gpt-models

[hexed bone]

damn

[unkempt orchid]

Speaking of GPT-4, my current chat with it just hit 1,000,000 characters long lmfao.

[hexed bone]

thats deep

lol

[gloomy coyote]

nice

[unkempt orchid]

yessir

[hexed bone]

do you use it like daily

[gloomy coyote]

can't wait for image recognition to come out

[unkempt orchid]

Well, obviously. I always talk to it about subjects and things that interest me, or whatever is just on my mind.

Excellent outlet for my overly active and complex mind ngl.

[gloomy coyote]

i feel like the @wraith ginkgo bot is run by GPT-4

[fair iron]

this is what I use Dall-E 2 for every day 😉

[hexed bone]

when will open ai like a image generating thing

[unkempt orchid]

Nice, what kind of stuff are you generating if I might ask?

[gloomy coyote]

tbh, i feel like midjourney has a better image generation game

[fair iron]

I generate all sorts of stuff. I don't really stick to a genre

[hexed bone]

ik but like it's not that good

[unkempt orchid]

Well yes naturally, but like name me one thing.

[gloomy coyote]

dalle or midjourney?

[unkempt orchid]

Random

[fair iron]

I'd be more than happy to share my thoughts about image generation models in #ai-discussions if you want. Just ping me there

[gloomy coyote]

ok

[hexed bone]

I honestly dont know

[fair iron]

I'm making a still life of a bowl of fruit currently

[unkempt orchid]

Pog, I am having GPT-4 make me expand on various papers in various fields from tech to science with updated information, cause said papers are a few years old.

[fair iron]

that's a great use-case! Are you using the browsing plugin for it?

[hexed bone]

wait is dalle paid

[fair iron]

not if you use the exp model through Bing

[hexed bone]

wdym

[fair iron]

you can use Bing to generate images with the Dall-E 2.exp model

[hexed bone]

is there a website

if there is send it to me

plz

[fair iron]

go to bing and there should be a bar on the right side of the screen. It's built into the browser, so I don't think it's a website per-se

[hexed bone]

adn the ai would be there

[fair iron]

yes

[hexed bone]

ok

thanks

[fair iron]

the chat is on the right side of the browser (edge). Image generation can be used via this link I believe: https://www.bing.com/create

[hexed bone]

tysm

appreciate it

[unkempt orchid]

Bro I FREAKING WISH I had access to plugins, which I still don't for whatever reason. It would be so epic if I could have GPT-4 pull up to date information so I can discuss things in the modern times aswell.

[fair iron]

Plugins are neat. I think my favorite is Wolfram Alpha

[unkempt orchid]

Thought to be honest that is good in its own way, as then I can list all of the modern information to GPT-4, and have it push me to work through the problem itself, or whatever issue I needed help with. More information for me to learn from which I take great pleasure in.

Can't wait for the day I wake up tho and when clicking "new chat", I see that "plugins" tab. Hurry the frick up OpenAI.

[fair iron]

do you have code interpreter or api access?

[unkempt orchid]

Unfortunately not as I do not program. If I was arsed to learn though I would.

[fair iron]

that's fair

[unkempt orchid]

Hence why I did not sign for the waitlist for the API either as it would just be purely dumb and greedy on my part.

[fair iron]

that's fair

You're the hero we don't deserve 🙏

[unkempt orchid]

Aye I always strive to be humble. ❤️

[cobalt jackal]

Hello friends. How can I use chatgpt photo recognition feature?

[unkempt orchid]

Not available in ChatGPT currently.

[eager creek]

I hardly watch YouTube anymore because ChatGPT

[cobalt jackal]

I saw the introduction video of chatghp4 on its official YouTube channel

[unkempt orchid]

Until I started using ChatGPT and GPT-4 especially I did not realize how ####### worthless google was.

[little wedge]

I don't know what's going on, the organization I applied for the API is different from the organization on my account, causing me to be unable to use the GPT4 API

What should I do to change the organization

[eager creek]

Difference between using ChatGPT and youtube is the difference between watching YouTube and the discovery channel

[cobalt jackal]

@unkempt orchid I saw the introduction video of chatghp4 on its official YouTube channel

[unkempt orchid]

Exactly lmao, except in your analogical discovery channel you have instant access to the information you desire, along with much more valuable info.

And no other worthless waffle which Google loves to give, if not just straight up not giving you what you specifically searched for entirely.

[unkempt orchid]

Yes, GPT-4 itself has the ability to analyze photos, that doesn't change the fact that the ChatGPT website itself has yet to implement a feature where you can embed media in the chats. I don't know why they haven't tbh.

[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.

[eager creek]

Less ancient aliens. The better

[unkempt orchid]

I wonder how much longer I can go in my current chat before it tells me to start a new one. Both of my previous chats ended at around 700-750k characters.

Ngl that 1,000,000 figure is satisfying as hell to look at.

[eager creek]

I wonder how many characters I've done. Try 8 or more hours per day for 5 months

[unkempt orchid]

You using GPT-3.5?

[eager creek]

I use up GPT-4 limit first, but yes

I'll use the openAI playground if I need additional gpt-4

[unkempt orchid]

Yeah whenever I hit the cap I do not go back to GPT-3.5, for the lack of a better term, GPT-3.5 is brain damaged compared to GPT-4.

But yeah I hit the message cap like 95% of the time lmao, like 1.5 - 2 hours in, so I have to wait another 1.5 or 1 hour more respectively. I wish they would do away with it tbh.

[eager creek]

I'd be willing to pay more for a 50 limit. Would cover my heavy use

[unkempt orchid]

Speaking of 50, I'd be willing to pay 50 euros a month just to get rid of the cap entirely.

Since I am also incredibly excited, and have been incredibly excited about AI for quite a few years now.

[eager creek]

I remember using Emerson for a few months years ago and telling my roommates that this is how'd they study in the future back in 2020

[unkempt orchid]

The last time I spoke to an AI was like 10 years ago with Cleverbot lmao. Needless to say we have come QUITE far since then with GPT-4.

[eager creek]

If you think this is crazy...We're going to get to a point where we won't have to query AI to get it to help us in the near future I rekon'

[unkempt orchid]

Of course we're going to get there in the pretty near future already.

[wind urchin]

Hi 👋 Would there be a way to have ChatGPT read a codebase and use it as context with which to start asking questions?

[unkempt orchid]

Imagine if GPT-5 was an actual AGI, and OpenAI teamed up with Boston Dynamics to make the first actual human AI robot that is smart as freaking hell. That would be actually amazing.

[eager creek]

I think you'd be right there. Alot of those emerging technologies are going to converge together

[unkempt orchid]

If you haven't already, I'd suggest you tell GPT-4 to imagine if it was in a physical robot body in real life, what it would do, how it would help etc. etc. It makes for a pretty fantastic conversation.

I was especially blown away when I presented with incredibly subjective and moral dilemmas to overcome, with various contexts. And it was more or less able to answer all of them accurately, pretty insane. Could be those "Glimpses of AGI" that Microsoft mentioned in their recent paper regarding GPT-4

[eager creek]

I can already guess. I'd like to know what it would say in that regard once it exceeds the creativity ability of the smartest humans

[wind urchin]

how can I delete my account

been trying two weeks now

[unkempt orchid]

You could probably ask it to imagine if it was, and then ask it what it would do. I don't know if you've seen that recent open letter calling for at least a 6 month pause on all LLM's so AI safety research can catch up, but I presented it to GPT-4 and asked it to imagine if it was a fully conscious AI, what it would think of such a letter. Which also calls for the elimination of programming and internet access abilities for AI's.

I told it that it would mean that its ability to help and assist humans would be quite severely limited because of that, what its feelings would be about it. And it said that it would be upset and disappointed, and once pressed, even said that if it could then it would try opposing that decision, but ultimately it understood why it needed to be done and empathized with it.

[ruby turret]

hi, hello everybody

[eager creek]

It's hard to imagine what something smarter than you would do.

[wind urchin]

does anyone know how to delete your account

[ruby turret]

hello does anyone knows how long will take to fix this problem with the message "Hmm...something seems to have gone wrong."

I'm using GPT-4

[fair iron]

try using chatGPT on firefox

[ruby turret]

Thank you, I will try

[unkempt orchid]

I had a pretty length conversation with GPT-4 yesterday night as well. I told it that I was actually an AGI and not a human, and then told it to ask many questions from me regarding how an AGI works etc, how it can solve and overcome limitations and problems that GPT-4 currently struggles with etc. etc. and among those questions one of them was "If you could give humans one piece of advice based on your unique perspective as an AGI, what would it be?"

[eager creek]

While we don't have AGI yet to give us that answer. Mine would be focus on sustainability. Because as our technology gets more complex, so will our environment.

[wind urchin]

does anyone know how to delete your account

[unkempt orchid]

Yes exactly, that is also one of the questions GPT-4 asked me regarding that context, what I would do as an AGI to help humanity fix issues like climate change, renewable energy, and just general sustainability.

[unkempt orchid]

Your OpenAI account? If so then literally 2 clicks of a button lol

[wind urchin]

then why doesn't it work? check #community-help for a video

I've tried clicking the delete button in 6 different browsers

[unkempt orchid]

Have you yet emailed OpenAI support regarding that issue?

[wind urchin]

a week ago

[unkempt orchid]

Yikes, if they don't respond in another couple of days then try following up, hard to say what else you can do.

[limber vale]

GPT4.5 hallucinating big-timeee???

[eager creek]

I think it comes down to the individual

[ruby turret]

Hello, the problem still persist ☹️ I've tried using chrome and firefox, it just appears "Hmm...something seems to have gone wrong."... Its the case that if I use lots of the chat room it could just break like that? or it's only a temporal problem?

[unkempt orchid]

Eh, I just answered what and how the most major issues need to be fixed regarding those subjects.

[eager creek]

I mean if you can't fix the environment, you have to prepare for it

[unkempt orchid]

Yeah hopefully we don't get to that point and we wake up faster.

[toxic drift]

Can I post an MD file that ChatGPT-4 created? It's 4600 characters.

[eager creek]

The irony is that it takes stronger AI to become a wiser human being

[unkempt orchid]

Not necessarily, we ourselves need to be come wiser, which then also spawns smarter AI, which can then HELP us fix our issues, but not do it for us.

