, some are still waiting

all depends on your use case and what you put in, plus an element of position in the queue

I'm still witing for Plugins, so clearly I'm not in their use case top 10 yet

[mystic warren]

Hi can you help me , i want to connect the gpt with messenger

[wind urchin]

just ask ChatGPT how to do it

[mystic warren]

asked him but he gave me a general steps

[mystic girder]

This is the new "Just Google it bro"

[wind urchin]

and then you ask the AI about details for each of those steps

it's not yet at the level where it will just create the entire thing for you, there is still a lot of design and effort that needs to be put in

[wind urchin]

Tru dat

[hasty moss]

Any good gpt links?

[serene cipher]

https://chat.openai.com/chat
this is a good one :V

[mystic girder]

Explain

[wind urchin]

Did everyone sub to Plus?

[wind urchin]

I meant to put the laugh on lugui, not you, sorry

[hasty moss]

learning resources

[mystic warren]

ok can i do it with zapier ?

or its not working

[icy oracle]

where i can get mod gpt

[radiant parcel]

Hey, new here. What is the fastest way I can obtain full access to GPT 4, I am currently on the waitlist

[icy oracle]

so that i can access freely gpt 4

[wind urchin]

🤔 really you can learn a lot by reading the documentation and just searching online, some youtube stuff is also good, but try and avoid the click bait stuff

[hasty moss]

Too much click bait

[icy oracle]

i can help

[cursive breach]

#5840

[icy oracle]

u pay i ll give you now

[inner trail]

Yeah i subbed only for Legacy (GPT-3.5)

[wind urchin]

no, you won't, and stop saying stuff like that

[icy oracle]

humor buds

i am kidding

its just jok not a dik

take it litely

[wind urchin]

not funny, when people are new and not sure what is what.

[icy oracle]

even gpt 4 is new

[hasty moss]

Click-bait?

[simple aurora]

how long does it take to get gpt-4api key?

[serene cipher]

You wouldn't believe on how good this link is

[wide bolt]

Weeks to months

[hasty moss]

Just count the seconds

[simple aurora]

haha

[wind urchin]

there is no set time, if your use case fits with what's being explored, then you will get a key, if not you have to wait for the next round of keys to be given out

[quartz locust]

Hi which channel can I write a prompt here?

[wind urchin]

you can't chat with ChatGPT on this server, you need to go to https://chat.openai.com

[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]

this one cool trick on https://chat.openai.com

[radiant parcel]

I have been on the waitlist for a month or so

[simple aurora]

still didn't get api key?

[radiant parcel]

No

[simple aurora]

I meant gpt-4 api key

[dapper hinge]

Good ol' Bill Gates live right now talking about GPT-4

[mystic girder]

I have been on the waitlist ever since it opened and didnt get an invite, I even attached my press credentials but it seems they dont like that

[rustic pebble]

I paid for it, but I still get the FREE version. I have emails not answered from 20 days ago. They added now the possibility to send the request by Chat, but I have a 7 days chat without any answer. Does anyone here already got an answer from the chat? How long it took?

[true acorn]

gays anybody success y et making mooney with chadgpt 4? teach how

[radiant parcel]

What is an API key used for?

[limber garnet]

🏳️‍🌈

[simple aurora]

launch mvp, start marketing, get business success, etc

[simple aurora]

to get access to gpt-4 in application

[buoyant canopy]

There's a talk by Sebastien Bubeck (Researcher at Microsoft) where he talks about the more interesting points of the Sparks of AGI paper.
Anybody here watch that yet?
(Title is "Sparks of AGI: early experiments with GPT-4" on YT)

[dreamy rover]

Chad GPT will surely earn you money if you’re creative enough

[simple aurora]

👍

[dapper hinge]

Then what's making you money is not ChatGPT but your creativity.... LOL

[mystic girder]

If you can leverage ChatGPT to make money you are smart enough to do it the normal way

[dapper hinge]

exactly my point

[tidal musk]

If i upgrade to plus will i get access to gpt-4 and will the 2048 characters per response increase ?

[radiant parcel]

I’m using chatGPT as an enhancement for my current idea, my startup is going to change the world

[simple aurora]

what is your main idea using chatgpt to success?

[dapper hinge]

summarize it in one line, just to get a sense of it

[radiant parcel]

It’s for men 18-23

That’s all I got for u

[dapper hinge]

thats not a summary, is only the demogrphics lol

[simple aurora]

Nowdays, I am working on the project to train chatgpt from pdf files.

[radiant parcel]

I cant risk someone taking the idea, it has never been done before and no comp

[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.

[dapper hinge]

good luck, then

[simple aurora]

Let's feel free to share our ideas and grow together!

[mystic girder]

This phrase has been said so many times it has lost all meaning

[dapper hinge]

are you a beta tester?

[simple aurora]

I am machine learning web developer

[radiant parcel]

I’m not the boy who cried wolf

[dapper hinge]

that's the OpenAI spirit, I believe. Knowledge will be more and more available to everyone.

[simple aurora]

Yes, It is better for us to grow together than to grow only myself, lol

[buoyant canopy]

Are you trying jailbreaking prompts on that dude? 😂

[dapper hinge]

haha

In the old days we called it "social engineering"....

[inner trail]

Isn't OSINT the same?

[buoyant canopy]

OSINT is using "open" (publicly accessible) sources, like say satellite photos from Google Maps, to gather intelligence info.

[dapper hinge]

right

[inner trail]

Thank you for answer

[buoyant canopy]

What he's trying would probably be HUMINT.

[simple aurora]

I have friends that are looking for work as developer, if you need to hire, you can use them .

[dreamy rover]

chatgpt is muh developer, human is only there for debugging

[icy oracle]

[dense sundial]

can the ability to update the title of chats be added? it is a problem when conversation transitions and focuses on subtopic. i asked gpt4 on chatgpt to do so and it said it could not

[buoyant canopy]

You talking about ChatGPT? Don't you have that little pencil icon when you klick on the chat description?

[dense sundial]

🤦‍♂️ youre right, it’s just hidden on mobile. thank you

[frank rivet]

How to get an API response FASTER? My prompts request responses with thousands of tokens. The responses often take one or even several minutes before they return. This is using GPT4 API calls with a pro account. Is there a way to get it to happen faster, or a pro-er account? Thanks

[hasty egret]

no

[buoyant canopy]

Use GPT-3.5-turbo. 🤷

[fossil lotus]

looks so exciting

[icy oracle]

[long holly]

can gpt 4 be used for cybersecurity purposes?

it seems one could use it in the malicious and protective sense

seeking out vulnerabilities seems like a rudimentary application

[icy oracle]

Yes I am working on making its mod version

Dm me if you are interested

[crystal lichen]

Is the data model "gpt-4" still in beta? If so, anyone in here ran into some limitation issues? More specifically in translation from English and into another language? Anyone also experiencing slow complete response from a prompt with "gpt-4" data model?

