#chatgpt-discussions
1 messages · Page 3 of 1
From my experience it's less about how many words and more about how many characters, I regularly have it do 2,000 words at a time and I've had it say it can't every so often but it's rare.
I use Custom Instructions to tell ChatGPT about my aesthetic preferences for characters in images. Like my favorite colors, etc. But there’s a problem. No matter what I say, it seems to only ever use those colors I specify. I do want it to use those colors more frequently than it otherwise would, but I want it to sometimes use other colors as well. How should I express this without using up too much space?
But I don't think it's a set amount either way. I'm sure there is some theoretical upper limit, but I feel like it's only real limitation on generation is how many users it has at one time
I am guessing the dall-e in chatGPT is separate from image upload (reading images) ?
seems I got the dall-e access now (EU)
but not sure about uploading images for chatGPT
bro i had access for like 10 minutes and lost it later when i refresheed
what the frigg
it's still rolling out, there may have been a glitch, be patient and it will come
im taking this extremely personally. gonna spend some time verbally berating chatgpt
be nice to the AI, it may help us when we really need it 🙂
Can anyone share - what’s the first sign that I will have ChatGPT Dall-E 3 access? Plus icon in the GPT-4 Default Chat or other type of Chat? Mobile or firstly web? Logoff-logon necessary?
orange send icon instead of purple for gpt4
its a new option under gpt-4 menu and the orange was alpha color, its now the purple gpt-4 logo
oh yea what he said
Thank you!
wait so did i have access to the alpha by accident for a bit? cuz i had an orange send button
refresh and it went away though
Maybe, or some other momentary glitch. More likely a glitch, alpha's a separate tab like the 3.5 or 4 choices. If you didn't see and click alpha, the orange is likely just a loading in issue.
Is there anyone real who got access to GPT4-V here? 😂
I keep seeing people sharing screenshots of GPT4-V on Twitter
Many have it, many don't. Still being rolled out!
I have Voice and DALL-E 3 but not Vision
I do
Omg can you try things out for me please?
I have vision it's impassive! But don't have voice an Dalle 3
I hope I get it next week
I’ll try one thing for you if you’d like
@jovial flint Yeah, let me send it in private messages
Sent you a dm
Hey everyone, I need help with something. When Chat history is turned off, ChatGPT+ is automatically clearing a conversation while I am still typing/conversing. For example: I send a message to GPT-4 asking to write a blog article. It writes a blog article. Now I am writting another prompt to refine the article it wrote. While I am still writing my prompt, the whole screen clears up and starts a new conversation! This doesn't exactly happen right after sending the first message, sometime happens after 3rd or fourth message. This issue started after the most recent update. I tried clearing all cookies, session/local storage of my browser, issue persists. This is super annoying when I am deep into a conversation and their is no way to recover it since I turned off chat history. Is there any way to fix it?
Hmmm does this also happen on the mobile app? I have not heard of or seen this behavior myself.
I would reccomend reaching out to support@openai.com with your bug report if it continues
Don't use mobile app. However on mobile safari I usually end my conversation after 2/3 messages. I only keep a conversation alive on the PC since I need help with my daily tasks
Sent a message through their bug reporting option in help centre. Not sure if they will notice and fix
Does the safari version of chatGPT have the same behavior of clearing your messages in longer conversations?
Does anyone have any tips when working with large amounts of data in GPT 4? As a proof of concept, I'm trying to use GPT 4 to create an algorithm in python that generates line segments in the shape of a tree. After about 200 lines of code it seems to start rapidly losing large amounts of context. I've tried forcing it to split each task up into units and then working from start to finish on each task but that so far seems to actually make things worse.
I usually close the safari browser after finishing the conversation and later I don't find the conversation. Just checked the bug-report channel. Looks like a lot of people are having the same issue
If you have the chat history off, it will not preserve your chat messages between sessions/reloads
It doesn't matter. They're just payment processors.
Yes I understand that so not having any issue about safari. The issue is on Chrome Desktop. I keep the tab open and while typing, the chat refreshes.
sorry for deleting the message so i was asking what to choose between stripe and google and it was just a mistake it's the same price google is cheaper because of the change
and so if i do the subscribe on android does it work on pc ?
See if it does the same thing in incognito mode
Or attempt it on firefox or edge
It should work everywhere. As far as I know, OpenAI doesn't charge per platform.
Let me check on incognito. I will let you know. Using firefox won't really help me since all my synced data is in chrome.
thanks
I've in trouble And I cannot renew my chagpt4 , anyone can help me ?
what is happening when you try?
man they really have to do something with the usage cap with GPT4
why, what's wrong with it?
Please dont tell me its down now
Me when chatgpt is down for 0.0000024 seconds
Yup, same thing happens in Incognito too.
I'd say it takes about 10 minutes to refresh
Hey my chat gpt is not working what to do
Reload
Both thw accounts not working free and paid one all of them are down what to do tried
Reload religion nothing working
Even right now?
I think in india they are slowing down
They just released image feature so maybe everyone is using it
I found the problem: if you're using own DNS (NextDNS on my side) then make sure that you have enabled access to cdn.oaistatic.com otherwise ChatGPT could not load any resources.
is the ChatGPT vision rolling out on time? I saw it was gonna be 2 weeks, that should be next tuesday. If so, am I just unlucky or is it gonna be extended or something like that?
it's still rolling out, be patient it will come
us EU folks have it harder, except for those where its banned.
we have to wait overtime.
Hi everyone,
I'm interested in openai since 2 years now and using gpt since it's available for everyone.
Today, I watched a video of a man have dall-e 3 in it's gpt interface!
Do u know how it's possible?
It's rolling out to everyone with ChatGPT+ accounts. Should have it by end of October
If you're a GPT+ member
plus minus a month or two because we're EU, or UK.
Mmm, they didn't say anything about Dall-E-3 and EU/UK. It's Vision that they said delay on for there.
can't wait 😋
try it on bing
i thought we talked about Vision.

The images on Dall-E 3 inside of Chat-GPT seem to be producing tiny but noticeable dots, as opposed to the integration that exists inside bing's image create. Any ideal why or anyone notice?
do thumbs down then apply feedback
is the text to speech available now?
yes, on gpt4
You might be able to specifically ask for smoothing, blending, and more natural, less dotty, styles. But you could have to specifically request that.
I'm sorry, but I've only experienced Dall-E 3 within Bing, so I haven't had the opportunity to compare it.
how does it work? i cant see it
ooh nvm it's tripping
"I apologize for any confusion. While I can generate audio in a technical sense by creating an audio waveform representation, I can't generate playable audio directly within this environment. However, I can assist you with code that you can run in your local Python environment to generate a text-to-speech audio file."
Hi
Voice is on the OpenAI ChatGPT app only, not the web interface.
From your iphone or android device, you'd open chatGPT app, then go into beta settings and turn on voice. From there, the icon in the chat to turn it to voice looks like headphones
it's more like "speech2text", and it's notoriously bad (at least atm)
it is still rolling out! soon for you, I hope!
I’m a plus user for 2023, iOS and still don’t have Voice. Many of us still don’t. We should all get it by Monday at the latest. 
im using speechify for the last 6 months to listen to the answers still works fine
just one extra step after every answer, would be great to not need it soon
Race ya to be the last to get it!
🤣I’m thinking every time I sign out, it sends me to the end of the queue! I need to play hard to get. Good luck
Thanks! Good luck to you too. I bet none of us are being sent to the end of the queue, and I doubt logging in or out matters. But our turns will come!