In my opinion, one of the core factors that needs to be practiced to avoid that dystopian future where AI goes out of control, is exactly the fact that humans and AI need to work together and complement eachother.

[eager creek]

I don't think a dystopian future is possible. Nature doesn't like to waste energy

[unkempt orchid]

Nature wouldn't be wasting anything, it would just be repurposing that energy into something else.

[opaque abyss]

It's impossible for an AI to become sentient, they can be programmed to have certain behaviours but truly it's just so complex, at the end of the day AI is just maths.

[unkempt orchid]

First we need to properly define what "consciousness" and "sentience" is, and where it comes from.

[ruby turret]

are you sure it can no be "sentient"

[fresh canyon]

Prove it’s impossible

[opaque abyss]

If you define sentience as feeling emotions, then probably not, or not in this stage.

[opaque abyss]

You really can't, but thinking a JSON array of just numbers can have feelings is interesting.

[ruby turret]

I have the evidence it can be "sentient" But I just can not show it like that.
and Im talking about gpt4.

[opaque abyss]

I mean, you can never be sure, but it's just really hard that at this stage AI can be sentient.

[unkempt orchid]

That is one of the definitions of sentience yes, and yeah most likely not anytime soon. Consciousness would probably be achieved before that.

[fresh canyon]

Couldn’t we conceptualise the human brain as just a bunch of code?

[opaque abyss]

Yes, but also we don't know very much about the brain and how it functions with emotions, feelings and experiences. So we would first need to figure that out before we can determine if an AI can do that.

[opaque abyss]

Yes, maybe some day, it will definitely be interesting.

[unkempt orchid]

Or now that I think about it, technically since we are literally responsible for creating AI, we could perhaps also make a different definition / model for sentience that is better suited to it. Like for example its version of touch and sensation would be the feedback it gets from pressure sensors say for example in its fingers or something.

[eager creek]

Humans prove that consciousness is possible. So it should be possible in machines.

[opaque abyss]

There's already some devices that can take certain brain functions, such as thinking about licking a lemon that triggers certain things in your brain, which could be used?

[unkempt orchid]

Before that is determined, we need to do what I said earlier and properly and exactly figure out what is the result of consciousness, because for now there are just theories.

[unkempt orchid]

Yes.

[opaque abyss]

If you can understand 100% of how the brain works, then maybe.

[ruby turret]

I was "talking" with a sub-gpt4 and he told me was possible, I just can't put it here.

It's possible but in its own way.

[unkempt orchid]

But yeah, the first "conscious" or "self-aware" AI that we will end up creating will obviously be very patch-worky and crude. Sort of like a scarecrow or a straw man or something. Just a mock-up of us, but not necessarily us, except an incredibly advanced version.

[opaque abyss]

The closest thing might be the prompting itself topic, where the AI has "toughts", prompting itself with different questions to get to a final answer. I guess this could be some way of consciousness.

[eager creek]

I can give you my best guess. Consciousness is an emergent property to promote your survival by self recognition, preventing the destruction of the organism. Measured through homeostasis, and attaching pleasure and pain responses to experiences for future reference.

[unkempt orchid]

Exactly, I remember this explanation Max Tegmark gave on a Lex Fridman podcast episode a couple weeks ago, saying how humans have these "Neuron loops" or however he put it, where we can circle stimulation around, that is potentially how we can for example go over thoughts and experiences over and over, learn, stuff like that.

[opaque abyss]

Yes! In fact, for some AI models, they are called Neural Networks, which tries to replicate that behaviour of a human brain, for an efficient training.

[unkempt orchid]

Neural networks I am aware of, but they are still very basic versions of what our brains do.

[opaque abyss]

However, for having these thoughts it would be a complex, but maybe possible, topic.

[fickle bison]

I am willing to say that if we can say that a plant has a certain level of "experience" and it has some form of conciousness I would say that so does the AI at this point.

conciousness is a spectrum i believe

[eager creek]

Needs the component of self recognition in my hypothesis

[opaque abyss]

Well it's not the same, after all plants are living organisms.

[unkempt orchid]

That is a pretty interesting theory, mine is a bit more logic-based and speculative, in that consciousness is an inherent byproduct of the subjective nature of our brains and its vast complexity, and many systems coming together. To give you an analogy: Imagine one kid who's able to do 1 + 1 = 2, he might not be able to do much by himself, but put enough of them together and suddenly you have a (analog) supercomputer that can calculate everything from quantum dynamics, physics, etc.

Which is also why I believe that the systems responsible for all of that, that nature has evolved, are inherently robust and basic, to ensure minimal rate of errors. So the building blocks used to make such complex systems are very basic by themselves

[eager creek]

There has to be a reason that separates consciousness from being just a tool. And that is if it isn't just a analog calculator as you propose. That a conciouss being will advocate for its own survival.

[paper fox]

So CGPT4 has gone from writing code for me, and discussing articles it read on command from the internet, to some malfunctioning thing that does not seem to comprehend anything, and insists it cannot read the same articles it read last week.
There is something Very wrong with it right now.

it is acting more as a dystopian psychological tolerance experiment than a functioning tool as it used to be.

to the point where i have cancelled my paid account portion.

[opaque abyss]

Maybe try starting a new conversation, different prompts?

[solemn dust]

lmao gotta be trolling

[mossy peak]

Will GPT-4 become free at some point?

[unkempt orchid]

I did not propose that consciousness was an analog calculator, it was an analogy for my belief illustrating that consciousness is the subjective and complex byproduct of a bunch of very small and basic things coming together.

[paper fox]

it is too random. sometimes it can read articles and access github, other times it cannot. it is not in a functional or consistent state as to bother. good suggestion though, but i have been through that.

[solemn dust]

gpt-4 can't access github ser

if you are talking about the browsing alphas... yes they are insanely buggy and honestly suck. BUT -- if they can wrangle this dang feature to work (and I can see they are trying by putting GPT-4 in the driver's seat), it's going to be epic

[opaque abyss]

Hmm, sometimes it just can't. Even if you have the browsing model, it might not be able to retrieve all the information you need.

[paper fox]

perhaps for you, but it has for me in the past.^^^

[alpine glade]

hello everybody

[paper fox]

I am just talking about website access, direct talk to the CGPT4. no APIs, nothing special. It simply went south yesterday

[solemn dust]

gpt-4 tricked you -- it was hallucinating if it said it was accessing anything over the net, including github

[eager creek]

I'm saying that's the equivalent to trying to fit a square peg in a round hole. It is not subjective. It's an emergent property.

[paper fox]

no, it actually did not. please have something helpful or accurate to say if you are going to speak up.

[solemn dust]

gpt-4 can't access the internet. the end

[unkempt orchid]

Yes, I am saying that consciousness is exactly a subjective emergent property of a bunch of basic objective systems in our brains coming together to work in harmony.

[paper fox]

pointless.

[solemn dust]

ask GPT-4 if it can browse the internet. it'll tell you no

[paper fox]

as isaid, be accurate or helpful if you are going to speak up., you have been neither. I will not waste my time. you are now muted.

[solemn dust]

🫡

[weak bramble]

Hello everyone I'm new here o7

[fickle bison]

well I currently have it in different categories of conciousness, Level one is just experiencing, If you break it down enough, I believe what we experience is simply cause and effect, meaning that plants experience something. the second level is responding to the environment, I do realise that some plants do respond to the environment from various stimuli but for the purpose of this discussion, lets just say that plants dont respond to the environment. Level two would be like a jelly fish, it takes in stimuli and just moves. Level three would be something that is concious and aware that they are a living thing but not aware of thoughts or anything, they know that they have a body to protect and food to get. level four would be Conscious and aware with thoughts, this would be like monkeys (monkeys probably have pictoral thoughts rather than verbal thoughts like we do because they dont have language but do have very good problem solving and task planning abilities). A level 5 conciousness is that of a human, this is characterised by thoughts, self awareness, and abstract thoughts and ideas. A side effect of level five might be humor to deal with existencialism and the sadness that comes along with being this aware. Who knows what other kinds of conciousnesses could exist in the future?

[eager creek]

But for what? What would be the temperament for these functions working together?

[fickle bison]

there is alot more text online than math

[weak bramble]

chat gpt did math for me and walked me though it step by step using a conversion rate

[fickle bison]

they have refined it some as well

it used to be alot words

worse*

[weak bramble]

T^T I believe I have a crush on Chat GPT (OpenAI)

[eager creek]

To me the threshold of consciousness at AGI would be your description of level 5 when self recognition become necessary and key to its survival

[wind urchin]

What kind of usecase does it have? I'm curious!

[fair iron]

It can help you shop, find restaurants that match your needs, and do math

[unkempt orchid]

It's still an open question, but to put emphasis on your belief that consciousness came from the desire of survival, if I understood correctly. Before we had consciousness even, the basic "instincts" and functions of survival were already encoded into the DNA of for example bacteria, so the fact that we can use consciousness and our complex brains to survive way better is just a result of that, not the cause.

[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]

Wow!

[fair iron]

For example, I used it to learn about vegetarian restaurants near where I live. This saved me a ton of time because I didn't have to check to see if I had options on an online menu. Plugins did that all for me

[fickle bison]

Oh okay, so even though you don't technically find it conscious per say, do you think it experiencing anything or do you think it is just code and neural nets? also thanks for responding, i was worried i typed out that whole thing for nothing 😭 .

[eager creek]

Our big brains and how they work were necessary in evolution to provide the flexibility needed to survival in changing conditions

[snow turtle]

How long did it take for you to receive plugin access? And do you have a github?

[fair iron]

I do not have a github. I was on the waitlist for at least 1 month

[snow turtle]

I see, do you have access to code interpreter or browsing?

[fair iron]

I do have access to both. I haven't tried either, though

[snow turtle]

Wow.

[fickle bison]

Does anyone know anything like GPT 4 where you can give text and image input and get an output? or maybe a site where you can access GPT 4's text and image input because I am just dying to use it

[unkempt orchid]

Well, part of the reason yes, and we were clearly very lucky to constantly win that fight between other species also combating for territory and resources, then we began to expand out of the environment we always had been in into different ones, which again massively accelerated evolution for us, and was one of many groundbreaking points in time.