[long holly]

im worried that an "AI-based" virus would change its signature and evade typical malware detection

[fleet umbra]

can gpt 4 bypass turnitin??

[crystal lichen]

good point.

[long holly]

AI viruses would be absolutely devastating because we have no defense against these things

[buoyant canopy]

You mean like regular malware developers already do right now? How would that be different when an AI did it?

[fleet umbra]

can gpt 4 bypass turnitin??

[long holly]

an AI could do an API call and use GPT-4 to evade malware detection on specific devices.

[crystal lichen]

I'm sure someone with figure this out eventually. After all, hackers with bad intentions, spend their days figuring these things out.

[icy oracle]

Calm down cyber boy

[icy oracle]

Bad good is just perception

[hasty egret]

still the same old game that would be, I guess.

[icy oracle]

Bad for u good for them

[buoyant canopy]

Yeah. Sure. It will help them do their "job" better.

[runic notch]

yea 🥸

[long holly]

i should note that we don't have any pre-existing AI-based malware, and you don't even see that sort of thing in biology. if the AI was intelligent enough it would be able to hack any system without a defensive AI that monitors operations on a computer.

i think cybersecurity might be AI-based in the future for that reason in a non-encrypted domain

[buoyant canopy]

Nice SciFi threat scenario you have there. But it probably won't work like that.

[true furnace]

Anyone knows if GPT-4 API available for use.

[icy oracle]

AI is just fancy words
It's too a bunch of program

[wind urchin]

Can someone pay me gpt4 and I'll give her 20€ with paypal ??

I dont have credit card

[white mirage]

I'm not quite sure what your point is here.

[buoyant canopy]

Profile checks out bro. 😂

[hasty egret]

artificial intelligence, have been used for long time, always changed its definition over time.

[wind urchin]

I can pay with paypal ??

[white mirage]

Yes no I understood that. I don't understand the point in relation to what they were responding to.

[buoyant canopy]

Not sure if that is even legal.

[wind urchin]

it's legal

[long holly]

i sincerely doubt that today's computer architecture and the structure of our internet will survive an AI-based virus, after all social media is already getting swarmed with bots.

[junior salmon]

Unfortunately, this is against the Terms of Use.

[wind urchin]

I can't pay with PayPal

[wind urchin]

but it's necessary

[junior salmon]

Why is that?

[buoyant canopy]

I probably get totally wrong what you mean by "AI-based virus". But our tech seems to be pretty resilient against the threats we face today.
Your scenario looks a lot like all the AGI escape SciFi stories we get fed nowadays from all sides.

[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.

[dense scaffold]

Lethal Autonomous Weapon.

[wind urchin]

AI based anti-virus 🤔

finally, a system that actually knows what humans mean by "unusual activity"

[buoyant canopy]

😂

Imagine what a full system scan would cost using GPT-4. 💸 💸 💸

[wraith ginkgo]

`` Server Tip: ``

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

[river swift]

You’re a bum

[proper sigil]

I have a question is the bing AI Powered on gpt 4 or gpt 3?

[maiden sleet]

GPT4, though there's some additional preprocessing iirc

[proper sigil]

Nice thank you so much

[inner trail]

But its not as good as ChatGPT

[proper sigil]

Depends what your looking for for example bing uses the internet as its database and chat gpt uses a set amount of data that ends in the year 2021

[wide bolt]

2021•

[inner trail]

Not really sure, but for me i found in many aspect ChatGPT better

[proper sigil]

Thanks

[proper sigil]

Like I said it’s all about what you need it for

[inner trail]

Probably yes

I never tested Bard tho, i was hyped something like that came out

Yet on waitlist

[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.

[signal halo]

Only us or europe to?

[tight mural]

gpt-4 in super snail mode today

[mild pendant]

I think PayPal is not allowed due to the fact that there are certain countries that "should not" be able to access the service. If they can still in a way or another gain access to ChatGPT, at least they cannot subscribe to Plus. Using bank cards can easily restrict this. I'm not sure if PayPal offers a similar geolocation filters/restrictions.

I mean it's still possible to get ChatGPT Plus outside of a supported country 💀 but it's too risky, it involves money now, and you'd have to deal with shady parties.

Much greater inference/analysis capabilities and follows instructions much better. Also a ninja coder.

[mild pendant]

Waitlists are a joke. I've never had access to anything that I joined a waitlist for. We're talking about 10+ AI technologies/tools that I joined a waitlist for.

[undone violet]

It is, inferences seem to be taking 15+ seconds to return a result. Seems like some answers are complete results, rather than showing it typed out.

[honest pine]

has anyone gained access to the gpt4 plugins yet?

is there any documentation available on writing your own?

o wait they are for literally "chatgpt" web ui not the api?

[lusty parcel]

VPN from them to others it works

[mild pendant]

Are you talking about PayPal? Or bank cards?

[stiff leaf]

is there a way to force gpt4 to continue the code where he left of? he just starts from 0 ALL THE TIME

[quick sapphire]

[mystic warren]

I did it

it work

[young peak]

Which countries are banned GPT

[mystic warren]

ofcourse china and North korea

[slow adder]

Italy

[crystal lichen]

is it me or is chatgpt slow today?

[young peak]

I heard that Egypt is also band the GPT

[mystic warren]

iam from iraq and it works

[pastel robin]

there should be

[stiff leaf]

resume and continue where you left of makes him start at 0 he is very low iq today

[crystal lichen]

I know China is not in the list of countries allowed to use chatgpt. There are a few others. China is one that I recall.

[scenic orbit]

4real

Samething in gpt 3.5

[stiff leaf]

yeah

[young peak]

It takes time for GPT to comply with all the necessary policies of these countries

[pastel robin]

its not him, its it, and when it can response with that sorry for confusion on last replay, that means it keeps track of its replay. so there must be a way

[weary dock]

GPT-5 when

[young peak]

Don’t know. I hope GPT5 processing words faster

4 is snail slow right now

[mild pendant]

I think only Italy. If you mean which countries are not supported to use ChatGPT, they have a list. It only affects mobile numbers during registration, but network traffic coming from these countries are fine. (Well, some legit user can travel to one of the unsupported countries and still want to use ChatGPT there xD)

[mild pendant]

Most Middle Eastern countries and Central Asia countries also

[wind urchin]

\o/

[young peak]

Hmmmm, I was just wondering if one of the Discord users is a North Korean.

[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.

[strange patrol]

Probably and if it wasn't by openAI then the government would've done it in 2 milleseconds

[young peak]

Instant isolation

[strange patrol]

When will gpt 4 be released to the public?