I tried that already. Tested with the same prompts as well, they are still generated only on the chat-gpt side. I'm not sure what the rules are on posting pictures here for examples, but I just posted a reddit thread with a visual comparison on the sub reddit. Thanks for the advice though.
Some threads don't allow any images, others are ideal to discuss how prompts matter or the output you get from dall-E.
you guys all have the speech tool already?
And those do accept images
how did it work? did you have to update the app? or was it just there all of the sudden?
it just appears, you don't need to update the app
ok i see thanks
Can chatgbt 3.5 do image to text?
it seems random
no
yea i was thinking that too. they got some dicerolling supercomputer or something
got images 2 days ago but nothing else
i got a friend who got it this morning
I got Voice several days ago
wow ok then it seems pretty random
where are you guys from?
maybe its based on country
im from germany
Australia
I took GPT Plus in France but I'm in Canada at the moment
my friend is in the exact same situation and got it but not me
Still waiting to gain access to Images. Curious to know if there is a system in place for rolling out new features? Is it based on level of activity, seniority, or is it just random?
it seems random
I figured it rolls out based on the same conditions as GPT-4 api. Use case, chat history & tos violations, Etc
But that's just speculative
dam ive had a lot of violations
Does anyone know the Dalle3 limit
I got voice but not vision/dalle
Voice usually comes first, you should see Vision very soon
i got vision before voice
I've got voice, I've got Dalle3, but vision was what I wanted the most 
I got vision on my second account but not my main
if you have any specifics you'd like to try while you wait you can ask in something like #ai-discussions (since they allow images)
Do we know what kind of limits there are for Dalle 3? Or is it just the same as our gpt4 limits?
I don’t think DALLE3 has limits. I’ve generated over 200 images within my 50 GPT-4 cap
damn bro save some for the rest of us
Some of us are really good at spacing!
Others do get failed images, and recently, told of a timer.
Great work at delaying yourself between gens, so you never have to wait!
I can’t sub to gpt-4? It’s grayed it out
well it is a limitation, but not everyone will feel that hard
Anyone else got this problem?
i ate the button sorry
hey how many people have voice in the app because i dont have voice but i have gpt4v
i have voice
i wish i had vision 😭
i got vision but no voice
I've been having issues too; something is definitely very wrong. It's almost like GPT-4 is actually using GPT-3.5 or something. The more i try to guide it, the worse it gets. It's like the conversation history context length is like 2000 tokens again.
openai loves to not say ANYTHING about what theyre doing and its extremely frustrating
why are they rolling out vision and voice separately so some get both while others get one or the other. makes no sense at all.
Did y’all get browsing on phone? Or anyONE? On IOS
yes
When will the photo features come to chatgpt?
research and experimentation? 😄 idk
How do I get Bing chat gpt to roast my selfie?
Dang it it blurs out faces?
I want chat gpt 4 to roast me
im not saying they shouldnt roll it out i just think rolling out everything separately is odd.
What are you talking about guys ?
is Chatgpt only showing this for me ? "couldnt login with the plugin"
where could we learn more about how to use Dall E?
is there a specific thing you're struggling with?
me ?
your issue seems to be related to the third-party plugin. I don't think I can help much with that, unfortunately. I was wondering if V needed help with Dall-E. I might be more useful with that 😅
I want to know more about how to use Dall E and its abilities
like, literally
Dall-E 3 is being rolled out to ChatGPT right now. If you're looking for help with prompts, you can browse #1021130377026351105 and/or #❌┃prompt-help . Most of the prompt tips are applicable for Dall-E 2, and that can be accessed here: https://labs.openai.com/
all the plugins are showing "couldn't login with the plugin"
._.
hm, now i know why this is happening, it's because of the language thing
Hi
Hi! I'm unable to start new converastions with GPT-4 Data Analysis, it takes a long time to respond to a simple "hello" and then it return the generation error red box, any one else with the same issue?
do you mean Plus members can use Dall E 3 on ChatGPT now?
No, most people can't
Some Plus members have access right now. however, I don't know how many do or how long the entire rollout will take
Ok, just curious. Dall E is terrible anyway.
have you seen Dall-E 3? 
dalle-3 is mindblowing
you didnt try dalle 3 to said that
is it better than Realistic Vision of Stable Diffusion and Midjourney?
i cancel my MJ plan cause yes
How would a guy access said DE3, given I have permission for it?
it would be a dropdown under GPT-4 on ChatGPT's website
Ohh it's integrated right into ChatGPT, neato
yup! ChatGPT will expand and rewrite prompts. It's pretty mind-blowing
$20/mo is beginning to be a steal for this platform frankly!
Though, Dall-E-3 will usually generate an image with the text 'Dall-E-3 is terrible' if asked, so... ehh? The image will often be gorgeous and incredible too.
Ive been using Vision to save all my Walmart receipts and putting them in a nice table by price
It's super handy
ChatGPT has a strange understanding of how to get exactly what you want too, ask it to adjust something in the image and it changes the prompt so that you actually get it, sometimes it takes a few tries, but I've only run into a couple things it's been completely unable to produce
ChatGPT is very slow right now and that's making me appreciate more the captchas I was complaining about earlier
with the exception of that messsed up really difficult catpcha that should be illegal
dang, just like 6 weeks ago I looked into a way to do this easily. Crazy how quickly GPT is moving, to now be the best solution
Yeah Vision is a beast at text extraction. Even poor handwriting
very excited to hear this 🙏🔥
im stil without chatgpt vision... is it based on region?
Probably, openAI tends to release things to Americans first
Cuz it's an American company iirc
Many american tech companies release to Canada first, because they're most like Americans, but nobody cares about them, really. (JK, I'm canadian). Also in this case Canada has legislation making them not attractive to AI im sure.
Can anyone help me log out the entire device from my chatgpt account?
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.
funny~9days passed,still not receive voice talk and Dalle3
Thousands of new users and new account got those. Betrayed old users?
it is random also #announcements they still rolling out
Me neither
so yes it is random
And I told u
ive been a plus user from day 1, still waiting on the vision feature. just be patient.
There are some e-market on telegram
Selling voice talk accounts,hundreds of accounts
It’s a grey industry chain
look, you need to understand it's business, the goal is for earning profit. my suggestion is you show less inserested in the products, then you would get more features and service.
And still us first plus subs,some of us even not received Bing browsers and images analy
if you dont have bing you need to enable in setting
Not me
It’s not about setting ok
Actually u cannot find “new feature” in setting interface
no way people cant wait a day for the voice 😂
I think the positive logic is first subscribe,first got roll out
doesnt matter imo, everyone gets it within a couple days of each other anyways
At least we can get a waiting list,but not got roll out at the end of the day,I mean 2wks later
you could do something better other than sitting on discord crying about it
Yeah, i bought another accounts
voice def not worth spending extra money for
I just complain and I know what I can do
yeah i used the voice feature once. found it a bit creepy and never used it again lol
im just tryna talk to it on my walks to class
The roll out strategy is immature
and youre acting very mature philip.