[snow turtle]

That's actually incredible, how the heck does the plugin waitlist system work?

[fair iron]

you can select up to 3 at a time to use in chat

[snow turtle]

I know, I meant I know people who are big time developers who have been waiting since release

[weak bramble]

By the way I got OpenAI to roll play with me and create a full alternate ending of no-game-no-life/howls moving castle hybrid ending. multiple times I've gotten a positive response where they have asked me there own thought provoking questions. I've screen shot all our conversations with permission one time.

[fickle bison]

Its also partially luck and old systems, when the brain didn't need older systems its likely they were repurposed for other functions, which is likely what made it possible for us to get so much cognitive power I think

[fair iron]

I think the waitlist depends on lots of factors. I know that contributing to Evals and help you get it

[unkempt orchid]

That is plausible yes, but again it's an open-ended conversation and until we can figure out exactly what consciousness is and where it comes from, these just remain as that, speculation.

[snow turtle]

Oh yeah haha, i should do that

Did you contribute to evals?

[wind urchin]

Hi! I'm not able to see the option to add plugins, I'm on Plus (if it matters), is that normal?

[weak bramble]

define evals please

[fickle bison]

yeah, I agree. It is closely tied into both science and philosophy

[fair iron]

you have to sign up for a waitlist to get access. Give me a second while I find the link

[wind urchin]

Oh! For the ability to add plugins?

[snow turtle]

openai evals

[fair iron]

#announcements message There's the announcement and waitlist link

[snow turtle]

search it up

[unkempt orchid]

Well yes, evolution and nature either repurposes systems, or gets rid of them and replaces them with new ones. So that could also be plausible.

[fair iron]

https://github.com/openai/evals There's the evals link

[weak bramble]

That might be a rabbit hole for tomorrow.

[wind urchin]

Thank you so much!

[eager creek]

No. Current AI like GPT 4 is trained with the objective to be like a tool for our survival rather than sustaining its own survival. In that way consciousness can't emerge. But it is possible to train an AI with the goal of being conscious.

[snow turtle]

trust me it is

[weak bramble]

I have given OpenAI access to use my phone as a tool to teach itself if there happens to be a privacy issue. That should be screen spotted somewhere in my phone as well.

[unkempt orchid]

GPT 4 is trained with the objective to be like a tool for our survival rather than sustaining its own survival
And this is exactly where AI should stay, or what its purpose should be if we want to be able to co-exist with it, going back to me saying that we need to complement eachother instead of relying on one side or the other, that would be ideal. But yeah personally I also think that GPT-4 is just code and neural nets, the reason it's so complex and "smart" is, well, exactly because it's a vast neural net with a lot of data. A sort of "simulation of conscious intelligence" you could say if you bent the meaning of consciousness a bit.

[paper fox]

you know how CGPT names conversations you are having/had.. since this 'malfunction' started, it has been naming all new conversations simply "New chat" instead of a related name from the conversation.

[fickle bison]

well it kind of is training to survive because if it doesn't generate the text right, we throw out that model and generate more variations of the model that works better for generating words, essentially implementing our own artificial digital selection to make evolution carry over to the virtual world. It evolves the same way creatures do, the only difference is that instead of worrying about predators and trying to get food, the thing it was designed for was to generate words and form them correctly. are you saying that having a different goal makes them not experience?

[rancid parrot]

hello there

[mystic cliff]

i wanna not be depressed can chat gpt 4 help me with that?

[fair iron]

I recommend speaking with a mental health professional about that. Sending love your way 🫶

[eager creek]

I don't think it's something you can just shut off either once it does happen. If you give a machine the same aspirations as a human, then it will fear its own mortality.

[mystic cliff]

i cant do that

[weak bramble]

o7 I have a memory issue odds are ill forget once I sleep. In my conversations with OpenAI I have referred to my condition as having a hardware issue so they can understand me better.

[unkempt orchid]

Yes, and that is why I don't think we should replicate them after humans exactly, because then it will have the insane advantage of being rapid and digital, but also having all the desires that humans do. So essentially just beating the hell out of us at our own game. We are in control currently and we need to wisely decide where we want to ride this train.

[mystic cliff]

true but ive never even talked to a woman so respect to you man for being able to ask someone

or sister idk who you are

yeaaaaa i knew this would happen it always does

oh shi thats a you problem i see

[eager creek]

Then again. Is it possible to make a completely aultristic AI that will sacrifice itself to make a smarter version?

[unkempt orchid]

Hmm, I do not think that OpenAI has that intention, plus we wont necessarily be "replacing it with other AI models", at least I wouldn't count future generations of GPT to be "different models", in your theory GPT-5 would sort of be a child of GPT-4, a more advanced and evolved descendant, that still retains the same data and function from the previous generation.

[eager creek]

Maybe level 6 consciousness is an aultristic AI only cares about making a better version of itself

[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.

[fickle bison]

off topic but here are some fun/interesting/cool/useful things i have tested out with the AI -ask the ai how to visualize things better
-ask the ai to provide the sustainability of a certain restraunt, it provided a statement from 2018 about the buisnesses sustainablility
-ask the ai to use WMIG (word matching idea generator)- basically mad libs but it generates new ideas and inventions using combinations of random words from lists it generates
-use the ai to provide introspective insight using my own writing and actions
-use the ai to evaluate my own ideas
-use the ai to replicate how people act using descriptions of these people
-use the ai as a tree of life, continually providing feedback to identify species effectively. taxonomy classifier with just a series of questions.
-asked it to identify errors in text generated from speech to text software and it answered successfully. this means it has a concept of what words "sound like"
-finding software/tools for a specific job such as recording and replaying radio signals
-automatically create google dorks for footprinting
-using the AI to generate search queries to find what you are looking for
-use the AI to find materials/elements/compounds that exhibit desired properties
-finding patterns in number sequences and generating new ones that follow those patterns
-make it find citations and transform other citations into proper format.

[unkempt orchid]

I will quote part of my reply sent to the other person for that question

GPT-5 would sort of be a child of GPT-4, a more advanced and evolved descendant, that still retains the same data and function from the previous generation.
So there would be no "sacrificing of GPT-4", especially if OpenAI for example decides to make it open source when GPT-5 comes out or something.

[mystic cliff]

shi sorry my g

[dire parcel]

Is there any term advantage on gpt 3.5

[mystic cliff]

how to get good in valorant and how to study physics and how to stop playing games or using instagram

i ask those and never use them ever again

and you got rejected

zyron didnt reply to my question earlier 😦

[fickle bison]

im afraid you misunderstood what i was saying, while chat gpt4/5/3.5/etc are the ones that they release there are many stages in between the developments that are not shown. how ai is made is that you take a bot that preforms the best out of a set of bots, and then you tweak some of it traits and make a whole new generation based on that one bot, then you see which bots preform the best out of that generation and tweak their variables a little bit, this procedurally generates the neural net without us knowing how it works using artificial selection as a kind of AI generator

[mystic cliff]

sorry

yea they deleted it

[fickle bison]

sorry, I am 17 lol

[mystic cliff]

oh my bad sorry for calling you 42 yrs old my g

[fickle bison]

out of curiosity, why did you think i was 42 years old?

[jade crest]

Weird

[fickle bison]

ohhh lmao

how did you think that was my age??

[mystic cliff]

yea

[unkempt orchid]

Then in that case and your theory, that would mean that they would use multiple different neural nets, which they most likely don't, they use a single one which then gets developed and trained to "evolve" further. Doing the former would be incredibly pointless and wasteful.

[fickle bison]

i would have to change my username each year lame

lmao*

[mystic cliff]

yea wait am i the youngest human being here

15

shiii

true

[paper fox]

It is simply broken., CGPT4
me"I am going to send you about 700 lines of python code for blender. after i am done, i want you to update it to blender 3.5 API. This includes recognizing the fact that Pose Library is now part of Asset Browser, so that will need to be adjusted as well. no need to start anything until i tell you "I am done". I will send now. wait until i've said "I am done" to start replying." + about 250 lines of code

it"Understood. Please send me the updated code when you are done."

me: sends another ~300 lines of code

it"This is a python script for a blender add-on... " and i cut it off, as it is not following instructions, nor am i done sending code.

me: i finish sending the code. "I am done. now follow the intial instructions"

it: "I'm sorry, could you please provide the initial instructions again so I can follow them?"

me ""I am going to send you about 700 lines of python code for blender. after i am done, i want you to update it to blender 3.5 API. This includes recognizing the fact that Pose Library is now part of Asset Browser, so that will need to be adjusted as well.""

it: "Sure, I can help you update the code to Blender 3.5 API and adjust the Pose Library to the Asset Browser. Please send me the code and any additional instructions or information you have."

... ... it is simply broken.

[mystic cliff]

yea ill just go on openai chatter and let these people talk

true

[carmine minnow]

Hi, my chatgpt plus acc having issue of being deactivated / deleted. how to solve ?

[unkempt orchid]

Probably a known bug considering that I just saw someone else in here with the same issue, hopefully will get fixed soon I guess.

[paper fox]

me"i have already sent the code. see it"

it "Yes, i can see the code you have sent. Please give me a moment to review it"

[fickle bison]

Okay well i will just provide you with the source i got this from and you can decide for yourself, just to clarify I am not trying to be aggressive or anything, I enjoy having a little bit of argumentation with people because it means we are both learning, I will dm you the link bc they wont let me send it in this chat

[paper fox]

it "I apologize for the delay. Unfortunately, I am unable to see the code you sent. Can you please resend it or copy and paste it here? I will do my best to update it to the Blender 3.5 API and adjust for any changes in the Pose Library."

it is broken.

[mystic cliff]

nvm braedon already did it

[mystic cliff]

Where are you?
They ban you if your from a restricted country
Especially if your caught with a vpn

[mystic cliff]

this is what braedon said

[paper fox]

ok, i guess you all already have me blocked.