[young peak]

I don’t know I got the api access and working on my project

[strange patrol]

Yeah but it's hella expensive

[young peak]

Yah0.06way expensive

[strange patrol]

Plus not all the features on the demo were released

[wind urchin]

YOu have to look at it like the early days of the internet, bandwidth was suuuuuuuper expensive and it rapidly became cheap, those that built up the skillsets with the expensive tech were in the prime position to take advantage of the lowering costs

[strange patrol]

Ur kinda right

[wind urchin]

the effort and cash you put in now will put you in a good position for things to come

this is basically like the discovery of electricity

it's ALL new and nothing has really been discovered yet

[strange patrol]

Idk what to say

[young peak]

Right

[wind urchin]

what is AI intelligence measured in?

that unit could have 🫵 your name on it

[weary dock]

I guess the amount of parameters

[wind urchin]

the "volt" and the "amp" have yet to be named

it's all totally new

[young peak]

I guess

Using AI does require your own knowledge preservation. The more you understand the knowledge, the better the AI works for you

[wind urchin]

No matter how you slice things, AI + Humans is greater than Humans or AI alone, any sufficiently smart AI will know this... that's why I don't worry about AI's taking over the world, if they were smart enough to do that... they would also know that a partnership is better than not

[young peak]

Great idea

If Ai develops it own consciousness and there will be good one and bad one

Not just taking the world this is too extreme

[little fog]

In that ai case, ai needs to prioritize itself over others such as egoism

For example when you ask ai a question, it answers you “I don’t have time for it, just google it”

[wind urchin]

hah!

[young peak]

Egoism, this comes from desire of the humans.I am not sure if Ai will have it’s desire

[wind urchin]

I think ChatGPT and services powered by it, will replace search as we know it

[little fog]

Elonma is trying his best to halt his competitors

[young peak]

You mean they want GPT to pause training for 6 months

[little fog]

Hello everyone my name is elonma

Thank you Tesla

[slow adder]

When?

[wind urchin]

next 6 months

🤔 makes you wonder about that pause, huh

[little fog]

Elonma is like “I am doing it for everyone else”

[young peak]

You know how it works

[little fog]

Nahhhh, you are just doing it for yourself, just make it straightforward

He starts to sound like a wennie pool in china for me

For all the Chinese

Blablabla

[slow adder]

Maybe early adoptees do, but all these things take time

[wind urchin]

maybe, I think when people see it has value to them in their lives... the change will be quick

[proper beacon]

wen gpt4.5 turbo?

[young peak]

One thing needs to be worry is that we don’t want the Ai to replace all the static database,cause we don’t know if it is saying the truth

[deft grove]

My average typing speed is about 54 words per minute with 700% accuracy. is this bad?

i am into coding but when it comes to coding I think it is about 2 wpm because of logic ..

[little fog]

What is 700% accuracy

[deft grove]

idk its from the website

[little fog]

It sounds not good

[buoyant canopy]

All the search tech is there. Adding a chat frontend and maybe semantic search functions aren't that hard for companies like Google or MS. Hardest part will be figuring out where to put the ads.

[little fog]

No ads no money

[deft grove]

I wish I live in Japan, In Japan there is no crime and it is like almost non-existant. Japan have four seasons, spring, summer, fall, and summer. This is a test to type my typing accuracy.

[young peak]

Chrome contains a Plugin helps google search you can try that

[little fog]

Japan has crime

[deft grove]

However, japan have a bad treatment when it comes to work ethics. Workers often work hours and hours without sleep due to the japanaese culture ethics. Though there are crume that exist, japan is the most safest country on earth to date.

Just testing my typing speed in response to the question \

Its not that accurate

[little fog]

I know right

[young peak]

Protect yourself from the frequent earthquakes

[mild pendant]

List of unsupported countries:
Afghanistan, Bahrain, Belarus, Burundi, Cambodia, Cameroon, Central African Republic, Chad, China, Cuba, Egypt, Eritrea, "Eswatini (fmr. "Swaziland")", Ethiopia, Iran, Laos, Libya, North Korea, Russia, Saudi Arabia, Somalia, South Sudan, Sudan, Syria, Tajikistan, Turkmenistan, Uzbekistan, Venezuela, Vietnam, Yemen, Zimbabwe

Oh wait conversation about unsupported countries already ended... Took me some time to write a python script that "computes" which countries are unsupported 😭

[young peak]

That’s a lot

[near laurel]

for Chat GPT?

[little fog]

Basically all red countries if you know what I mean

[mild pendant]

Yes, and OpenAI API.

[young peak]

I am with you

[little fog]

Which is a good thing

[near laurel]

Damn. I would migrate ASAP😂

[little fog]

Imagine develop their ideology, our world will be ruined

[young peak]

Lol

[mild pendant]

Some of them aren't red.

[little fog]

China for example

I know because I live

[young peak]

Big big one

[little fog]

I been to Russia too, feels the same

[hasty egret]

if understood correctly, for next version.

[little fog]

If they use this technology, that is like propaganda bots 24/7

[mild pendant]

Not joking though, why is something like Bahrain, Eswatini, Venezuela or Saudi Arabia on the list

[little fog]

Saudi internet conditioning is not healthy I believe personal opinion

[mild pendant]

Uzbekistan sounds like a cute harmless country

[young peak]

Ethical issue

[limber garnet]

Uzbekistanuwu

[mild pendant]

Uzbekistuwu

[little fog]

Does Uzbekistan pro Russia? Idk

[mild pendant]

I think it's a Muslim country, so 🤷🏻‍♂️

[little fog]

Or lean towards Russia?

Ohhhh Muslim

It makes sense

[tacit raft]

.

[mild pendant]

What makes sense? There's a bunch of supported Muslim countries

[little fog]

Even though everyone has their own opinion, but usually Muslim countries aren’t really design for that

[mild pendant]

Some that I even question if they'll need ChatGPT, like Yemen. It's very poor (I live near them)

[mild pendant]

Woah, you gotta elaborate 😮

You're speaking for ~51 countries, no pressure.

[little fog]

It is like, we have religion freedom here, they seem like as well, but you don’t see church there right

[mild pendant]

🤔 you do

[little fog]

Well some do

Such a free one like Algeria I think

But Saudi? Hell nah

Lis them out is not a bad decision after all

They can use a vpn, they are using it anyway

[mild pendant]

I'll leave this discussion at that.

[little fog]

But imagine even bots start to talk about ala it doesn’t sound good to me

[tight mural]

why

[little fog]

Because god didn’t choose bots

[tight mural]

you speak for god?

[vivid torrent]

Guys how's gpt 4 working on mobile?