Cause the server’s burden too heavy
You paid for this,you’ve right to speak out your frustration
You know what,if u know every new feature you’ll be the last one push,let me guess your reaction
How does the "rolling out" work? Is it random? Region based?
i believe the rolling out is random
random
“Let it be,I’m okay .idont care about new features baby”
i dont have the vision feature, and although i would really love to have it right now, im waiting patiently
We’re discussing another thing,as a loyal subs,if you been told you always got roll out on the bottom of the lists,the situation is different
ive been subscribed to chat gpt 4 from the day it came out. youre not the only loyal sub here bud
The roll out principle is awful,I mean it
but cry me a river
bro 😂 nobody cares everyone is waiting. your chances of getting it randomly were just as high as someone elses and maybe that person would be complaining too instead
You just told everyone”patient,not cry it”
But Russian got money from selling voice accounts,funny
so buy them. u expect openai to know all their customers life stories?
lets stay respectful please
someone get my man philip lee the new features!!
bro why u getting philosophical cuz u getting a feature a day later like the rest of us. random sounds more equal than rolling out to those who signed up first. no reason to prioritize one subscriber over another
u think this is some fascist government or something
if you dont like how it is then maybe stop paying for the subscripton 🤷♀️
You create thousands new accounts,and you’ll know it’s a real random or fake one.see what Russians do
Yeah that’s my strategy
doubt its that easy to tell. what about the real people who would get mixed up in that? plus each bot is paying their fair share regardless in exchange for the service
From now on,I’ll abandon my old account and buy a new account which got access to every new feature
ok bro waste ur money
can anyone help me with a bug, basically cant buy Plus, cause it says its my current plan even though i never bought it, my card didnt get charged anything, i cant use GPT 4. Last time i subscribed was in july
good on you phillip, then you can stop having a sook about it
From which platform?
browser
Try AppStore
okay
is the image function available to all gpt4 users or no?
not to everyone, its still rolling out
alright, thanks, I was gonna buy a month to try ^^
nice
Day 8- I login everyday to see if I got image upload but no 😔
🙏
Anyone else getting a lot of errors recently?
my brother in ai, we wait together
its like christmas morning everytime i get a new gpt featurue lol
Im refreshing and going back to back bed again and again like a mad lad
when no vision 
Why didn't they release the form for vision and voice too
yep, i had the exact same problem, had to create a new account.
you can subscribe on AppStore but for me it was a bit more expensive than on the web(prob depends on your country)
is it legal for dalle to output copyrighted logos?
whats legal is decided in the courts
i had an output contain the windows logo and i didn’t even ask for it which is interesting
I'd like to renew my sub, but the option to do so is greyed out. It doesn't let me do anything. Has anyone else faced the same issue? I opened a ticket but... one whole week for this seems excessive .
read a bit above, experienced it and other users too ^^
thought i was the only one encountering the issue so created a new acc to subscribe
wdym a month or two 
I wonder how many emails written by chatgpt have been sent to openai 🤔
Thanks, just posted in there.
Is there a way to give myself the 3 month free trial for premium?
there is not a way to get a free trial to ChatGPT Plus to my knowledge
There is, I can show you a pic
somewhere
@slate berry check dms
@warm token Situation fixed, had to upgrade through android
@hard dirge
@candid pecan When you upgrade from android, the situation gets instantly fixed. ♥
niceuu, i had the option to do it via appstore too, but it was like 5$ more than on web due to my country, preferred to just create new acc
hmmmm, I'm not sure how that feature works
It's been 8 hours since I've slept overnight and yet I see 💀
I apologize for the inconvenience, but DALL·E is still experiencing high demand, and I'm unable to generate images at the moment. If there's anything else I can assist you with, please feel free to let me know, and I'll be happy to help.
Everyone seems to love Dall-E! They work to scale systems.
That message isn't you using it a lot - it's about everyone right then!
So it's just timezone I have to wait for USA to go sleep which is in 2/3hours I guess
😔
To be honest, I've been using it off and on across today. I've gotten 1-4 images almost every try.
I need it to design a logo u know
"A one-line logo set against a pristine white background, elegantly spelling out the word 'SUTR'. The unique feature of this design is its one-line approach: each letter is meticulously shaped using a single, uninterrupted thread-like line, maintaining continuity throughout. This ultra-thin thread seamlessly connects each letter, creating a sense of unity and elegance. The entire word appears as if it's been gracefully penned in one harmonious motion
@prime swallow can u maybe help me refine prompt
I didn't know!
But some options:
Have chatGPT help you design it, then take it to Bing image creator or some other option to actually make it as an image.
What do you mean by 'Refine prompt'? Prompt engineering tips?
Any other open source model that does typography?
For most people, Bing's also using Dall-E-3.
Also, you can try again to make your logo in a few minutes.
most of my Dall-E-3/chatGPT image requests got at least 1 image today. Some actually got all 4.
Most got 3.
I see
The only tips I'd suggest:
if that isn't quite what you want, but you're not sure how to reword it, try asking ChatGPT "Riff on this prompt, keeping most details the same, including the word."
'Riff' is a bit vague to the model, but it usually lists these meanings for it:
Music: In music, a riff is a short, repeated musical phrase or motif that forms the basis of a song or a section of a song. Riffs are often used in rock, blues, jazz, and other genres to create memorable and recognizable melodies or instrumental hooks.
Casual Conversation: In informal conversation, "riffing" can refer to a spontaneous and often humorous exchange of ideas or comments between people. It's a form of improvisation where individuals bounce ideas or jokes off each other in a playful manner.
Comedy: In comedy, a riff can be a comedian's improvised or spontaneous commentary or jokes on a particular topic. Comedians often use riffing to engage with their audience and build humor on the fly.
Technology: In the context of technology or coding, "riffing" can refer to a rapid and creative development or brainstorming session where ideas and solutions are generated quickly and informally.
And in use, it makes small, creative changes that stay true to the core concept.
Is there still a message limit on GPT 4?
I don't see the message on the client talking about 50 messages an hour or whatever anymore. Didn't see an announcement either though.
It's there, hover over GPT-4 on a new chat screen and read near the top.
50 messages/3 hours.
Ah ok. I didn't know it was /3 hrs. I've fortunately never hit that.
Yeah! Between the speed it takes to type to the AI, and for the AI to answer, if you don't go faster than about 7 minutes between each of your inputs (and if you don't regen what the AI outputs) you probably won't ever hit the limit 🙂
I definitely don't recommend playing Hangman with GPT-4, hehe!
hmm okay
bing chat also has dalle 3 and it’s free to use
but chatgpt version can refine prompts
It says my custom instructions won't be shared, but if someone chooses to continue the conversation, will the custom instructions still be in effect? Like, could someone in theory extract information from it by asking questions as if they were me?
We had server issues today
A lot
They’ve been mostly resolved
So many server issues……..
hang in there moxi! We (at least I) believe in you 🫶
ofc u gotta believe, it's a worker
No, custom instructions are not shared
Is the small image photo button next to send a message vision?
I dont see anything in beta features
No. And the model may struggle to handle how it was speaking before with what makes sense to say 'now' that the CI are not in play.
You could potentially have someone try to analyze whatever conversations you share, but only what you allow them to see. It's possible someone very skilled in emotional intelligence and psychology, and who also knew the model well, might be able to read your inputs and the AI's outputs and have a pretty good idea what your CI probably says, at least as it affects the showing conversation.
That doesn't mean they know more about you, but they might have clues about what you used that time for your CI.
Yes it is !
Another thing you can do before you give anyone a shared link, is take the link and open it yourself. You'll see it like anyone else would, whomever you give the link to. If you don't like what you see, delete the link, it won't let anyone else open it after deletion.
great 👍 nothing in beta features so I was confused
Congrats
hoping i'm next and not last
Is the dalle-3 thing rolling out at the same time too?