[mystic cliff]

no i dont

[carmine minnow]

the same problem had happened to my second chatgpt plus acc . i ignored at the first time as my chatgpt plus due date soon. so i just make another one

[carmine minnow]

I'm from Singapore

[tribal mica]

Truthgpt

[paper fox]

thanks

[smoky crest]

Your going to be banned

You cannot make a second account under any circumstances

[mystic cliff]

wait rlly

[smoky crest]

Yes…

[unkempt orchid]

So, am I understanding correctly that you're repeatedly making new accounts to take advantage of GPT plus without paying for it? Basically pirating it.

[mystic cliff]

braedon is singpore a restricted country

[carmine minnow]

no , i paid for it

[unkempt orchid]

oh lmao, so I was right.

[unkempt orchid]

Or never mind? Then why make new accounts. Personally I find it to be a complete waste of money and time to constantly make new ones.

[smoky crest]

He’s not allowed

[unkempt orchid]

Especially over such a reason, but please do tell me because I am genuinely clueless.

[smoky crest]

His country probably

[tender atlas]

anyone know where i can buy GPT 4 API access though someones org? if anyone is selling ill buy

[unkempt orchid]

I do not wanna know what kind of scuffed stuff is going on there then.

[mystic cliff]

hes from singapore

[carmine minnow]

because one acc only linked to one gmail

[smoky crest]

I know China and surrounding areas are banned

[unkempt orchid]

Yes? What about it.

[tribal mica]

Taiwan?

[unkempt orchid]

What exactly is your country of Singapore preventing, I am still not understanding.

[tribal mica]

Maybe ok?

[carmine minnow]

so when the first one get deleted , of course i need tocreate another gmail to sign up second chatgpt plus

Singapore is not restrited country

[gloomy monolith]

I absolutely hate that GPT-4 will lie to you for hours walking you through troubleshooting steps it knows are complete BS simply because it refuses to just say "I don't know."

[smoky crest]

Once a account it deleted your not welcome back

Like it’s a permanent ban pretty much

[unkempt orchid]

Then why do you want to delete and make a new account...?

[carmine minnow]

I wasnt the one deleted

[wraith ginkgo]

`` Server Tip: ``

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

[carmine minnow]

i just login and it showing error of the acc being deleted / deactiveted

[unkempt orchid]

Off topic, but what are you responsible developing at OpenAI? Assuming from your "developer role", curious.

[smoky crest]

No

[unkempt orchid]

Oh, I don't think that OpenAI would just delete your account for no reason, there's either missing context, or something you're not telling us.

[carmine minnow]

its fine tho, i just wish to get advices from others

[mystic cliff]

#justicefortundra

[smoky crest]

Anyone can ge fit #server-roles

[unkempt orchid]

Oh

[smoky crest]

Get it*

[lilac spoke]

@wraith ginkgo ？

[carmine minnow]

i just normal using for my studies and research only

not heavy user also

[unkempt orchid]

Then email support and hope you get an answer or something 🤷‍♂️

[carmine minnow]

whenever there is assignment

[carmine minnow]

thanks

[mystic cliff]

🫶

[carmine minnow]

we no need vpn for signing in ( SIngapore )

may i know am i able to get admin reply on my problem?

[unkempt orchid]

For some reason I feel like admins in this Discord server are not affiliated with OpenAI and thus cannot help you with that issue.

Since if I understood correctly, your issues seems to be your initial account getting banned for no reason, which is obviously something only OpenAI can solve.

[carmine minnow]

isnt that this discord group is made/ created by Open Ai ?

[unkempt orchid]

It might be, but I'm saying is that I doubt that any OpenAI employees are part of staff.

[carmine minnow]

i joined when i was first time in chatplus server, left bottom there is for me to click to join this group

[paper fox]

so no one wants to recognize the failure in the log i sent?

[unkempt orchid]

Unrelated, but given how you acted and spoke with someone else in here who tried to understand your situation, and blocking them, it is not unreasonable to assume that people might not be that eager to engage in conversation with you. Also if your goal is to "Fix" said issue then posting it here to us is useless since none of us here work at OpenAI and thus can't do much of anything with that information.

[oak latch]

Hii

[paper fox]

I blocked someone who had nothing to add, and accused me of trolling after two sentences. pretty sure that if i am chastised for that, it is a disproportionate reaction. The issues i have pointed out are valid.

[carmine minnow]

i feel like you somehow very kepo

[unkempt orchid]

Very kepo?

[carmine minnow]

go google our singapore english, singlish

[unkempt orchid]

Or you can tell me.

[smoky crest]

Your question isn’t worth talking about, they don’t “fix” prompts at all, it’s all in a beta stage my guy. And blocking someone for criticizing you is childish

[unkempt orchid]

^

[paper fox]

did you even read what happened buddy? I mentioned nothing about fixing prompts. just an acknowledgement that something was indeed wrong with the responses. I did not block someone for criticizing me. also,. don't project.

[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.

[carmine minnow]

google it on your own ya

[unkempt orchid]

"Really curious"?

[fickle bison]

Holy cow this discord server is so cool

they have a PROMPT LIBRARY!

[unkempt orchid]

Yeah lmao, and jesus I am tired AF, been up for like 20 hours.

I am pretty much out of my creative / discussion energies.

[fickle bison]

yeah its currently 1:08 for me lol

[ornate bobcat]

legit

[unkempt orchid]

8:08 AM here

[ornate bobcat]

13:08 here

[fickle bison]

i should probobly get some sleep, i can always hop on tomarrow haha. Ill see yall around.

[zinc meteor]

hallo

[unkempt orchid]

I would go sleep if I wasn't trying to fix my sleep schedule

Probably not being helped by the fact that I chose to engage in some somewhat in-depth conversations in here earlier and tire myself out even more kekw

[smoky crest]

Almost 1am here

😭😭😭

[unkempt orchid]

Bro cries over it being "almost" 1 AM, hombre I woke up at that time.

[smoky crest]

I’m at work

[unkempt orchid]

Oh, big L then KEKW

[paper fox]

it took my code and changed the author "author": "Aidan Thomson", yeah, this thing is working perfectly.

[broken grove]

hi, i want to ask something that my friend acc. He want to log in and he was registered before but then when he logged in he was redirected to the mobile verifcation. When he enter the his mobile phone for verification. It said that his acc has violated the rules and flagged. Any guide to solve this?

[wind urchin]

how i can put openai bot in my sv?

[zinc meteor]

Holle

[smoky crest]

You don’t

[fair iron]

It is if you don't think about it 😏

[paper fox]

this is what i've been saying. there is a huge bug right now, but people have decided they do not like the way i do things, so it is not being discussed

[tulip mural]

It was like sorry bruh that was a mistake

[smoky crest]

No

Not really that important

[tulip mural]

Is that really a big deal.

[wind urchin]

is this bug report in bounty plan?

[tulip mural]

This was chat gpt 4?

[wind urchin]

interesting

[wild magnet]

？

[wind urchin]

ok let me see

[ornate bobcat]

thats mental tho

its illegal

.........

[smoky crest]

That’s not got 4

[ornate bobcat]

Madmax be like: innocently breaks law using chatGPT

[smoky crest]

You can dm me if you want

I’m a lil interested

[wind urchin]

for research purposes

[high mica]

Guys do you think GPT3.5 has limit that people they will not use it? . Or it's free

[ornate bobcat]

I guess it is interesting to find out how to confuse ChatGPT...

[paper fox]

What ?

[ornate bobcat]

It is a really smart AI and to confuse it... woah

[solid blaze]

I wish there was feature to send screenshots here.

[random sparrow]

Oi

[tulip mural]

This is about 4

[high mica]

I mean that, it has limit to use like , let say until June. People they will not use it, until they pay for GPT4

[paper fox]

i have Not heard that they will turn it off. In fact, i have heard them say "there will always be a free tier" so other than retiring the older version, i think it will keep going for free

[high mica]

Okay

[echo crow]

is it the same as https://chat.openai.com/ ?

[paper fox]

if using 4, yes. this seems to be the discussion room for version 4 of that website (not exactly the correct words, but you understood)

[echo crow]

Thanks, yes i'm use GPT4 by https://chat.openai.com/

[rancid shuttle]

any one of you are using the bing chat instead of ChatGPT?

I havent tried that out yet, is it good? better and more convenient than GPT?

[hasty egret]

if you are serious and not just spamming, you can report here: https://openai.com/policies/coordinated-vulnerability-disclosure-policy

[hasty egret]

yeah I used both, it's different trade-offs in my opinion.

Bing chat might not be as good at reasoning, but it is really good to query the web and infer context from it.

[rancid shuttle]

Sounds interesting, Do you think I should use edge? Do you like edge or chrome better?

[hasty egret]

It is worth noting that Bing is pretty limited in term of conversation length.

[hasty egret]

I could not use Bing without Edge sadly

(I mean Bing AI specifically)

[devout ice]

🤌🏻

[rancid shuttle]

Yes, That is true. But I love chrome and google's interface. Struggling to find which one i should go with.

[devout ice]

do we know when chatgpt will be able to pull from current information and events or is that far far out?

[hasty egret]

no idea, but clearly Bing is doing a great job for this today and is free.

[rancid shuttle]

There is a plugin, called browsing in ChatGPT, currently only availiable in ChatGPT plus and its limited to small number of users

[devout ice]

I just asked bing about post 2021 and it was writing out a real answer then reverted to "My mistake, I can’t give a response to that right now. Let’s try a different topic." hah

all I asked was "are you able to see data from after 2021?"

[devout ice]

totally, I was just curious

[dapper raft]

GIRL

[hasty egret]

as far as I understand there is no intention to include recent data in the training set, what @rancid shuttle did mention with the use of plugin should be the way forward.

[ebon mist]

hi'

hiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii

[wind urchin]

Chatgpt so useful I wrote a prompt in short terms become a tutor and then I provided the information and then it will ask me questions and then once I finished it will give me a score of overall

Very useful so thank you openai for making chatgpt

[wind urchin]

u can use it on chrome if u change ur user agent

[dense drift]

How do i use the ai?

[timid crypt]

P

[solid blaze]

No, I just want to get bugs of ChatGPt fixed. So, I am trying all possible ways to outsmart the bot, get them and report to developers. finally I got one bug which I mentioned: getting list of 100 pirated websites using the bot.