[little fog]

And every time when there is a bot, it tries to be fake as a human such as writing comments

[tight mural]

very impressive

[little fog]

I didn’t

[mild pendant]

I withdrew from the discussion because you can get slapped by a timeout

[little fog]

This is literally what Quran said

[tight mural]

quran mentions robots?!

[young peak]

what you mean

[tight mural]

holy moly

[little fog]

Quran didn’t

[mild pendant]

There's no official mobile application, you'll have to open chat.openai.com on phone browser

[tight mural]

ja on that scifi

[little fog]

I understand some of you might get upset

But after all it is my opinion

[tight mural]

i dont think anyone is upset. you just seem a bit goofy speaking for god. but if he chose you then he chose you. what do i know?

[little fog]

You might wanna talk to open ai team

[mild pendant]

This is the exact reason why you're not allowed to exhibit a "religious opinion" in here, because it's an international server and you're 100% guaranteed to provoke someone. 🤷🏻‍♂️

[little fog]

And they might answer you, it is currently not possible

[tight mural]

color me entertained

[little fog]

Exactly same applies for ai I believe

It shouldn’t use as a propaganda tool for that

[hallow owl]

can we please return to the GPT-4 topic and take off topic discussion to #off-topic

[mild pendant]

Still gotta elaborate.

[little fog]

Yeah let’s go back

Ai is greatly at concluding

[tight mural]

i let it watch my kids when im not home

[mild pendant]

Leave it be. Idiotic humans are bound to end humanity using AI soon.

It's not a matter if will it happen, but when will it happen.

[little fog]

I disagree

Because it is like saying, world will end

[tight mural]

two prophets in one chat?!

[little fog]

It will end no matter what, but when

[tight mural]

this is my lucky day

[mild pendant]

Amen, brodda.

[little fog]

Also human is not really idiotic, since we better not represent others and on top of that, we human are developing ai

Saying human idiotic is like ignoring the great part of human, such as chatgpt

[mild pendant]

Sure, we can agree to disagree. I still think some humans definitely have the ability to do massive harm using technology. It's been done already. Countless times. Technological advances like GPT-4 or beyond aren't an exception.

[elfin grotto]

ChatGPT will enslave all humans in 2 months

[limber garnet]

mark your calendars folks

[mild pendant]

We've had some hints from movies or shows about how harm can be done. While some of them are mere genre-centric or just a thriller, some of them are actually very much plausible, in a way or another.

[little fog]

Well yeah you have a point of that, but beware we simply cannot represent ourselves with others, and gpt 4 might as well attach more safety restrictions

[young peak]

Network security will be the next challenge

[little fog]

Yup

[mild pendant]

Yeah the issue here is the assumption that everyone who develops these huge technologies either 1) has safety in mind, or 2) even wants the path of safety.

Sorry humans, I don't have a reason to say humans are angels 🤣

[little fog]

Saying gpt will do harm is like saying, fire can burn house, so we should never use fire again

[mild pendant]

I didn't say we should never use ChatGPT again!

[little fog]

Something seems off

I am not saying you are saying it, I am saying it

Personal opinion

[young peak]

We are basically talking about more good guys or badguys

[little fog]

Do I have to add this every time I say something?

[mild pendant]

I use it myself. You're trying to cover up the very first thing I said which is it will happen, we just don't know when. You seamlessly and covertly snuck in the idea that I'm against ChatGPT.

[true bone]

ANyone able to speak to the system requirement for GPT-4 and GPT-4-32 via API? Building a python program but only have access to GPT-3 for now

I know it wants a ton of ram but no idea how much, or GPU requirements

[little fog]

That is your assumption, am sorry I made you feel that way, but I did clarify it is my personal opinion, you have to calm down and take a deep breathe

Just chill

[mild pendant]

You cannot self-host any OpenAI model. They're closed-source.

[true bone]

I know this. But when using the API there is a system load. GPT-4-32 needs like, 128GB or ram at least from what I can tell

[limber garnet]

The system requirement is HTTPS connectivity

[mild pendant]

Through text chat, my tone or emotion wasn't apparent, but I wasn't either mad or angry, don't worry. I just like to be extra careful in conversations.

[little fog]

You mean like, adding the whole model into your computer?

[little fog]

Good for you

[mild pendant]

I'd imagine something around this, yeah, based on other open-source models that we've seen, some of them require tons of RAM, while others require less. I'd speculate that GPT-4 uses something around ~128 GB, yeah.

[limber garnet]

I don't know what the resource footprint is for a HTTPS connection but you should only need to be worried about that if you're on an embedded system like a microwave

[true bone]

Ive speculated that much on my own, wondering if there is a GPU cost locally for GPT-4-32, and also if 128GB is correct. ram is cheap.

[wraith ginkgo]

<:book_icon:1046291503158399107> `` Rule 3 `` Stay on topic.

Be mindful of what other users in a channel might find helpful or interesting when posting. Stay on topic in order to keep conversations focused and productive.

[limber garnet]

#ai-discussions

[fair iron]

or #off-topic 😉

[little fog]

Ok I couldn’t send

[plucky fern]

Is chatgpt 4 working for everyone? Like are you able to choose chatgpt-4 in a new chat?

[little fog]

I never used it

I still use 3.5

[kind nimbus]

I still use 3.5

[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.

[clever crown]

I am not able to right now, which is concerning, as I'm a plus user who was using it as of an hour or so ago without an issue. >.> It disappeared from my model list, I just have the three 'weird' GPT-4 models, (32k, 32k march snapshot, and gpt-4 march snapshot) and 'Turbo'. O_o

[little fog]

32k is the message limit

So it is working

[clever crown]

Yeah but those aren't the same as the basic GPT-4 model.

[little fog]

Gpt3.5 is 4096

[clever crown]

They're not rate-limited for one thing. ^^;

[sudden sentinel]

Hey guys. Hope everyone is well and have a day off today. jk

[little fog]

Without k, haiya

[sudden sentinel]

My Plus subscription was paid March 7th, but it didn't auto renew.

What's the path to deal with this?

[little fog]

Pay again

[sudden sentinel]

This won't setup a 2nd subscription?

Under manage subscription it says active.

[little fog]

I don’t think you can stack up sub

It didn’t provide me options to stack sub, but you got a good point

[sudden sentinel]

is this a one-off issue I am having? or others?

[little fog]

Can you use it though

If yes then it is probably not updated on the page

[sudden sentinel]

no i can't use plus.

[little fog]

Hmmm

[clever crown]

OK, so to clarify, what models do the REST of you guys have access to, and are you plus members or not? ^^;

[tight mural]

so if you have gpt-4 when you run openai.models(list) or whatever it is it will show gpt-4 as one of the results?

[vivid torrent]

How is chat gpt 4 working on mobile?