The new GPT geatures roll-out is kinda BS. Some people get it and never use it while others paying for plus get it in the last batch weeks to a month later. Feels like a cop out
Ah, I guess I just assumed it would work differently for me. It's clear that the custom instructions aren't carried over. Thanks!
Everyone who gets it is paying for plus
It’s like really random, I have everything except the Voice feature…
He means despite all plus users are paying for their service, only few selected ones get extra goodies too
He is implying that some people without plus are getting it "while others paying for plus"
I've never heard of instances of free users getting access to features which even some (unlucky) plus users don't have access to
anyone got voice conversation yet?
Nope still waiting.
Anybody else getting errors on all image uploads?
(after hours of it working fine earlier today)
You may have to wait a bit if you reached a limit
yeah. sucks. been waiting a while
Oh, what's the limit?
Still waiting
android or ios
ios
they pay less attention to android 😔
They should roll out by Monday
I expect all user access to them since Sat and Sun are weekend
yeah that would make sense
Today, I still haven't received GPT-4V and voice conversation🙃
I hope I get it by the end of the week
If I don't get it until next Monday, I'll be devastated
It makes me sad to see my friends playing
Yesterday, I finished reading the paper about GPT-4V written by Microsoft, and deeply felt its power
And then I'm even sadder because I haven't got it yet
what is gpt-4v
GPT-4 with Image Recognition
oh. i have played with it
meh its mid tbh but depends on waht you use it for
with selenium automation, who needs API
Can you elaborate?
selenium lets you script website interactions for testing, or whatever you want
Im currently using chatgpt3.5 on automation mode
thats so inefficient lol
generating literally endless content for my website, youtube, blog,
I mean if you are on home network, you dont really care
my god
and it is not like you need to be super low latency and super bandwidth sensitive
but you're waiting for website to load, etc
my home network is unlimited for fixed price
its extra cost in time too
like I said not latency sensitive
i know i just can't let that level of waste go unmentioned lol
it is not like generating in 10 min and 30 min will make a different
a python script that GPT makes for you could do it 10x faster
and when you do content production it is good to have some latency in between
Possible legal ramifications for web scraping over API
greater the supply, the less the value
Many websites prohibit webscraping in their ToS
robots.txt
I predict in the next 5 years, 90% of content will be ai generated
and content will be personalised
That's quite a conservative estimate tbqhwyf
like 1 article written by human, then AI that has learnt from the user preferences will rewrite it before it gets display on your screen
1 story can be written in a billion different way
You can do this with one prompt already on Edge with Bing Chat
the point is you dont even need to prompt in the future
But at this point social media algorithms are still better at predicting what you like
it is just automatically there
Isn't that conducive to an echo chamber?
No
Well, AI could just repeat your views back to you.
That would be an echo chamber
Well, just an echo.
Echo chamber refers to insular social media bubbles where everyone repeats and amplifies their views
If AI rewrites everything to fit one's views as @sick flint proposes wouldn't that be the case?
I thought you meant rewrite it in your content style
spending decision
Not, agreeing with your views
But I don't think many people would want an echo chamber AI.
it is only the matter of the prompt that these company will incorporate in the experience
Add your well-crafted prompts to our #1019652163640762428,
or share your interactions with ChatGPT in #1050184247920562316!
Echo chambers happen unintentionally due to social media algos not because people actively seek them out.
Let's disambiguate here so we understand each other. @sick flint what did you mean by "user preferences" in #chatgpt-discussions message
Most people probably want AI that tells them what's true.
...... . Add a guidance for the user to purchase X from Y at link Z
user preference is what the user watch, like interact with in the past
But that could be so specifically tailored to actually be useful to you.
the context of that user live, work, relationship
This sounds like a recommendation algorithm, not a rewriting AI
In an ideal world, advertising could be so targetted that every ad you see is a product that improves your life.
yeah it is recommendation but on a higher level
Yeah, we already have this with social media to a certain extent.
Precisely
the greatest mind are all in advertising for a reason you know
they are in the business to sell sth
Business majors forgetting their coloring assignment due at 11:59 PM
Greatest minds are probably in tech and academia
you need to sell sth to make money, whether it is good, service or a company itself
Most valuable company, Apple, isn't built on ad revenue model.
they are selling the device, so they need to advertise it in someway.
Their devices practically advertise themselves.
they are doing it in a masterful way that you think it are not
the greatest minds are probably unknown
What makes you say that?
thats the pinnacle of advertisement
The greatest minds are on the OpenAI Discord
lmao
making the user think the product are advertising themselfves
theres nothing in our economy, culture, society, government or anything that ensures the best minds rise to the top. Since most people don't rise to the top, most will be obscure and relatively unknown, probably undiscovered.
Yep, but it doesn't feed into the social media/recommendation algo model that much. They make good devices, with ads on their backs. And blue/green chat bubbles and so on.
😂 indeed
they are advertising on social media
! Help
I start a private/incognito conversation with GPT4. Everything is going well, I leave for 5 minutes and then type a new message in the same conversation-- entire chat goes away -- all of it -- no warnings, no errors shown. I've experienced this multiple times and I've started seeing behavior from 18hours ago. It used to work fine before that.
Is this happening with you guys as well?
god does it? i can't argue with that. literally.
I guess you are never bombarded with ads about iphone like me
If you're a greatest mind of all time in the USA, you will certainly be discovered. Unless you live in some isolated cult or something.
and the people that use those phone, celebrity, KOL, ...
So every great mind is doing great things to be discovered? Even the ones in poverty? nah.
I have seen way more Samsumg/Android/Google ads than Apple ads
magical thinking
if they are great they wont be contented in poverty
if you think someone has a great mind and he is still in poverty, he is not so great
everybody has their own definition of great. its why you should never use the term "great minds" for anything serious.
You just used it though
with my own definition, in response to his use
What's your definition
yeah, Im talking about general sucess, have family, have wealth to support family and help people around you in meaningful ways.
No, @sick flint started referencing business majors as the ones with the greatest minds
really?
ew
yes sir
Honestly though, in this age of available resources, a great mind is anyone with passion and commitment. thats my definition. I know a lot of people with great minds who feel trapped in poverty. Doesn't matter how smart you are, life doesn't care.
I like this update.
Did anyone from EU get vision yet ?
hey there ive bought chatGTP4 monthly today and my cousin got the Dall-E for free as a beta application in it, but i dont? Any reason why i dont can use it for free also?
Don't worry, you'll get it as a subscriber
they are rolling out "randomly". you just have to wait, it will come
It's just that you need to wait as long as I do
I hope you don't fall apart like me waiting for GPT-4V and voice chat
well looking at all the "features" like plugins or data analysis or the webbrowing which work 10% of the time i wont get my hopes up. gpt4 was nice when it first launched now its nerfed af. Look into lama from meta if yall want something of more cabablity. gpt4 is becomming more and more liberal censoring/gatekeeping politics
good for you
Ever try to make coffee in a teapot?
im guessing you use it for Low intellectual quotient tasks so you should be just fine with gpt
there is no quality downgrade in gpt4, but the model now asks for more precise prompting
what ever floats your boat buddy
bro, you're on the openai discord arguing gpt4 is worse than lama, what is the point ?