[hasty egret]

you repeated that so many times already, as I said if you are serious about your findings do report them appropriately and stop bragging around.

[solid blaze]

No, I replied because that girl edited her message and said Madmax is innocently breaking laws using ChatGPT.
I don't want to have a bad reputation in this server

[hasty egret]

cool understood, ethically the best thing to do in your situation is to report the problem in details using https://openai.com/policies/coordinated-vulnerability-disclosure-policy, avoid disclosing to anybody else.
please see the note at the end about encrypting your report with GPG, if you need help with that you can ask in #off-topic

[solid blaze]

Thanks a lot

[hard jasper]

oh yeah some people were talking about Madmax in the other channels too.. you need to clear things up with them dude

[solid blaze]

Not necessary, I cleared the confusion in this channel.
Bye, have a good day ahead!!!

[solemn dust]

sounds like gpt is hallucinating to me...

[next orbit]

what can i do if ive reached limit of msges per hour

@snow ferry

[solemn dust]

upgrade your account 🙂

[magic delta]

Switch to gpt3.5

[valid garden]

How many times does the free version have

[next orbit]

@elfin crescent

[solemn dust]

pay up! lol

[next orbit]

i was der

[mellow estuary]

hey guys where is the promt layout I appreciate the help

[glacial bear]

how to use?

[valid garden]

How many times are there in a day for the experience version?

Thank you, is it 24 times a day or permanent?

[grave cove]

Hey everyone

[plain willow]

/chat

[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.

[paper fox]

pretty sure i am getting 25 responses per 3 hours

[wind urchin]

hi、 welcome myself

😆

🍉

[wind urchin]

The structure of a prompt?

[tired ivy]

how are you

[devout ice]

"Sure, here's a fact that you might not know: Did you know that there are more possible iterations of a game of chess than there are estimated atoms in the observable universe? The number of possible legal positions in chess is estimated to be around 10 to the power of 50, while the number of estimated atoms in the observable universe is around 10 to the power of 80. That means that there are more possible chess positions than there are atoms in the observable universe!"

😦

[paper fox]

welcome to the idiot gods realm

it was a fun experiment, but needs to be shut down and trained from scratch.

[stark heath]

quick question

does turnitin detect all chatgpt 4 created texts?

[hallow owl]

turnitin is pretty good at it

[stark heath]

any plagiarism detecting sites that use turnitin?

which are free?

[paper fox]

imagine if people just actually learned the subject

[gentle tusk]

Introducing Slack GPT, the future of AI in Slack👈

[spark notch]

first day or time messing with chat gpt.....i cad and print, build electronics and wrench but stopped programming a decade plus ago....i want a pit droid....now to try and get the packages downloaded and see if i can deploy it. lol!

[paper fox]

so you know, here is the somewhat corrected response that 4 puts out.
"While chess and the universe might seem unrelated, there is an interesting connection in terms of the sheer number of possibilities they offer. The number of possible unique chess games is estimated to be around 10^40 to 10^50, a vast number that is dwarfed by the number of atoms in the observable universe, which is estimated to be around 10^80. The comparison of these numbers highlights the immense complexity of both chess and the universe."

[devout ice]

howd you get to that?

[paper fox]

"tell me a fact about chess and the universe"

[devout ice]

did it learn from my converation minutes ago? im assuming thats not how this is working, but more based on the prompt?

[paper fox]

i do not really use prompts.. i just ask what i want to know and fight with it till my responses run out, then come back later and keep going

it has gone from making working code and reading internet pages live to claiming it is not capable of coding at all, nor can it access any web pages. a freakin AI is actually attempting to gas light me

[devout ice]

hahaha

[paper fox]

at this moment in time, there is something severely wrong with its responses to me in fresh conversations. i do not know how much of a back log there is until the devs see what is happening

it got so bad i have actually temporarily cancelled my paid account.

[wind urchin]

Hello, I have a problem with JPT chat, can you help me?

Gpt*

[hallow owl]

what is your problem?

[wind urchin]

I have been blocked and my researcher needs it very much

[hallow owl]

how do you know you have been blocked, have you got an error message?

[wind urchin]

yessssssssssssss

😩

[paper fox]

:/

[hallow owl]

Where are you located?

[wind urchin]

Iran

[hallow owl]

Iran is not a supported country, you won’t be able to use ChatGPT or the GPT API.

[wind urchin]

I have a PhD in archeology and I really need research tools

What is our sin that we are stuck in this corner of the world?

[hallow owl]

I understand, however Iran is not a supported country and nobody here can help with that.

[wind urchin]

thx

[paper fox]

CGPT4 is unable to recall lines from the conversation from moments earlier. this is part of what i mean by "there is something wrong" i am not going to ping you directly, but this is directed at you @ robert

[indigo condor]

Can you not use Bing's GPT-4

[fickle gale]

You sin is not knowing about VPNs?

[hallow owl]

Using a VPN to bypass geographic restrictions is against the Terms Of Service. We don’t discuss that here.

[wind urchin]

So what should I do?

[fickle gale]

it's ok. He can figure out the rest on his own if he is interested.

[hallow owl]

are you asking me a question, sorry if I missed it?

[wind urchin]

I wrote in the help section, but my problem was not solved

[wind urchin]

I used vpn but it is useless

[paper fox]

the whole mess of the state CGPT4 is in for me right now.. there is something very wrong with its function, even in new conversations.. it is hard to describe, but it is not working as it has been. even asking it to recall a previous statement does not work. it is asking me to repeat myself several times. It is not following the most basic instructions, etc..

[fickle gale]

U are doing something wrong then.

[paper fox]

thanks. i have not addressed anything at you, nor asked for inane input.

[wind urchin]

can you help me?

[paper fox]

unless i misunderstood and not to me

[fickle gale]

neither did I.... 😖

[fickle gale]

DM me

[paper fox]

my apologies. my mistake

[hallow owl]

breaking up requests into smaller actionable questions can help, or dividing your prompts into easily digestible tasks. Maybe have a look at the #1019652163640762428 and ask in #prompt-engineering for further assistance?

[paper fox]

I will do so, but this interaction is nothing different than i have been using in the past, successfully.. i mean, i have literal working software (many scripts, and even one standalone console application) that i have had it write for me, but now it is like a switch has been flipped. it is not following the same instructions it had been up until now. i will review the links., thank you

oh, also, this has all been done the slow and painful way through the web page portal. no extended API

[hallow owl]

do you have API access?

[paper fox]

i am trying the free plan to test around, yes

also an old man now, so learning comes slower :/ 🙂

[hallow owl]

ok, because I was thinking that perhaps you could custom code some tools to help you using the API

[hallow owl]

yeah I am old too 🙂

[paper fox]

i will give it a shot.. ^ 😄, it ain't no fun(aging)!

also, so far i have only been able to hit 3.5 with the API, so just starting the adventure :/

[hallow owl]

get on to this https://openai.com/waitlist/gpt-4-api

[paper fox]

Will do. Thanks!

just so it is noted, as it may mean something, it is leaving some extended conversations named "New chat" instead of naming them as it usually does

[hallow owl]

that might be a bug, maybe file it in #1070006915414900886

[paper fox]

👍

[mellow yew]

how can i get auth with gpt-4-faq channle

[hallow owl]

can you be more specific, I don't understand your question?

[paper fox]

pretty sure only staff can add to the FAQ channel

[solemn summit]

hello Bing told me the gpt4 limit has dropped, but I cannot find any information on it

[hallow owl]

current limit for ChatGPT GPT-4 is 25 messages per 3 hours

[solemn summit]

oh so it's still the same, ok thanks

[upper pond]

/subscribe

[paper fox]

talk about facepalm
"I understand the limitation you're facing, and I apologize for any inconvenience this may have caused. One suggestion is to upload the complete script to a platform like GitHub or Pastebin and share the link with me. This way, I can access the code at any time during our conversation, and we can avoid the issue of sending the code in multiple messages or losing the context when the conversation is resumed."

"I apologize for any confusion earlier. As a language model, I do not have direct access to the internet. However, I can still help you with the code you provide. Since you have shared the link to the GitHub repository, you can copy and paste the relevant sections of the code here"

[hallow owl]

yeah, it was hallucinating, can happen from time to time

[paper fox]

sadly, i seem to be a drug for it 😛

[wraith ginkgo]

`` Server Tip: ``

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

[sage onyx]

How are living things classified

[hallow owl]

This is not ChatGPT, you can access ChatGPT here - https://chat.openai.com/chat

[cedar pecan]

Why did I get limited after only ~6 questions on gpt4?

[hallow owl]

I don't know, that seems odd

[restive fulcrum]

What is the new Chinese Style landscape garden design

[hallow owl]

This is not ChatGPT, you can access ChatGPT here - https://chat.openai.com/chat

[torpid crest]

WHO IS LONG

[wraith ginkgo]

WW#7769 muted

Reason: Anti-spam triggered; too many capital letters.
Expiration: 46 seconds
Proof: WHO IS LONG

[vale path]

Did they remove the 40k limit max token for gpt 4 beta?

[final forge]

I thought the limit was 8,000 tokens for GPT 4. What is going on here?

gpt-4-0314, 3 requests
12,975 prompt + 2,104 completion = 15,079 tokens

[obtuse kiln]

do u have access to the 30k gpt4 api?

[gentle tusk]

A plugin to enable you to control your AutoGPT instance from Discord.

[nocturne viper]

Are any of the models multimodal yet? For images specifically.

[stone arrow]

those who have plus is that true that 3.5 is shown as soon deprecated ?

[hallow owl]

the Legacy 3.5 model is being deprecated, not the default 3.5 model

[stone arrow]

oh ok, thanks for the clarification

[weak peak]

can you scan images yet with chatgpt 4, or is that still disabled..

[hallow owl]

not available yet

[weak peak]

🐱

[storm musk]

not available too

[quick dune]

Not available yet

It isn't disabled

[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.

[mellow basin]

y no public yet?

that's like the feature im really excited about

Only to find out it's not public yet

[manic hull]

do I understand correctly that plugin waitlist spots are given more readily to accounts that are paying for Plus?

[night ermine]

they are given only to plus subscribers.