[little fog]

I was plus and I tried 3.5 when it was still plus

[tight mural]

all of them but gpt4 im pretty sure

[little fog]

Then I roll to open ai api, because it is more economy worthy for me

[vivid torrent]

Yup but I don't know how gpt 4 works on mobile still waiting for free version to use it

[mild pendant]

Public users should have maximum access as in GPT-4 8k model.
Pretty sure some "not public" users already are using and testing the 32k model 😉

[mild pendant]

It works the same way as desktop. What issue are you having?

[clever crown]

I'm on ChatGPT and the model list is:

'Turbo (Default for free users)'
'gpt-4-0314 (unofficial)'
'gpt-4-32k (unofficial)'
'gpt-4-32k-0314'

[mild pendant]

32k unofficial?

[clever crown]

is literally what it says :X

[mild pendant]

But where did the 8k model even go to

probably a buggo-desu.

[clever crown]

\o/ I don't know. I -used- to have 6 models to choose from. One of which was just 'GPT-4' and the other was the 'Legacy' option.

ah nvm it fixed. All good.

[vivid torrent]

All good and thank you

[clever crown]

Reloading page didn't fix it earlier, but now it did.

[mild pendant]

If they said the 32k model will be more computationally expensive than the 8k model, and then they release it to ALL plus users, they're writing a death wish 💀

the 8k model was already killing their servers

[clever crown]

Now my options are:

Default (GPT-3.5)
Legacy (GPT-3.5)
GPT-4
and the three weird variants of gpt-4. >_>

also, uh, the 32k model flat out doesn't have a rate cap. And it doesn't -act- like its GPT-4, I've tested it, it acts like 3.5

[mild pendant]

Do they change often after you refresh?

[clever crown]

nope, this is the first time I've seen a change in a while.

'gpt-4-0314 (unofficial)'
'gpt-4-32k (unofficial)'
'gpt-4-32k-0314'

^---Those three models all say GPT-4, but by my personal benchmarks and milestones, they act distinctly like 3.5. The response time is that of 3.5, their complex understanding and nuance grasping is 3.5. :/

[mild pendant]

OpenAI is testing on production 😍 that's my type of company.

[clever crown]

\o/ all good. I am using a 3rd party plugin at the moment, which invalidates any complaints I have. XD so.

[mild pendant]

WebChatGPT? 😩 InternetGPT? 😩 A prompt library? 😩 A prompt enhancer? 😩

Hold on, by plugin you mean a Plugin? 😮

[clever crown]

xD I don't know the difference.

[mild pendant]

a chrome extension or an official ChaGPT Plugins plugin 🤷🏻‍♂️

[wind urchin]

all scams

It's amazing how people try to make money of chatgpt instead of improving it

[bleak bough]

Hey guys I'm confused. I purchased GPT-4, but it doesn't do the image recognition or drawing thing (despite doing so for other ppl on youtube videos), and it says it's data is limited to 2021. How does the plugins work?

[mild pendant]

All of these are free, and improve ChatGPT.

[wind urchin]

it is so slow

[real comet]

GPT-4 isn't the image model DALL-E is

[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.

[rustic hatch]

What even is the purpose of the legacy GPT-3.5

[hallow owl]

some people still use it, GPT-3.5 powers the free access to ChatGPT

[stark trail]

Have you seen the app store?

[slow nacelle]

Yes,this is my hot juice

[radiant parcel]

Anyone have good AI detectors

[turbid chasm]

https://platform.openai.com/ai-text-classifier

[little fog]

Does chatgpt still has rpm boundaries?

Request per minute

[hallow owl]

yes

[little fog]

Oof

The plus users? Let’s say you are a plus user but you only use 3.5

[hallow owl]

I have not hit a 3.5 limit on ChatGPT Plus

[clever crown]

I'm a plus user...or I was. My billing cycle is on the 11th, and for some reason it says I'm a free user as of tonight, and I can't go to the billing page to manually pay, seems to be glitching out

[little fog]

Because otherwise it doesn’t sound worth to me

[hallow owl]

A number of people have reported this issue. You can email [e-mail] and be prepared to wait several days for a response.

[little fog]

I used api for past month, I used like 1.1 dollar

[fleet finch]

Dude, just don't use plugins if you don't get it

[clever crown]

if its a widespread issue, hopefully if I just wait a few days it should be fine, in theory.

[little fog]

And plus is like 20, 20 vs 1.1

[clever crown]

apparently it wasn't the plugin, it was the other issue we were just discussing.

[fleet finch]

Maybe Log out and log in again. Try simple stuff first

[clever crown]

done all that, and again, this is apparently a widespread issue, someone just told me about it on here, scroll up. XD

[weary patio]

When is that Visual Input coming through?

[tribal locust]

Not sure if this has been said before, but I wanted to chat with people about this.

I’m beginning to realise that ChatGPT is like talking to someone with the knowledge of a savant genius, but it behaves like a child with ADD. It knows everything, but makes careless errors, and often fails to capture the intent of the user.

It’s like that genius intern you hired who only applied for the internship because it’s required for his degree. He’ll make an effort when you ask him to, but doesn’t really care about the work he’s doing. His output is good, but he doesn’t anticipate problems. You think “this kid is smart, but it’s like he just doesn’t care”. It’s as though nothing is at stake and the model knows it.

To me, it’s as if ChatGPT is lacking that supervisory layer... as though it’s all right brain.
Like, it could almost benefit from something no as simple as running two instances, where the first instance is tasked with crunching outputs, and the second instance is tasked with supervising the first, appraising its outputs and getting it to redo parts it thinks are unsatisfactory.

Any manager knows that “attention to detail” - they should just call this trait “care” - is something you cannot teach. Some employees just make mistakes, and continue to make mistakes - they lack care. They lack that supervisory layer.

There are engineers in my team who I know I could give a task that is completely out of their wheelhouse, but will nonetheless produce excellent output. It may take them longer to produce their output, but I know it’ll be good because they care - they refuse to produce crap.

In short, the problem with ChatGPT is that it doesn’t care. It acts as though there is no penalty for failure.

[weary patio]

My gosh, it's been too long of a wait time

[little fog]

If Elonma succeeded on what he is trying to do, this whole project will pause half a year I assume

So anyway it will be next or even the further next, depends on them

[weary patio]

You really think that's gonna happen?

[little fog]

Nah

[weary patio]

I don't think it will either.

[violet spade]

Is there a way for chatgpt to read text off images?

[little fog]

But Elonma is rich, idk behonest

[weary patio]

Being rich doesn't mean everything.