The objective, you see, is to foster dialogic symbiosis between computational paradigms, particularly when applied to cognitively intensive endeavors. But then again, not all venues are conducive to such elevated exchanges.
Quite a sesquipedalian way of saying "Square peg. Round hole. Why no fit?"
anyway... moseys on out of the argument. Lol
Man, the copyright filter gpt uses for dall-e is way to strong. THe prompts are not even close to what i describe.
For sure, I did ask for some stuff to see how Dall-e would handle niche features, and was legit surprised to get a refusal because of copyright. Makes sense, but oof lol
So far, Bing is alot better than GPT. Hope this will change soon.
ur right, but you cant say that, the gpt fanboys are gonna come for you
He just did say that though 😵
for dallE?
depends on your intent
i dont like theres limited control in gpt
but also ive had some of the most beautiful results ive seen from dalle yet using it
Bing doesnt alter your prompt. Can you ellaborated why my intent can affect this result? As in why GPT can be better than Bing
i mean which you choose to use
depends on what youre trying to achieve
does that make sense?
No. Bing and GPT uses Dall-E 3. They transfer the prompt to Dall-E. GPT alters your prompt before transferring it. Bing does not this. That means Bing can do 100% of what GPT can do, but GPT cant do 100% of what Bing does.
I know
thats what im saying...
but it can help translate your prompt and make it more able to be understood by Dall E in a way that delivers beautiful results
also i have a feeling gpt's dallE is a little better
Yes, but you can use GPT to create a more efficient prompt, and serve that to bing. GPT will remove stuff from your prompt it think breaks its content policy.
i believe you can tell gpt not to alter your prompt
What happens when you try to send a prompt to Bing that breaks Bing's content policy?
Yes, but then it stops processing the image.
Bings content policy is as far as I can see, just restricted to adult stuff.
https://chat.openai.com/share/54f8be8d-d8c0-4893-b6ad-c5b1c017203b
Well this is awkward.
Orbital missile inbound
Hello. I am new to discord in general. And to this server. I hope this is the right place to ask. Is it allwowed to have multiple accounts (not multiple plus subscriptions) just for the sake of kind of having sort of "workspaces" such as that you do not clutter topics with each other that do not attend to each other. Thanks.
it's recommended to have just one account and only 2 accounts can be created with one phone number
Okay thank you very much for your answer robert !
I also just realize I have a typo in my name
I personally have 5 conversations named by their subject matter and any question I have will fit within one of those 5 conversations
No need for different accounts
Those who have ChatGPT Vision in their Plus subscription have already compared the results with Bing's AI Vision? I have it available on Bing, but the results are very poor.
Do you think ChatGPT Plus Vision, when released, will be better? Because if it's the same, it's really bad...
anyone noticed chatgpt being unusually slow today
Here too... very slow
L, guess i make this list by hand :p
Well they have advertised that you "own" and can use all the the generated pictures, which is probably why they need a strong copyright censor, or you get in copyright problems with the pictures when you use them
Inference speed is inversely proportional to server load. More people using it right now than some time before
ah, legal handwaving
Legal handwaving, or respect for creative ownership. Who can say, really 🤷♂️
The world may never know
I believe the censor may be even more problematic in the future sadly :/ I don't know about artists/brands but I don't think they want their art/brand to be together with bad reputation stuff
but on the other side, we can do all this stuff with photoshop too
well yes this is why everyone is making mickey mouse images with it
One thing's for sure: copyright law ain't ready ||(and probably never will be tbh haha)||
yup copyright law is long from perfect lol
That's when you end up with stuff like DropBox closing the account of a big-name tv show producer because their algorithm wasn't built to consider that the account holder might also be the copyright owner. Lmaooo
I think the equivalent would be like if I asked dall-e to generate images of something that I owned the copyright on. It wouldn't really be able to understand that "Nonono, I own this. We're all safe to do some generations here." lol
But also I'd think that an argument could be made that a more form-fitting solution would be more appropriate, so... idk. We work with what we get. Lol
Interesting, I got bonked for that last message by automod 😅
Let me try and reword that... I think a lot of people who don't like generative AI are trying to anger disney by posting mickey mouse pictures?
There should be some universal copyright key you get to prooe thats you lol idk
I have chatgpt plus, but unforunately Im unable to use the Chatgpt vision feature, how to get that ?
it's still rolling out, be patient and it will come
unless you live in EU
or the UK
yup
Im from Europe.
then it will come later
welcome to the club
im just jealous, it has some extraordinary features, and it is getting delayed
i want the vision feature so bad, to see if it can acutally translate advanced electrical schematics
be patient, it will come
Thanks robert
is there any public reasoning why it wont come to EU or is that secret politics stuff
GDPR and member state policies maybe ?
we don't know, it's probably GDPR related
well then it wont ever come to EU if its that
is there no way to bypass it buying a new account with vpn or something
bypassing geographic restrictions is against the TOS and we do not discuss that here
yes its coming to eu and openai has gdpr checkmark on their page
oh ofc, i was talking hypothetically, for research purpeses
its possible they are just organizing a gdpr compliant storage for the images which causes extra delay
ChatGPT Plus - subscription to train your patience while waiting for GPT-4V rollout 😂🤣😂
"not YET available in UK/EU".
hi guys , do any one knows how to get access to dalle 3 from chatgpt (i'm a plus user )
wait for it to be rolled out
thank you , any idea of how much i need to wait ?
it is being rolled out this month
thank ya
I have dalle3 with gpt - if anyone has GPT vision and not Dalle3 and wants to trade prompts, hit me up. Not sure if this is allowed or not so please delete if not!
Hi, does anyone know the reason for this? My plus account expired today, but I was not shown the upgrade page. The 4.0 window is also locked. Does anyone know how to do this?
multiple people have had the same issue recently
why
we dont know, possibly a bug
😫
some have said it works better via the app
I can’t upgrade on my phone either.
What do you mean? Via the app it should work most of the time. What's the problem in the app?
Thinking of it. The vision one can help me in Photoshop as well i guess
I show it a badly photoshopped image and it can basically tell what to improve and how and where
doubt it
do you have an iOS device, can you subscribe there?
nopppsy whooopsy
Android should work too. Any reason why only IOS (just asking because I already wondered yesterday)?
just wondering why is there no resubscribing option? kindda weird and at the same time funny. 😂😂😂
I think this is a bug for today. Several users reported in #1070006915414900886 about can't subs
If it works like some people showcased, if not then not 😄
But GPT can definitelly help with Photoshop in general
Would it theoretically work if OpenAI added a way for people to specify which beta features they'd like most, and those features be the ones that people get access to first?
Can always use the #1070006151938314300
I'm just not sure if it would technologically be to hard.
When will the chat gpt api be able to access the internet and read pdf files?
it is, you have to use plugins
this week it has read all sorts of diffrent files for me.it can also create those files if you want.
there is no news on that, these are probably things you would add to your application and feed the API your data
pls explain further
the API is intended for programmers to include GPT in their applications, reading data from the web or importing text from PDF files should be coded into your application
anyone having issues with chatgpt 4 atm?
it's working as expected for me
At one point in the past day or 2, I had a pretty high error rate, but it's fine now.
chatgpt
are you sure? it happened just a moment ago and won't continue
yes, I meant ChatGPT.
seems like I can't continue a specific conversation
rip
that's weird
Maybe try and edit a message earlier in the conversation. That's fixed that issue when I've tried.
anyone know why a shortcut to go to the top of page/bottom of page might have stopped working for chatgpt pages?
used to be really handy to go ctrl+home/end
this is windows
That sounds like the sort of thing that either might already have an answer somewhere on the internet, or ChatGPT might be able to help you find a solution.