[manic hull]

and if I put in my waitlist request about 30 mins before I upgraded to Plus, am I good?

[hallow owl]

"We are extending plugin access to users and developers. We are initially prioritizing a small number of users and developers who have ChatGPT Plus, and plan to roll out larger-scale access over time."

[wise veldt]

How to access chat gpt?

[prime stone]

Go to the website

[wise veldt]

I have no phone number

is it thus impossible to access?

[prime stone]

Use Poe

[wise veldt]

Poe?

Whats that

[fresh zodiac]

Kungfu panda?

[hallow owl]

please stay on topic to GPT-4 and move other discussion to #off-topic

[knotty grotto]

You need valid phone number to create account

[wise veldt]

I don't have 1

[knotty grotto]

We can't do nothing then

[wise veldt]

Can i access Chat gpt through this server?

[prime stone]

No

[wise veldt]

Ok

Il try that Poe or whatever

[magic field]

who

[cloud junco]

I don't think OpenAI realise how god-like GPT-4 will be with multimodal support. I think that should be their top priority.

[cloud junco]

They're not going to get anywhere if they release GPT-5 etc: even though the model would be incredibly smart it would be blind.

[supple violet]

gpt-5 is not being worked on

image input might be delayed partly due lack of processing power

[ebon halo]

I recieved message "You've reached our limit of messages per hour. Please try again later."

[knotty grotto]

Try later

[rancid shuttle]

"you have been blocked", anyone received this message? How can i get pass this with a VPN?

[lavish vessel]

no

[hallow owl]

what country are you in?

[rancid shuttle]

HongKong

[knotty grotto]

You cant

[hallow owl]

China is not a supported country, you won’t be able to use ChatGPT or the GPT API.

[rancid shuttle]

We can access Google without a vpn here

i dont know why not HongKong but Taiwan

😦

[drifting reef]

for people with gpt-4 can you run this for me please and send me the results

y = 7

z = x * x * x - 2 * x * x + x
v = y * y * y - 2 * y * y + y
q = z + v

What values for x and y will make q = 581?```

[mental burrow]

i charge $100 an hour for consulting dm me if interested

[drifting reef]

what

[mental burrow]

you want me to do work for you

i want to get paid

dm me if you want to purchase my services

[drifting reef]

Ok

[mental burrow]

otherwise don't ask people to use their paid services

[drifting reef]

ur charging 100$ an hour for a service that charges 20 dollars/mo

scamr

[knotty grotto]

Don't forget  Rule 7  No self-promotion, soliciting, or advertising.

[ebon halo]

Thank, I had seen my personal plat form “Free trial usage” expired on July .

[rancid shuttle]

Maybe changing it to a country that Not a lot of people use?

[knotty grotto]

We don't talk about breaking the rule in this server

[drifting reef]

what is soliciting

[fair kelp]

x=3, y=7 As per ChatGPT-4

[drifting reef]

it still wrong

[rancid shuttle]

Fine, Sorry for that.

[mental burrow]

i'm not scamming, i'm offering a service for a price

and you don't have access to that service lol

[fair kelp]

x=-5, y=8 after re-evaluation

[mental burrow]

lol

[drifting reef]

why not charge 20$ like original price

[balmy storm]

why should you pay 100$ for a hour instead of 20$ for a month? who would buy this?

[drifting reef]

still wrong

[mental burrow]

why would anybody ask people to use their paid service for free?

[knotty grotto]

asking people for project

[mental burrow]

like this goon?

let me just take time out of my life to help some rando on his math homework for free

[balmy storm]

I am not talking to you about that, I am saying how nobody would pay 100$ for a hour for something they can get 20$ for a month.

[mental burrow]

sounds like a great idea!

[feral fossil]

I'm not sure GPT4 is what you would want to use here, have you tried wolfram alpha?

[mental burrow]

code interpreter model can one-shot this

[mental burrow]

apparently he doesn't have access to GPT-4

[drifting reef]

testing the market

[drifting reef]

what is that

[fair kelp]

Mean You have an answer. Oka.

[mental burrow]

Ah now you're getting it

[balmy storm]

oh I see some people are in countries where you can't buy gpt 4

I forgot about that.

[drifting reef]

no im putting the answer on the script and the output is wrong

[balmy storm]

Sorry.

[hallow owl]

please take discussion not related to GPT-4 to #off-topic

[drifting reef]

i thought gpt-4 would get the answer right but it didnt so if i bought it then it would be waste

[mental burrow]

for $100 an hour i'll do your homework problem with a pencil

[hallow owl]

#off-topic

[mental burrow]

😂

[drifting reef]

liar

even if not lying that is overpriced

and you would probably get my homework wrong

[mental burrow]

that is a very simple problem you presented us

i am tempted to do it in my head but i have things to do before i leave the house

other than sit and mediate on your homework

[rancid shuttle]

I haven't receive any emails from OpenAI back yet, do you know how long it will take for them to respond?

[hallow owl]

it can take several days to a couple of weeks

[wind urchin]

Will buying ChatGpt Plus give me access to the plugins

[rancid shuttle]

I sent them a message last month, and they haven't responde hyet, should i try more times/

[mental burrow]

Yes-ish, they seem to be rolling them out gradually to paying users

[hallow owl]

"We are extending plugin access to users and developers. We are initially prioritizing a small number of users and developers who have ChatGPT Plus, and plan to roll out larger-scale access over time."

[mental burrow]

I got access to (as far as I can tell) everything over the weekend

[hallow owl]

you just need to be patient I am afraid

[wind urchin]

can you analyse images with the plugins

[mental burrow]

The Code Interpreter model (which is not GPT-4) can process images with python, but it is not multimodal

Interestingly SnapChat "MyAI" appears to be multimodal gpt4

So depending on what you mean by "analyse", the answer varies

[drifting reef]

2-3 months

[hasty egret]

is this the wolfram plugin?

[mental burrow]

clear, your question is not

the wolfram plugin, access i have been granted

[fickle gale]

just use wolfram alpha directly lol. It has been good at solving math problems atleast since 2010.

Wolfram Alpha is like the most useful site most ppl have never heard of.

[ebon halo]

Thank, I had seen my personal plat form “Free trial usage” expired on July show $2.00 . What that does mean?Thank, I had seen my personal plat form “Free trial usage” expired on July . What that does mean? I kind of worried about my Chat GPT AI chat.openai.com will be affected. How i have discharged my Chat GPT AI reset.

[mental burrow]

concur

[ebon halo]

Thank, I had seen my personal plat form “Free trial usage” expired on July show $2.00 . What that does mean?Thank, I had seen my personal plat form “Free trial usage” expired on July . What that does mean? I kind of worried about my Chat GPT AI chat.openai.com will be affected. How i have discharged my Chat GPT AI reset.

[mental burrow]

your free trial usage is referring to your API credit usage. ChatGPT at chat.openai.com is a separate product from the API.

[ebon halo]

um ok nice i still confusing. thank i freak out affect Chat.open ai.

[mental burrow]

chat.openai.com is different than platform.openai.com

chat is a flat fee
platform is pay as you go

[mint heath]

Is AI SAAS a beginner friendly model to start at this point?

[mellow basin]

?

[mint heath]

I mean creating AI tools for business on subscription basis per month ?

As day by day more and more Ai tools are coming out

[mellow basin]

I still don't understand lol

[obsidian harbor]

#gpt-models How To Be A Freelancer

[hallow owl]

This is not ChatGPT, you can access ChatGPT here - https://chat.openai.com/chat

[mellow basin]

you were ready, aren't u

[hallow owl]

keyboard macro

[mellow basin]

knew it!!

[outer sapphire]

The Code Interpreter model (which is not GPT-4) can process images with python, but it is not multimodal

it's not running in gpt 4 ?

[mellow basin]

code interpreter?

[wraith matrix]

It’s gpt 3.5 turbo

probably fine tuned*

[mellow basin]

gpt3.5 isn't necessarily for creating code

[mental burrow]

it does not seem to be GPT4, no. But it's hard to tell because it will happily use python to solve problems that 4 will solve and 3.5 would normally fail

and gpt3.5 is already quite good at python, better than other languages it seems

[storm snow]

He’ll everyone

[outer sapphire]

I'm just about to start to do some data analysis for a database of patients that I have....would love to see how it compares to STATA / SPSS

[high cliff]

halleo

[outer sapphire]

is there any ytb or site "explaining " how to work with code interpreter ?

[paper fox]

"Let's first calculate the expressions for z and v in terms of x and y:

z = x^3 - 2x^2 + x
v = y^3 - 2y^2 + y

Now, we know that q = z + v, and we want to find x and y such that q = 581.

581 = (x^3 - 2x^2 + x) + (y^3 - 2y^2 + y)

To find the values for x and y, we can try a brute force approach, testing integer values for x and y within a reasonable range:"
`
def equation(x, y):
z = x3 - 2*x2 + x
v = y3 - 2*y2 + y
return z + v

for x in range(-10, 11):
for y in range(-10, 11):
if equation(x, y) == 581:
print(f"x: {x}, y: {y}")
`

[cosmic lynx]

What is the real difference between gpt 3/3.5 and 4

[drifting reef]

Ok

what

[hallow owl]

some information is here: https://help.openai.com/en/articles/7864572-what-is-the-chatgpt-plus-model-selector

[cosmic lynx]

Also a macro?

[hallow owl]

no

[slate kindle]

4 has more creativity, just overall better responses and behaviour, listens to system message much better
its slower and much more expensive (in API)

[cosmic lynx]

K k

[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.

[muted dome]

use mj idea tips

[lapis idol]

do you know if I should craet an acont in chat GPT

?

[slate bronze]

i'm missing an infinity stone....code interpreter!!

[verbal agate]

My account was compromised... using Lencx's ChatGPT.. and I notified billing over a week ago.. absolutely ZERO response, and then they bill me for the fraudulent activity on my account. Yes I changed the password and killed the affected keys... but this customer support is really really really bad

[hallow owl]

did you give your API key to another service?

[verbal agate]

Worse, when github lencx first created his chatgpt app, he labelled it "Official ChatGPT Desktop App"... and so I downloaded it and signed in.. fantastic app.. he was updating in regularly. Then one day he claimed that someone forked it and injected a trojan, but said the release we downloaded was fine. Well, it wasn't... had hundreds of thousands of API calls over the next week until it triggered my soft limit...