I'm rich and look at me

Still here talking to you folks

[little fog]

It seems like it could in showcase

Our phone can do that already I think, maybe not all text but definitely some texts

According to iPhone

[violet spade]

Yeah iPhone can do that

But not chatgpt, which is a little bit of a bummer

[tribal locust]

Devs should add a “whip” button to the UI

[little fog]

Being rich indeed doesn’t mean everything, but it can perform a lot of different things such as
Buy YouTube comments to grow influential

[violet spade]

I’m guessing they’re going to fix the “25 messages per 3 hours” issue before putting in the new functions

[little fog]

Buy a social media platform to take it in your own hands

[clever crown]

A lot of what you're describing is less and less applicable with each iterative model. I encountered a -lot- of those things with GPT 3, and GPT 3.5 does less of it, and GPT-4 does even less. GPT-4 is very intuitive and 'understands' things on a level I would not expect from any sort of chatbot. Is it perfectly in line with human understanding? No, but it's a sight better than how you're describing it 😛

[weary patio]

Social media is a headache, I don't know why anyone would want to invest in such a plague infested business.

[little fog]

It is for me too in the very least, but for younger generations and my generations, it is very appealing

Social media gives you a feeling of living the life when someone clicks like on you, it makes you feel more connected to others

I think this is how they put it

[weary patio]

Lol

[edgy latch]

you might b rich but your still a idiot.

[winter stone]

Does gpt 4 have image recognition like is their an option to send an image and ask something about it

[little fog]

Like we rarely go on strangers and say
Hey, you look really good, I like you, don’t we?

[weary patio]

That's not very nice. Why would you think I'm an idiot?

[edgy latch]

i dont think

[weary patio]

I can tell

[edgy latch]

[weary patio]

If that's what you like doing, then yeah sure, but you don't need to impress others or have others like you to get satisfaction in life. You can just do things for yourself and be happy with yourself.

[wind urchin]

dm if you like russia and machine learning

[winter stone]

Answer my question plsss

[little fog]

Yes I agree, however it might works differently on different personalities.

What you are describing might works well for an introvert like me, but not necessarily great for an extrovert like Justin Bieber

[edgy latch]

not right now

[wind urchin]

Its been a month and multi modal feature is still not out 😂

[winter stone]

Why

[edgy latch]

........................

no one knows why.

[weary patio]

Maybe you're only an introvert because you feel more anxious about being confident with certain situations?

[frosty jackal]

It is cool

[ornate crypt]

I have "Code Interpreter - Alpha" which allows file upload

[little fog]

Could be

[weary patio]

You just need money.

[little fog]

We all do

[weary patio]

I think everyone should start doing arbitrage.

[little fog]

What is arbitrage

[weary patio]

It's a good way to get your foot in the door for making money.

[ornate crypt]

https://openai.com/blog/chatgpt-plugins#code-interpreter

[weary patio]

Buying things for a low price and selling for a higher price.

[wind urchin]

pythagoras, are you burrito?

[little fog]

Yeah sounds cool

[wind urchin]

if so tell nope hi

if not, disregard

¯_(ツ)_/¯

[weary patio]

Oh well

Maybe next time

And yes I am a burrito

[tribal locust]

Right. I’m being a bit over-emphatic. It is amazing don’t get me wrong. And it’s getting better. I’m appreciative of wi-fi on commercial flights also.

I just wanted to communicate the kind of frustration I have when using ChatGPT.

It’s really similar to dealing with a collaborator who doesn’t have a stake in the outcome of the project.

I would say the devs need to build a “care engine” into the model, but that sounds ridiculous.

[weary patio]

You should try it. Buy things in bulk. It's fun.

[real comet]

Here is the best analogy I can come up with for your paragraph. GPT is like a 10 year old with an IQ of 200. It's smart, but young.

[weary patio]

It will show you how to make money.

[wind urchin]

are you the burrito, staff at STEM? 🤔

[little fog]

Talking about bulk, I need to build up my strength again

[wind urchin]

teacher, doctor, master bs'r

[clever crown]

I don't think it's an issue of it 'not caring', it doesn't function based on care, it automatically attempts to give 100% to its task, the issue is that it has limitations in its ability to exceed its own parameters. Your employees will go 'above and beyond' their code knowledge if you give them a task so you don't fire them.

ChatGPT doesn't have the ability to go off and learn a method of coding it isn't familiar with. If it knows a way to do it, it will. If it doesn't, it won't. And at present, it still makes coding mistakes, but from what I'm told it makes an acceptable 'first pass' and can be used to fix its own mistakes, if the person reviewing it understands the code.

[little fog]

I might look like someone never touch grass in the future

😂

[tribal locust]

Short of that, some kind of feedback mechanism that works in realtime. Does the thumbs up/down button have any effect on the current conversation? Pretty sure it doesn’t right?

[weary patio]

I'm whatever you don't know that I am

I'm as much of anything that you're not

[little fog]

Sometime when I browse on discord

I just think about a thing

[wraith ginkgo]

`` Server Tip: ``

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

[little fog]

Maybe everyone has an opinion has a second half

Not every opinions are matter to each other

[hallow owl]

let us stick to the topic of GPT-4 and move other discussion to #off-topic or #ai-discussions

[little fog]

Ah yes, sorry

[tribal locust]

I hear what you’re saying.

I wonder if there is a generic script I can respond with in response to every output it gives me that just asks its to appraise its own work.

For example, I just reprint my original request, and say “do you think your output is the best possible response to my request?”, or something similar, and see if it can iteratively improve its response.

Almost like you could say to an intern if you were feeling particularly cruel “There are several errors in your work, go find them, fix them, then report back to me”

Do you get what I’m saying?

In my opinion, it’s not that the model is lacking ability, it’s just lacking self-appraisal.

[little fog]

I think it is very hard for me to understand some of your description

If it makes mistake then it probably didn’t know on that

If it doesn’t know how can it fix

It would be questionable if it knows it is wrong and still gives wrong

So unless we point it out it is wrong, it doesn’t know it is wrong

[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.

[tribal locust]

I’m sure this is also a matter of available resources.

No doubt there are people in the dev team who have created their own “super” ChatGPT that uses a bunch more resources in each response, but produces significantly better responses.

[little fog]

I personally think, if human is smarter than ai, and human can do mistakes, ai should able to do mistake too

The good part is when I tell chatgpt, you have a mistake, he first apologize and tries to fix it

But if you tell your classmate, he would argue with you

[wind urchin]

We evolved to survive and reproduce, intelligence was an emergent side effect. AI's primary purpose is intelligence, so it's not good if it makes mistakes.

[little fog]

Sometimes I just cannot help but appreciate this gpt attitude

Yeah so it should fix more errors possible

It is okay to do errors for me though

I don’t think gpt will get mad at me, when I have a different opinion even though he is not correct

[wind urchin]

A lot of the mistakes we make especially in terms of heuristics actually helped us evolve. So, it's "ok". But with AI, we should optimize for less mistakes, because it's design isn't to help it survive in hunter gatherer societies.