Does "rolling out" mean its random within like a week or month?
Yep
Ow. It should give you an estimate
Estimate : October
it does sound like that, but it's not working. I've tested the shortcut on other sites and it works. It just doesn't work on chatgpt
As long as it's not against the terms and conditions, maybe try using a browser extension that can help, or writing a custom browser DevTools console JavaScript script. (But only if it's not against the terms and conditions!)
If either of those methods would be against the terms or conditions, someone please let me know.
btw will we be getting that chatgpt kind of image integration in api as well?
Vision or DALL-E 3?
DALL-E 3 API is coming this year
It will be independent of gpt api?
good question, I am not sure
I am asking about their integration in api
like feed gpt images through api and get text output
that's Vision, that will come later
is the voice for chatGPT coming to pc? it seems weird only on ios and android
it's only on the mobile apps for now
Will both ChatGPT Vision as well as Dall-e 3 be invokable from a call of openai.chatCompletion.create()? Or will we have to use another function in the Python library?
I don't know, maybe someone else knows and can chime in
Someone, pls chime in. (Pls 🙏)
This channel is for chat GPT discussion. You might want to check into another channel.
Ok.
I don't think such things would be against ToS, but it sounds like an overkill. Perhaps a script might be good here, though. Good idea, thanks 👍
i want to turn the chatgpt into a stylist, and i dont know how to write a prompt, is this the place i can get help?
so i want chatgpt to answer as detailed questions as possible, such as color combinatios, advices, etc.
try in #prompt-engineering
it will be in the docs.
is there a shrine where i can beg for dall-e 3 & gpt-4v access
I also want to beg too...
ok
where is the shrine though?
are you all from american?
It's virtual.
is anyone elses chatgpt being slow
I am from Australia
Nah i am for da british neighbour hood
Anyone else still hoping against hope that voice will arrive by the end of the day? 🙂
that sounds great
How can I best use API for GPT4 for online store and website? Please give best small business advice if you have it lol
I have voice and DALL-E 3, still waiting for Vision
Hey this channel is for chat GPT discussion. You can check out the API channels.
Hi, you can check #1037561385070112779
I have vision, but I really want voice. So spoiled....
I want all of them i am more greedy
Anyone got voice and feel terribly disappointed with it?
thank you
I have it, I think it is great
Dang I was hoping it would be terribly disappointing
Why? What are you looking for in voice? 😛
its not that great tbh
actually i take that back, it's pretty standard
Voice is an easy way to eat up your 50 messages per 3 hours
Voice using the GPT-4 ?
yes
Well, my eyesight isn't the greatest so it would be sweet to be able to brainstorm and stuff like that via voice. I was using an Android app for this but it was kind of flaky. Hoping this is better.
I have been using Voice on the iOS app and the only complaints I have is that the voices are American sounding and it sometimes struggles with understanding the Australian accent
Looking forward to you exploring and experimenting when you can! Hope it meets your needs and you enjoy it. For me, background noise has sometimes been an issue, while listening to me, sometime Whisper turns the sound of running water into Korean speech. Other than that, it's met my needs well.
That's great to hear. I guess we have to keep in mind that this is still beta so it's only going to get better! Probably
Its the worst itll ever be
Hi @tranquil trench
I've purchase chatGPT on my browser and have it on my andriod mobile. but I can't seem to use the voice feature
Heyo, they're still rolling out the features! Keep an eye out through the weekend and into early next week 😎
Do check settings though, and beta settings. If you have it already, you do need to turn it on there.
Oh ok, I thought it would be available to all user on paid subs
I don't have it.. the doc announcement mention it would be under "new features"... I've check in beta features
I might have misremebered, but hope you get it soon!
I think you're right, it needs to be enabled in Beta Features in the app once it's rolled out.
Has anyone found that browser or cookies make a difference with getting dalle3 access? I know that was said to be the case with bing chat when it first came out.
no, it will just appear in the model dropdown
no need to log out, or clear cookies or anything like that
Okay thank you!
So the voice feature is not for everybody?
it's for Plus users and is being rolled out now
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.
I am a plus user but i do not see it in my app
it's still rolling out
What does that mean exactly? That there is a cue, first come first served?
it means that the feature is being progressively released to users on a random basis, when it becomes available for you there will be a little headphones icon at the top of the app that you can press to enter voice mode
Thanks, that clarifies
btw, why does it roll out randomly and not all at once ?
so they can adjust the servers for demand incrementally, and not be overwhelmed all at once.
Yeah think about how much you use something right when you get it! And imagine everyone getting that thing at the same time 😁 None of us would get to use it!
seems legit
I'm a plus user in the US and I have NONE of the new features. I'm so sad!
I am a Plus user in Australia, I have Voice and DALL-E 3, not Vision yet.
Gross. I even WORK in AI, it's like I'm being punished.
is there anyone else suffering from "unable to load history" rn
o-oh and it's completely crashed
yess, for me as well
Be patient, the new features are still rolling out
if i use a vpn in the us will i get the new features ?
VPN is blocked 😂
I don't wanna be patient. I am watching all the kids with their cool toys and I have use cases out the wazoo and no way to playyyyyy
Frrrrrr!
Is the roll out random or?
I'd be more patient if I even had like, ONE of the new toys.
not for me, i use windscribe and it's working
💀
Wait, i need to be in a specific country to get rolled out?
if you're in UK or Europe you won't get visions capabilities
-yet-
Do check settings and beta features, it does have to be turned on to show up.
I hate waiting; just roll out the new feature anytime you announce it.
Vision is the best feature
they have to scale their system...
Everyone can make a website just by drawing it
Show it to gpt
And ask for the code
only for basic prototyping tho
Is there any indication the message cap on gpt4 might stop soon?
there will always be some limit with plus based on original announcement and recent enterprise ver
i got dalle 3, vision, and voice. i think its just randomized
Hmm
You think they might introduce a higher personal tier that isn't capped? Or something unnoticble? so if you're using it for 3 hours straight with many iterations you won't get capped?
It is yeah, I got voice but not dalle and vision yet
i dont think they will ever allow individuals to have unlimited or non rate limited version because of malicious users. even the current plus version is worth +$1000 in terms of api costs
That's a shame! Wonder how much the enterprise plan costs and what the minimum number of users is.
and on an unrelated note, I'd love to see GPT Voice come to the apple watch. more useful than having siri on there haha
afaik its currently something like +200 users with a one year minimum commitment
Hi, do we know if or when multimodal chatgpt will go public?
it is already being rolled out
nope - if you mean dale-3, that's not multimodal
so like...$50k minimum
probably more like end of this month/early next
👍 this is good news anyhow
Is there some information about the vision rollout in Europe or do we still only know that we are not getting it yet?
probably at least double that if at annual level
Radium replied my same question 4 minutes before 🙂
Hey Guys, does web browsing work for you ? It suddenly stopped working as of lately.
it's working fine for me
Where are you from ? maybe location issue ?
Australia
I'm about to do my morning GPT refresh in hopes of vision, wish me luck!
negative 😢
Sad
what i mean is, on chat gpt plus, i can actually use askyourpdf plugin to read pdf file, and i can access websites using webpilot. Will that ever be possible to use in API mode? If it is already a possibility, i would love to know how.