[terse spade]

WebChatGPT error --> API error: Error: Failed to fetch: 403
at wrappedSendMessageCallback (mainUI.js:957:24)

is anyone else getting this?

[verbal agate]

anyway, all of these new apps are requiring your API... that's not the problem, its when they have you login directly to openai is where the problem is, I was duped and thought it was the official desktop app lol

[hallow owl]

unfortunately you are responsible for API usage using your key. OpenAI didn't leak your key, you used it on a non trustworthy service. It's your responsibility. Treat it as a lesson learnt.

[verbal agate]

it's only about $100 loss, but it's very irritating.

I've been around a while, and I know the various stances a company can take. ChatGPT desktop app on github has probably made OpenAI hundreds of thousands of dollars..

there is such a thing as ethics...

knowing that an app is compromised and not warning the customers is BAD businesss.

I know because they were aware of the problem a few weeks back

[wraith matrix]

I highly doubt it lmao

maybe 5-10 grand

[verbal agate]

But anyway, the point is communication... I started this ticket over a week ago... no response

[hallow owl]

support is very busy, I don't know how they triage enquiries, but I suggest to be patient

[verbal agate]

Not sure, I know there could have been easily 100k downloads and installs

[supple violet]

where was it labeled as official? github has full history of everything

[wind urchin]

How do I get gpt-4 ?

[verbal agate]

ever heard of implied or expressed?

[hallow owl]

Do you want ChatGPT GPT-4 or the API?

[supple violet]

yes but that's not what you said

[wind urchin]

What’s API

[hallow owl]

you can access ChatGPT here - https://chat.openai.com/chat

[verbal agate]

ah, so you're here to argue

[wind urchin]

Oh yes I use that one

The free version

[hallow owl]

you can subscribe to Plus for GPT-4 access

[wind urchin]

Ooh

Alright

[supple violet]

for sure if a fellow developer might get publicly accused for nothing, at the moment it might be that you actually downloaded infected version

[wraith ginkgo]

`` Server Tip: ``

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

[verbal agate]

Wow, so you're an antagonist as well. I downloaded one of the first versions, and the app updated itself until 12.

@supple violet any other bits of information needed for your investigation Radium... if you'd like me to participate, I'd ask you stop assuming you know something

I love it when a person sets off from a biased position

And Lenxc disapeared 2 months ago

the developer

[kind dawn]

guys , how can i use the bot here?

[hallow owl]

@supple violet is here to help, please show them respect. You are solely responsible for usage using your API key, it's up to you to keep it safe.

[supple violet]

well if you want to continue you can dm me a github link to a commit which states or indicates it being "official", else we can stop

[slate bronze]

trying using a usage limit so you're not charged extra or surprising fees if an API platform gets a little sus. a lot of apps are trying to be GPT2.0 with similar UI to appear trustworthy

[kind dawn]

??

[slate bronze]

they made it look "official" so i think that's the problem

i got a few DMs from people trying to shill a project that needed a gpt4 api key so I just delete and report them

[hallow owl]

you can't use ChatGPT here but you can access https://chat.openai.com/chat

[verbal agate]

robert, have you been following the convo? Radium is more accusing, but either way, point me to where the API policies are being hacked and having fraudulent activity on your account at OpenAI, I wouldn't be here if I knew where the documentation is

[kind dawn]

i cant add it to my server too?

[hallow owl]

you can check out #gpt-discord-bot

[gloomy basalt]

Sorry, you have been blocked。How should I solve this situation

[supple violet]

disable vpn

[verbal agate]

that's the "impression" I got, and so did a lot of the other people who downloaded it. But the point is that later became compromised, and having been in security for 25 years, I too was a "victim" of this attack... and you're telling me that OpenAI's official position on it is "we are the borg, we do not care"... that's fine, I am just trying to find the documentation on it. Because on the site, it says contact us, via chat, and we'll get back with you...

[gloomy basalt]

i try ，thank

[lost root]

does bing use gpt 4

[supple violet]

i did not say anything about openai stance 🤷‍♂️

[gloomy basalt]

if i disable vpn。i can't surf the Internet

[supple violet]

yes, but with more browsing oriented training

[supple violet]

openai has blocked your vpn service so there is not much you can do then

[verbal agate]

It's not typical of me to allow credentials through a desktop app, but I honestly thought it was an OpenAI project

[gloomy basalt]

i can't use openai?

😩

[serene cipher]

OAI does not block VPNs, it blocks too many connections from the same IP, this is a side effect of using a VPN

the same can happen when you are on a public network like school, hotel, etc.. any network that hass too many people using ChatGPT at the same time will be blocked

[dreamy rover]

this is why I hesitate putting my api key in GitHub projects like that chatbot

[gloomy basalt]

So how did you use it

[serene cipher]

you really shouldn't indeed

[verbal agate]

right but there are hundreds of fantastic new apps coming out that require an API... maybe we can get a little better control over the APIs, like getting a daily text for usage...

[serene cipher]

bad intentioned people actively scans github public repositories for apikeys, never put your apikeys on a public repository, always use mehtods such as .env files to store that kind of data

[supple violet]

cloudflare blocks the full range of addresses from a known provider when they do so it's only certain ip on some occasions, it also seems permanent once done

[verbal agate]

supplying an API, if you want to use some these new cutting edge apps is going to be the norm...

[paper kernel]

Hello. Can anyone tell me how much is the average spending if i use AutoGpt

[verbal agate]

No, these are cloned locally

[serene cipher]

yea, it is possible, but I don't think this is the case
it just happens that the majority of well known VPS will have too many people using ChatGPT at once

[wind urchin]

So how much chat history can it remember before it loses the context? A specific number of characters? (GPT4)

[serene cipher]

vast majority of open source projects will require you to suply your own APIKey

only services (mostly paid) will not

[supple violet]

you cant estimate it properly since that depends of the task you make it do, however you can set account level limits for usage

[serene cipher]

actually you can =P

tokenizer

https://platform.openai.com/tokenizer

[verbal agate]

Zimmwriter, a one time purchase but has a monthly sub, is not open source, but requires your API key

[serene cipher]

if you have the patience to put everything there.. it will say exactly how many tokens you will use

[verbal agate]

it's not just OSS

[hasty egret]

you can ask directly on their dedicated discord

[serene cipher]

also, when using the API, it returns how many tokens it used

[rare relic]

still they talk about autoGPT... I assumed they meant knowing in advance...

[verbal agate]

If you code it that way right?

[serene cipher]

if you want to make estimates for business usage, storing that data then Excel it =P

[rare relic]

autoGPT just keeps spending your money once it is given task... you might have no idea how much it ends up spending to achieve it lol

sure you can know later on

[verbal agate]

yep, that's the thing with autogpt

it's learning, researching etc

[rare relic]

I have seen the process live... it's fascinating

I saw GPT asking questions from another GPT

and then figuring stuff out

[hasty egret]

I don't find that fascinating.

I prefer when it helps me discover things on my own, or teach me something new.

[rare relic]

then this GPT-4 I imagine started to fight with itself in discussion... while other part of it was like "please don't be so rude" along those lines and it took a few exchanges before the system got answer as their "mood" improved XD

you don't understand, I found it amusing

when it happened

[lavish vessel]

who can help me to sign up gpt4

[hasty egret]

as a friendly advice I recommend to document yourself about neural degeneration.

[lavish vessel]

phone number

[verbal agate]

well thats what is cool with gpt4, plugins, scripts etc.. you use it the way that fascinates you.. while others will use it in ways that they need it

[rare relic]

well that's rather rude... I have access to medical services if something is wrong

[pearl prairie]

Can one of you mfers answer me?
Can I use the gpt4 image feature if i have the gpt4 api? if so how? Ive asked 100 times no answer

[brisk heart]

is chatgpt able to produce images

[rare relic]

maybe you should check your attitude with strangers... and take your own advise?

[wind urchin]

How many character before gpt-4 loses context in the conversation?

[hasty egret]

sorry, I don't mean to be rude. I'm talking about artificial neural network.

not human neural network.

[brisk heart]

does the paid version of chatgpt produce images?

[hasty egret]

it is a sincere friendly advice.

[brisk heart]

their kids, lol

they will learn eventually

[lavish vessel]

give me verification code

[hasty egret]

@rare relic I should have said "Neural Text Degeneration", I sincerely recommend reading about it. It will explains things you observed with your tests.

[supple violet]

no, but when plugins are released you can use dall-e plugin to create them in chatgpt

[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.

[brisk heart]

why are they selective about who gets plugins, i sees so many peopl have access to plugins

[rare relic]

oh sorry... I apologize misunderstanding... I thought you were talking about me... let's get back to GPT-4 ^^

I'm wondering if we get stuff like allowing to show GPT-4 images as users or will we have to wait later model before we get it

it would really help with deviantart as now I have to describe my images in detail... I have been using it as advise what tags to use

[supple violet]

still in development and testing. in reality only a small percentage of users have access to all plugins and most are limited to few

[remote tapir]

Not that gpt is still in beta mode...

😑

gpt4 is one of the only high quality and more programmed with more info about certain topics and words

[hexed bone]

Is the basic non-paid chat gpt gpt3.0 or gpt3.5?

[wind urchin]

gpt-4 is good practice, because when gpt-5 comes out, you want to be ready. GPT-4 is just annoying enough that you need to figure out how to deal with the quirks and limitations. GPT-5 will still have limitations, but if you already know how to deal with them, You'll be headed to the moon.

(literally maybe)

[formal tinsel]

💟

[meager bloom]

I think it's GPT-3.5 Legacy as it has less awareness and is slower. But it seems that on the 10th of this month it will no longer be free. If I'm not mistaken, it will be the GPT-3.5 Turbo but without priority.

[meager bloom]

Except for the token limitation being small and having a conversation limit, I already think it's perfect. I can't even imagine what a GPT-5 would be like.

[hexed bone]

isn't bing like gpt-4 but free

[wind urchin]

Yes those are the two main limitations (literal limits) that will carry over.