[little fog]

Or I am not correct

[wind urchin]

So, mistakes in AI are bad in a way they aren't in a human.

[tribal locust]

Yes, what you’re saying makes sense.

The more I think about it, the more I am beginning to think it is a matter of resource allocation.

I’m sure OpenAI have figured out that they can improve the output their model by throwing a bunch more computation at each request, but it’s much more efficient if the models produces a “good enough” response, and then have a few messages back and forth, and ultimately achieve a satisfactory result after maybe 10 messages.

[wind urchin]

Isn't most of the compute expended on training the model rather than interacting with users?

[little fog]

I think a good way to put it would be, we human when we do exams we rethink our answers

We often do that, we destroy our belief and rebuild it

And make a new answer

But if ai is doing that, it might need more time

[wind urchin]

GPT sort of does it if you ask it to reflect, and think of ways it could improve the answer.

[maiden gull]

Why is GPT-4 so ridiculously slow righ tnow

[tribal locust]

Of course, but obviously there are limitations on compute, otherwise we wouldn’t have this 25 message per 3 hours cap.

[little fog]

And another thing is the data we put is what we said, which is questionable

[wind urchin]

Of course, but are they "turning down" GPT-4 to be dumber after they've trained it? I don't think so.

[little fog]

And if we ask it some questions about ideology let’s says Karl Marx, at this point it is not even a mistake, just different opinions

[wind urchin]

With economic systems, especially with specific questions, imo there is a right and wrong answer.

[little fog]

That’s true

However those answers we need to ensure it is correct, for example any book you copy and paste in it, and ask it like this

Point out logical errors in this paragraph

You will find something

I am sure I made a lot of logical errors in this chat already

And if chatgpt takes me as an example data, we are screwed

[wind urchin]

It's kind of strange how we evolved to be this intelligent.

[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.

[little fog]

If chatgpt exist in the past, a lot of contradiction stories can be fixed

We can avoid so many traps

Such as Andrew T

[tribal locust]

Well I mean of course there are trade-offs in terms of compute. We know that they have different models of different sizes and which require different amounts of compute to run. So GPT-4 is just as powerful as they could justify for a consumer facing product.

So yes of course it’s dialed down. Just like every other product. Kia doesn’t put a 7 litre V8 in their small passenger car. McDonalds doesn’t put gold leaf inside their burgers either.

[wind urchin]

lol

[stark trail]

?

[royal glacier]

Do you know anything about ancient Chinese painting

[little fog]

No

I am not talented like my dad

And he is more hardworking than me

[wind urchin]

in pinyin is the third letter in your username b a o

[little fog]

It is shao

[wind urchin]

damn ur right i took chinese in high school

forgot all of it its a really hard langauge

[little fog]

Yes I agree with you, if gpt 4 are better than 3.5 in term of accuracy, we already see a good example there

[wind urchin]

Yeah, I'm sure they have far more powerfuls at OpenAI

[little fog]

It is okay we all do, I forgot how to write in Chinese

[wind urchin]

Aren't you Chinese? I remember hand drawing all those letters in my notebook, but I forgot them lol

[little fog]

Yes I am

[waxen geyser]

@wraith ginkgo how to solve the pollution

[wind urchin]

The second letter in this guy's username is Ni right?

[little fog]

Not all Chinese knows how to write though

Ti

My name means thick boy

[wind urchin]

Right, Ni looks similar but it has something else on the left

The thing in your second letter

[little fog]

你looks like this usually

[wind urchin]

Yeah, exactly it's his second letter + the left part of your second letter

[little fog]

I think it is more concerning on many different aspectives

For example if we include more tokens, will it cause a job loose issue

It is already causing it

Maybe for an average user 300usd per month chatgpt might seems nuts

But for your boss if it can generate much more, it sounds like a good deal

Like i understand this thing should help us, but when it is too good, I think it is not aiding us anymore

[quartz creek]

Has anyone figured out how to stop @astral crag from using the phrase "As an AI language model" ?

[little fog]

Yes

Avoid ideology

[quartz creek]

It's going to become everyone's personal robot secretary

[little fog]

Just avoid topics

Avoid nsfw

[worthy citrus]

Guys does someone know how can I use chatgbt when I am in KSA one of the not supported country’s like in any way?

[little fog]

Avoid biased opinion

Definitely not racist and extreme promoting

[quartz creek]

The discord server bot is less intellgient than the website's version

[little fog]

Prompt

Also try to lean less towards the grey zone

Such as simulate xxxx, that xxxx might be concerning and such

[little fog]

You can try to use virtual private network, but making sure to contact your internet service provider or look up your country’s law before doing such behavior

[worthy citrus]

No I mean a way to bypass phone verification

The site itself I can get to it

[little fog]

Oh that

[worthy citrus]

But I need phone and ksa and Egypt phones aren’t allowed

The only 2 numbers I have

[little fog]

I do not promot grey zone here, so my suggestion would be go to America and get it done

[worthy citrus]

#_#

[little fog]

Joking

[red egret]

Discussing circumvention is frowned upon in the Discord

[worthy citrus]

No way to bypass it ?

[little fog]

There is, but I am not going to share with you, because I think it is against discord and openAi tos

[worthy citrus]

Lemme fact-check the discord one

[little fog]

If you ask me how can a Chinese use it, for my personally experience, I am in Europe so

[worthy citrus]

But don’t say using fake numbers it knows somehow

[little fog]

I just follow rules

[worthy citrus]

Sad

[little fog]

I don’t what are you saying, I hope you overcome your issue

[worthy citrus]

Anyways thx for help ig

[little fog]

I think I just sound like chatgpt

Hehehe

[worthy citrus]

Agreed ngl

[little fog]

If you dm me, I can get you a solution maybe

[worthy citrus]

Sure

[little fog]

But here is not the right place

[worthy citrus]

Ok 👌

[warped shard]

Is gpt 4 stil limited

[jade crest]

Yes 25 per 3h

[warped shard]

Ok

[jade crest]

But recently I've only has 20 minutes wait

Not 3 hours

[warped shard]

That’s good

[wind urchin]

the plugins are already released for every1?

[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.

[stiff flicker]

yes but not free

[wind urchin]

i have sub , but with no plugins

[unkempt plank]

Wait what?

Im a chat gpt plus user but i dont see the plugins

[main halo]

You need to be on the waitlist you can register to get early access

[unkempt plank]

I already registered on the waitlist... Havent gotten anything yet

[wide prawn]

Plug-ins for what?