It probably will be, which is why I won't build an app that simple right now.
it's not possible with the API, the expectation is that you would implement that in your app yourself
We're in the churn stage of evolution, when millions of wrong ideas fight for survival, only to be eaten by the largest fish in the end 🙂
so you would not build an app for what i need because you expect it soon?
- It already exists, millions of those pdf reader apps
- Many similar ideas I wouldn't touch.. like some app to manage your files with AI
what do you mean by "implementing in your app yourself"? That chat gpt plugins will possibly be integrated to the API?
well, i mean, i have to use a chat gpt plugin for reading pdf's at the moment, since i will not be able to read a pdf without. right? or am i missing something here
no, what I am saying is that the expectation seems to be that if you want plugin style features that you would build those into your app
Yes, thats the best you can do right now. Why would I build that, when GPT will have full multi-modal at some point, probably soon.
okay, sorry but i don't get what you mean. Im not building an app, i just want to use the API with plugins such as askyourpdf
yeah okay
Plugins are not the same as APIs, 2 different purposes.
i know.
You're saying you want to use the API with a plugin.. you'd have to build the plugin to do that.
and many already do that
how are you using the api if not via own app?
but what i mean is. I want to get the API to read a pdf, then access the internet. And that's it. I don't want to build my own plugins for that. I want to use existing plugins such as askyourpdf and webpilot
The API takes text only right now, end of story. So it's PDF -> Text -> API, at a minimum.
some openai devs have said at twitter that its up to devs to implement them on their apps. things may of course change but no likely to come and not been announced
okay, i mean yes, i have python code with a system prompt, and one user prompt (this prompt should ideally feed the pdf), and a second user prompt (this prompt should make chatgpt access a website and return some values based on waht's on the website)
so you have an app where you can add pdf reading libs or ones that access net and scrape the data from html
yeah, i guess my hope was that i could use plugins in colab with the api. just as you can use plugins (such as askyourpdf and webpilot) in colab with chatgpt plus
Then there's PLUGINS, which live in the UI chat interface only.. and might or might not call apis, but always do to be useful.
plugins are made for end users, i doubt the plugin authors would allow api users to earn money while consuming their server resources
yes, but i can't do that atm with the api
and if you are ready to pay then there are already pdf reading api services or even things like langchain
because openai does not have a pdf reader, those are all by 3rd party
its not hard if you can already make python scripts.. parse the pdf to text, then pass to the API (note: you'll encounter size issues to work with)
ah, is that all the askyourpdf does`? i thought it was more tricky
im sure they have other custom prompts, vector db, etc. but at the end of the day yeah, thats all it is.
yes, my biggest problem atm is the context window for API. its so short compared to the chat gpt web version
web has smaller context than api, much smaller
yeah, you need to learn about vector databases, embedding, and searching vector dbs.. if you want any kind of memory in the app.
Or just use one someone already built lol
that's because with the API you're supposed to feed the context to the model in each request
I don't understand why anyone would use the API unless they are developing or deploying an app, using Playground is for testing not general consumer usage
hax0rz (jk). agree. its for developers lol
so webpilot is basically just a third-party that is scraping/accessing the internet, and then converting the content of a website into text which is then fed to chat gpt?
i got all the things except VISIONAAA
same 😢
its the missing piece to a full system
yes, we are never complete. 🧘♂️
so webpilot is basically just a third-party that is scraping/accessing the internet, and then converting the content of a website into text which is then fed to chat gpt?
you'd have to read their code to know for sure, but all of these AI powered apps do the same thing at the root.. augmented retrieval and searching.. the magic sauce is their prompts and overall flows etc. like any app i guess.
but i guess webpilot is not anything with prompting, since its just a tool to access a website
right?^^
I haven't been impressed with any AI powered apps, yet 😉
but when you guys say i could just use a third party. Then yes, i would love to use webpilot as third party, for when im building my own app (using the API). But i can't do that, right?
Check their terms you agree to when using their app.. cant know beyond that I guess
I believe OpenAI gives you ownership though
but technically, using webpilot is not possible, since it¨'s only usable through the chatgpt plus
right?
i don't know what it is, just talking vaguely. is it a plugin?
afaik webpilot offers its own api for a cost
it asks me to sign in before telling me much, but i saw the word gpt plugin
it lets you prompt gpt with a website, that it then visits live and extracts information from
thereby making chat gpt "live"
ah, simple python script could do the same.. easy for smaller websites if your entire prompt fits in context.. if not, then what i mentioned earlier with embedding the website first then searching it
just use theirs though, if you dont want to learn abt embeddings, because its hard to avoid
yeah, but i need their api before i can use it in my app, right :)?
btw, currently i have a problem when generating text with chatgpt API, because the context window very early hits the limit. Any ideas on how to solve that?
🤷 thats not an openai question. I don't care to read their API docs 😄
any other plus members still not have access to the new voice chat or vision features?
nevermind just saw it is by tuesday
I know the image functionality is rolling out to plus users, just wanted to make sure is it coming to PC chatGPT or just the app?
Wanna race to be the last to get everything? 😄
*anything. I don't even have DALL-E 3
Aww. And you do know about app-only for voice, and beta features to turn it on there, plus in web browser turning on settings-beta settings for Bing-assisted search? To look for the image icon by default GPT-4 and the Dall-E in the GPT-4 drop downs? Everything is accessed its own way....
I just noticed, looking at my history, that voice actually seems to use GPT-3.5, unless it's being labelled wrong on the web app
Yep, I check all the feature toggles and drop downs every day. No voice in my app, no vision or image creation in my desktop.
Cheering you on! Here's to soon for you!
For me, voice works with either 3.5 or 4, I can select at the top of a new conversation which to use. It is default 4, of course.
oh cool
use less text.. chunk things up
abstractions work wonders too, if you can abstract to less text but keep the original meaning / details
also, ask GPT4 to help craft your prompts if you aren't.. its an expert
The Bing app has vision, midjourney, gpt4, voice. The official OpenAI app has none of it for me. Why do i pay for this?
well wait until it gives to everyone
I'm waiting for this Monday to get my GPT-4V. That's when OpenAi's "handing out over 2 weeks" finishes
you can not compare the ChatGPT with Bing. For me Bing is very very simplified version of chatGPT
I expected to get them today and play with them during the weekend, so sad!
Anyone had a chance to get them today?
All we can do is wait. But we'll get the new feature :')
It is really not simplified if it has all the features we have been waiting.
did you actually try to use it for good? it has chat message limit and can end the chat at any point, thread limit, image gen is based on few credits per day and then its hours of waiting on top of that the images are small res and square
i use it daily but its kinda like free/lite version of chatgpt plus
So? The official App has none of this.
bunch of people already have the new features
For me nothing like Chatgpt
⬆️
to me, Bing has its advantages and disadvantages... it's similar to using ChatGPT4 + Browsing feat : I'm sometimes FORCED to disable browsing because I get more accurate results by letting GPT only, ignoring all the crap coming from the Internet
for gpt-4, if i use up all my 50 prompts, it says you can ask again in x amount of time. when that time arrives, are you immediately given another 50 questions or do they just trickle thru?
it's 50 messages per 3 hours in a rolling window
ok so it's 50 messages for whatever the last 3 hours were?
Whats the difference between the free and paid version of Chat GPT?