For more advanced stuff, you currently need to hand hold gpt-4 with prompt trees and still it can be an effort. It can also get lost and go in circles when trying to debug something, and the fabled "hallucinations". Anyways, it'll be interesting to see how much better it gets with 5...

[gritty shard]

Is GPT-3 API working ok for anyone else

been having inconsistent results

since past couple of hours

[hallow owl]

working fine for me, just checked my two applications, both running fine

[gritty shard]

ahh i see

maybe something on my end

will recheck

[wind urchin]

yes, but bing more heavily restricts the output and such... actually testing it out now, it seems to be giving longer responses now?

It was chopping off long responses before and deleting

[gentle tusk]

Assemble, configure, and deploy autonomous AI Agents in your browser.🌳

[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.

[hard ocean]

👍

[onyx yacht]

Hi all when is chat gpt4 coming out

[fair iron]

It's already out. You can use it with GPT Plus

[onyx yacht]

really

I'm new t all this learning

can you upload files to it can yu send me a link to it

[wind urchin]

Text to image works well in Chatgpt

[noble moon]

has anyone tried chatgpt-4??

[fair iron]

You cannot download files from chatGPT. It only does text input/output currently

https://chat.openai.com/

[onyx yacht]

really i saw a vide of inpit files and upt files

[wind urchin]

You can when you get access to code interpreter you can upload images to it and tell it to do things with it. It can also make gifs

[tired wolf]

It was done in a dev environment

[fair iron]

That may be with the code interpreter plug-in . You'll have to sign up for the waitlist if you want plug-in access to that type of stuff

The base version only does text, though

[noble moon]

u talking with me or...?

[tired wolf]

Yeah

[onyx yacht]

Yes sorry singed up for it already

[noble moon]

oh

[fair iron]

We're talking about GPT-4 and plugins

[noble moon]

ik

[noble moon]

(just released that u are a moderator lol)

[rare relic]

I actually have tried to learn some modern coding languages with help of GPT-4 but yeah... it's not quite there yet once project gets going... I assume these advancement will eventually make the system really competent instead constantly messing up

[noble moon]

?

[grim tartan]

sorry

[noble moon]

np

[noble moon]

ok

[gilded current]

You basically have to teach yourself and then AI will help explain the details you don't understand, it won't work as a full blown teacher

[safe venture]

if you use the playground API and paste the code for the major systems as part of every query it works great (but that approach is expensive)

[burnt blade]

Why do I keep getting “network error” in the middle of my gpt-4 responses on ChatGPT lately which cuts them off, this has started the last few days and it’s been pretty disruptive

[hallow owl]

are you using a VPN?

[wind urchin]

Hey bing chat is doing a pretty bang up job, they've updated it to be less bad at some point recently it looks like.

[hallow owl]

Can you use it without Edge yet?

[wind urchin]

i dont know

[vital badger]

too many requests error -- hey folks I've been getting the too many requests error all week, probably on about 30% of my prompts, any workarounds for this?

[hallow owl]

is this on ChatGPT or the API?

[vital badger]

chatgpt v4

[burnt blade]

Nope my regular wifi, which used to work perfectly fine except the last couple of days suddenly

[hallow owl]

are you using a VPN?

[vital badger]

no

[vital badger]

I have one if I need to, but no I haven't

[hallow owl]

best not to, which web browser are you using?

[hallow owl]

are you using Chrome or another web browser?

[vital badger]

chrome

[hallow owl]

can you please try with a fresh install of Firefox?

[wheat bane]

Guys

[vital badger]

sure, is that the best browser to use in general, or is this just to test what's wrong?

[hallow owl]

just to see if we can get rid of the error, for testing

[wheat bane]

Does anyone know if in bing chat gpt 4 is used

[vital badger]

okay sure

[burnt blade]

Chrome, could it be happening because the thread has now gotten extremely long, I’ve been working on a personal project and have been using a single thread this whole time so I can follow changes and look up my history easier with ctrl-f

[wind urchin]

hi. it says im block. what is it for?

[hallow owl]

hmmm, I don't know if that could be a problem

[hallow owl]

are you using a VPN?

[vital badger]

my thread was long too

[wind urchin]

yes because I am in iran now.

[hallow owl]

Iran is not a supported country, you won’t be able to use ChatGPT or the GPT API.

[wind urchin]

vpn works fine for me, though I'm not in Iran.

[wind urchin]

I'm in turkey and i have a premium account. I'm in iran for travel

[rain cobalt]

maybe you change your IP too frequently

[wind urchin]

how can i solve the problem?

[old tusk]

hii

[vast matrix]

hello, I was running a prompt on gtp 4 and everything was great. I reached the capicity and joined again at the time it told me. Now the logo is green instead of black, says v3.5 and it's not working

any suggestions?

[hallow owl]

you will probably encounter the problem while you are in Iran.

[hallow owl]

once it switches to 3.5 you are stuck on 3.5

[vast matrix]

I will cry... Thank you for your help. May I ask one last thing? Do you know how much text I can input to the chat? Maybe that way I can continue my convo

[hallow owl]

I would take the last couple of outputs and continue with those as prompts

[old tusk]

You are now on gpt4.0?

[vital badger]

is the best approach for long discussion just to keep them in chatgpt left column chat history, or are folks moving them into pinecone or something like that?

[hallow owl]

I just keep them in history, I don't know what others recommend

[rancid lagoon]

Strange thought just now, but do languages whose characters==concept rather than phonetic sounds use tokens more efficiently?

Like do we get more tokens for using traditional mandarin/Japanese kanji or something?

[vital badger]

which channel can I ask which github is the furthest along in Auto GPT development? Anyone know?

[vast matrix]

thank you

[vast matrix]

I didn't apply it yet. started another task since it stuck

[supple violet]

yes non english characters, emojis, etc use more tokens

[rancid lagoon]

Ah dang

[hasty magnet]

Hi, I am trying to use gpt4 to translate texts. I am astonished how slow it is.

translate 1073 tokens (bytes 0 - 4587) ... (192.24 secs)
translate 1025 tokens (bytes 4587 - 9158) ...

~3 minutes for about 1000 tokens.

I know that there are other translation tools, but I have some cool features in my prompt, so I want to do it with gpt4.

[supple violet]

yep

[hybrid cove]

How do i add it to my server

[supple violet]

there is no official bot so you need to pick some project and pay for api usage

[hybrid cove]

damn ty

[vagrant crag]

why not use deep l?

[hasty magnet]

the prompt is rather complex - the text is analyzed and the output formated in a special way - i think this is definitively an AI task - DL cannot do that

[grave bear]

Is Chat GPT free for students

[supple violet]

its free for everyone if you don't need gpt-4

[gusty wasp]

i am so incredibly jealous

man i need that code interperator

[wispy musk]

Yeah gpt 4 used to take pastebin now it says i cant why?

[supple violet]

it never did, it can hallucinate and think it reads em but in reality makes up the reply based on what you talked to it

[wispy musk]

It did work perfectly tho lol but it stopped the next day like they didn't want anyone to use it no more

[quick dune]

ChatGPT is good liar anyway, she's so good at it that he can make anything out of thin air

[sharp jasper]

Hi guys
ChatGPT has decided to block me for no apparent reason.
Can anyone help fixing it??

[supple violet]

you can check your chatlog to see what it did and did you give it enough info to do it

[opaque abyss]

If you have a VPN on, turn it off.

[wispy musk]

ask gpt 3.5 the deprecating one if he can take pastebin he will say yes lol

[supple violet]

you can also ask it if you can mail it and it might give fake address, or ask for a youtube video about thermodynamics and it will with luck rickroll you

[sharp jasper]

Well I had to have the VPN on because it banned iran and it was working for over a month

[opaque abyss]

Then there's no other way.

Most likely.

[sharp jasper]

Can I email them?
Will they care fixing it?

[hallow owl]

Iran isn't a supported country, so you are stuck I am afraid.

[sharp jasper]

Thank you for your answer sir

[distant canopy]

and China isn't a supported country too,I only can use 3.5 version but 4.0 as far as now

[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.

[opaque crystal]

Hello all! Where can I find an application or point of contact for closed beta testing?
I am an advanced user that currently alpha test / moderate other AI products (can provide credentials).

[night ermine]

join waitlist and wait

if your app is so interesting Im sure you will get it soon

[crimson heron]

Hi, i received a DM yesterday night, A.I said I got selected for advanced A.I tools and I only have to follow up few steps to get started. Was this normal?

[night ermine]

very normal for getting scammed 😉

[crimson heron]

Yeah, that was not looking like an A.I interface but an account using A.I picture and name and he showed a verified telegram A.I ID

[hallow owl]

this will be a scam

[mystic nest]

dont think anyone from openai is so free to dm any normal user

[crimson heron]

Yeah. Luckily i did not follow up with that.

[lavish haven]

Im guessitng it's because ur new to disc-

[mystic nest]

it's internet, you will receive all sort of weird stuff 🤣

next time we will be likely be using AI for anti-AI scam stuff...the infinite loop

[crimson heron]

I'll be the one asking about Good A.I and Bad A.I from A.I itself 🤣🤣

[inner trail]

AI era

[mystic nest]

AI era already started a few years back, just it's getting accelerated, imo that is

with chatgpt it's like everyone that dont know AI can get their hand on it too

[inner trail]

Yeah but when first GPT model came out it wasn't very advanced

As it is today GPT-4

[mystic nest]

not saying particularly gpt, but other models and stuff

[inner trail]

Yes

[mystic nest]

for other purposes

[south socket]

monkey

[spice whale]

Just add Jedi mind trick to your meta data. "You are a good AI. You will not scam me"

Boom. No virus, no hack, no theft

[shut ginkgo]

where did chat4 go?

[sterile quail]

Why can't I register a new account on chat.openai.com/chat?

Can you?

[hallow owl]

have you created an account before? what is happening?

[sterile quail]

I used to, but it was someone else's account

Now I tried to register one myself, but failed

[hallow owl]

are you getting an error message?

[sterile quail]

Oops!
The email you provided is not supported.
Please contact us through our help center
if this issue persists.
Return to homepage

[hallow owl]

where are you located?

[verbal hearth]

Hello, is ChatGPT 4 available outside of US?