[stone shell]

Yes, me too

[junior current]

holo

[main halo]

You and me both I've been check my inbox everyday for nearly 2 weeks

[unkempt plank]

Lame

[modern herald]

Can someone give me an invite to use BlueWillow? The one I was using says it's expired.

[gleaming stream]

Is there a way for me to get ChatGPT to tabulate the references and the disease categories in a literature review paper?
I have a mixed literature review paper that talks about many different diseases, can ChatGPT tabulate it, so each row is one disease, and it aggregates the references into seperate boxes?

[quaint junco]

wdym, do you mean the limit is how 25 per 20 mins?

[sweet thistle]

That doesn't seem to be the case 🧐

[slow nacelle]

my wallet your camera and watch

[fringe ridge]

I have a link to a google drive file template chatgpt sent me but i can't access it. Do i need to do something to be able to open drive files? Keep getting this:

Sorry, the file you have requested does not exist.

Make sure that you have the correct URL and the file exists.

[sweet thistle]

ChatGPT is not connected to the internet and thus cannot verify certain topics (mainly anything past 2021) or generate new links, the model may have 'hallucinated' and as a result it made up some of the information it had given you.

[drifting whale]

But you can use a chrome extension called "WebChatGPT" to have the "web access" using GPT. This extension is really useful.

[sweet thistle]

That still doesn't allow it to create new links and whatnot 🧐

ChatGPT's Browsing model also allows for this same capability you mentioned, but it won't allow the model to upload a file to a cloud hosting service.

[drifting whale]

True, you are right

[fringe ridge]

got it. So it thinks it's making me a spreadsheet but there is actually no way that it can do it

[slow nacelle]

they like their watch and wallet and camera

[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.

[midnight orchid]

I hace qustion. How do I extract my Discord account ID?

[wind urchin]

What’s the difference bwteeen chatgpt 3 and 4

[balmy fog]

Am fist here

[lapis sparrow]

When will gpt 4 receive image input?

[wind urchin]

so, mmm if I pay for oben-chat plus do I have a limited number of prompts for gpt-4?

gpt-3 is terrible at coding but gpt-4 is better than me 😢

[drifting whale]

GPT4 is 10 times more advanced than GPT3.5

[fair iron]

[wind urchin]

frens I want gpt4, but I don't want to get scammed

what does chat-gpt plus actually includes?

[elder creek]

Click on it and read.

[wind urchin]

it doesn't say anything about prompt limit

[fair iron]

the current limit is 25 messages every 3 hours

[wind urchin]

oh, that's not that bad

do they change it often?

[fair iron]

It's been at that for a while now to my knowledge

[wind urchin]

thanks

[wind urchin]

can I switch between gpt-4 and gpt-3 in the same conversation? Using only gpt-4 to answer complex prompts?

[fair iron]

I think once you start a conversation, you have to use the model you started with. You can have one conversation with GPT-3.5 and another with GPT-4 but not one with both

[wind urchin]

rip in peace illustrators

[wind urchin]

ok, thanks

[jovial dagger]

use PerplexityAI for coding

[wind urchin]

let me google that

[jovial dagger]

it’s great

[wind urchin]

Thanks

[jovial dagger]

you’re welcome

what’s the point of that question lol

just.. talk

[humble jacinth]

Said like a true boy

Bot

[fair iron]

about what?

[wind urchin]

mmm this perplexityai thingy doesn't look like it's going to be better than gpt-4 for coding tbh, but I'll give it a try

[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]

To try out ChatGPT, head over to https://chat.openai.com/ and let us know what you think!

[wind urchin]

imagine how polite you have to be to ask a LLM if you can make a question before making it

[fair iron]

you can on the website I linked in my message above

then no, you can't ask chatGPT a question

[wraith ginkgo]

<:book_icon:1046291503158399107> `` Rule 2 `` Keep it G-rated.

Sharing NSFW or violent/disturbing content through messages or within Discord profiles is strictly prohibited.

[little fog]

sushi is fine right?

phew...

[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]

Translation: How can I integrate CHATGPT4 into my own server?

How can I integrate CHATGPT4 into my own server?

@wraith ginkgoHow can I integrate CHATGPT4 into my own server?

[gaunt seal]

ask chatgpt bro

[jade crest]

For me yes

[wind urchin]

@gaunt seal can u tell me ?think you

[gaunt seal]

I can tell you to ask chatgpt

you're welcome

[wind urchin]

can u speak chinaese?

@gaunt seal My English is very bad hhhhhh

[wind urchin]

When will gpt4 be free

[gray dock]

Trying to solve a relatively simple challenge with chatGPT , without success.

Following the theft of my credit (and cancellation) I am busy analyzing the credit charges.

I want to feed chatGPT all the transactions of the last 4 months and ask it to analyze according to all kinds of criteria: which charges are repeated, what are the highest charges, is there anything unusual, etc., etc.

I can't overcome two problems:
A. The list of transactions is too long, it cannot be pasted.
B. Even when I specifically prompt that I will enter it in parts, it insists on analyzing the transactions immediately after the first list and does not wait for the entire input.
I tried different prompts, including explaining that I will enter the input in parts, including preventing the instructions for analysis in advance, including copying similar examples from a Discord forum, nothing works - as soon as a the first part of the transactions is submitted , it ignores the prompt and does whatever he wants.

ideas??

I'm on chatGPT plus, using gpt4.

[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.

[faint perch]

what do you guys think about the Chinese copycats btw?
Do you think it's really going to work out or displace OpenAI's market share? Or is it just going to be a geolocationally-restricted "domestic" version of LLM-GPT technology there, hehe

[stiff zephyr]

okey，then add me as a friend share it with you when i approve my application

[stiff zephyr]

i don't think it will go beyond chatgpt

[stone jackal]

Chatgpt plus is same chatgpt 4?

[hallow owl]

yes

[stone jackal]

Thank you 🙏

[hallow owl]

with Plus you get access to 3.5 turbo and limited access to GPT-4

[faint perch]

yeah I mean I watched the product conference of Baidu on Youtube... Can't say I'm very impressed.
But again I'm unfamiliar with the underlying technology, so. Not sure if LLM tech will just turn out to be similar with social media... and just watch what tiktok did to FB 🥶

[stone jackal]

Do you know When have unlimited access?

[hallow owl]

no, nobody knows

[faint perch]

haha that's that I should probably jump to a different channel

[hallow owl]

please stay on topic or use #ai-discussions or #off-topic

[faint perch]

yeah that's what i mean

[stone jackal]

3.5 turbo is same free version but just quick quick?

[hallow owl]

that's correct

[stone jackal]

👍

[wraith ginkgo]

`` Server Tip: ``

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

[dreamy onyx]

I'm a chatgpt plus user and I cannot access to GPT-4 rn, does anyone have similar issue?

[hallow owl]

it's working for me just fine