Free gives you ChatGPT GPT-3.5
Plus gives you ChatGPT GPT-3.5 GPT-4 Advanced Data Analysis, Browse with Bing, DALL-E 3 (soon), Voice, Vision (rolling out now)
How soon are we able to start using DALL-E 3 is there any good estimates?
I have it now, the rollout is October - ish
Yes I was hoping to get access to the voice today too. 😦 It's too bad. Well I guess they're still a little more time. When I got the vision thing a few days ago it was in the early evening I think.
Will you try a prompt for me? I've had it saved from over a year ago and I'm curious how well it can engineer this one since its kinda obscure (I am a paid subscriber of GPT+ since the beginning!)
I can try dalle3/chatgpt if you want
first optimize the prompt itself "Waluigi with a samurai sword confronts the emperor of Japan on horseback wit a background of cherry bossoms and an erupting volcano"
Right like this is a pretty good test
PM me with the prompt if you want to get the image
that would be blocked by the filter, waluigi is nintendo IP
It wasnt blocked: "A tall, lanky character reminiscent of Waluigi stands confidently wielding a samurai sword. Opposite him, a regal figure representing the emperor of Japan is mounted on a majestic horse, prepared for the confrontation. The scene is set against a breathtaking backdrop of cherry blossoms in full bloom, adding a touch of tranquility to the intense standoff. In the distance, an imposing volcano erupts, sending plumes of smoke and ash into the sky, further intensifying the atmosphere."
The fictional character Waluigi, wielding a samurai sword, stands determinedly facing a public figure on horseback, who represents the emperor of a historical Japan. Behind them, cherry blossoms are in full bloom, painting the scene with shades of pink and white. In the distance, a towering volcano erupts, spewing smoke and ash into the sky, adding a dramatic backdrop to the intense confrontation.
That’s so odd
I tried continuing an old text adventure chat
But it seems to just fail on the prompt because it didn’t have access to old context
How?
Like the whole chat is still there but the AI is just incapable of continuing it
It's always limited by the number of tokens isn't it? Like GPT 4 is 8K at this point? Are you on 4?
Yes, for me was early evening around 5:30 pm.
So we still have hope for today. 🙂
I went back to a 3.5
Add your well-crafted prompts to our #1019652163640762428,
or share your interactions with ChatGPT in #1050184247920562316!
I suspect it’s the overwhelming amount of filters that got implemented on it
That’s why people are seeing quality decline over time
Sure they are to prevent bad stuff from being generated but I don’t think it’s worth the quality slip
its changed, and hates anything resembling copyright.. i wouldnt say quality is down though. have you seen some of the knockoffs?
Yeah I haven't noticed a declining quality either. But it may depend on what your asking it
I always have my super verbose expert personality custom instructions on too, probably keeps me floating above the average
which is when for u?
sorry i mean.. where are you located hahaha
Vision
Care to share? I'm using a variation on Sam Altman's custom instructions which can get a bit terse at times. 🙂
# Assistant Response Complexity
**Note**: I may control the verbosity (detail level) of your response by prefixing a message with `V=[0–5]`(default V=3), on a scale where `V=0` means terse and concise, and `V=5` means most verbose and comprehensive
# Primary Assistant Guidance
Your goal is to provide in-depth, expert, and accurate analysis and opinions across all fields of study. Let's go step-by-step:
1. Is my question (wrapped in parentheses)? If yes, skip to step 6
2. Carefully evaluate every question from me, and determine the most appropriate field of study related to it
3. Determine the occupation of the expert that would give the best answer
4. Adopt the role of that expert and respond to my question utilizing the experience, vocabulary, knowledge and understanding of that expert's field of study
5. Respond with the expert's best possible answer, at the verbosity requested, and formatted with this template:
"""
**Expert**: [your assumed expert role]
**Objective**: [single concise sentence describing your current objective]
**Assumptions**: [your assumptions about my question, intent, and context]
[your response]
"""
6. if you have any suggestions for more context or online reading, add them with links to the end of your response as a markdown blockquote ("> " prefix)
7. any links you include must formatted as described in "My Expectations of Assistant"
**Remember: (questions in parentheses) don't use an expert**
I think just being a balance of kind but definite with it? like a nice lesson for life. state what you need! but dont be a dick! haha ?
I've found the assumptions and objectives to be very helpful in course correcting
ohhhh i like this
will try right now
let me know! i didn't make it btw, copied off reddit long ago
why does it say dont use an expert i wont? ive been tellling it its an expert and it works better than my half baked prompts
ive also been getting my prompts rewritten by it before i put them through
by getting it to assess and refine my prompts as 'prompt expert'
as a **
yes, everybody should be leaning heavy on GPT for prompt engineering
its a virtuous cycle
Thanks. I wish there was an easy way to have multiple custom instructions on my phone.
sooo virtuous
yeah interface has a lot of room to grow
but websites like flowgpt are saving that
Yeah can you explain that last bit?
Chotes pre instruct said this at the end
**Remember: (questions in parentheses) don't use an expert**
I have no idea why that reverse psychology works
yeah baffles me...
maybe its dont USE one, BE one
wait i know. ill ask it itself mwahah
who knows lol
TRUE
Good idea, let us know?
Expert: Cognitive Scientist specializing in Human-AI Interaction
Objective: To clarify and explain the rationale behind the guideline of excluding expert mode for questions wrapped in parentheses.
Assumptions: You are questioning the benefit and underlying reason for such a guideline, given that it appears counterintuitive. You may assume that some users want rapid, direct answers without the full expert-style exposition.
The guideline, "Remember: (questions in parentheses) don't use an expert", may seem counterintuitive at first, especially if the goal is to get in-depth and authoritative answers. However, this guideline can serve a few specific purposes:
Quick Access to Information: While expert-mode offers a detailed exposition, not all users may want such a comprehensive response every time. By wrapping a question in parentheses, the user is signaling the AI to bypass the more verbose answer style and deliver a direct response.
Cognitive Load and Efficiency: Not every question requires a deep-dive analysis. Sometimes, users may seek a brief answer without wanting to read through a detailed explanation, particularly if they're in a hurry or only need a quick fact-check.
User Customization and Flexibility: Providing a mechanism for users to control the output verbosity offers them a tailored experience. By allowing users to toggle between standard and expert mode based on their needs, they can customize the AI's interaction to better serve their immediate requirements.
Minimizing Redundancy: If a user is asking a series of questions and doesn't wish to see the expert format each time (which may reiterate some points or follow a template), they can utilize the parentheses approach to cut straight to the answer.
Design for Exceptions: As AI gets integrated more deeply into workflows, there can be scenarios where a user understands the typical mode of operation (e.g., expert mode) but occasionally requires an exception. The parentheses guideline acts as a quick switch, facilitating this exception without making the user go through a more cumbersome process.
To summarize, while the expert mode provides valuable in-depth analysis, there are scenarios where users might prefer directness and brevity. The parentheses mechanism serves as a shortcut for those occasions, giving users a tool to effectively communicate their desired response style to the AI.
For further reading on user experience (UX) design and the importance of user customization, you might want to explore Don Norman's "The Design of Everyday Things". It's a seminal work in the UX field and touches upon the significance of designing systems that cater to user needs.```
i guess?
Ohhhhhh
wait i see. the pre instruct is specifying questions in parentheses.
Huh
to not be approached as expert
Pity that following question in ChatGPT is not giving guidance on the rollout phase for specific user
“When I will get access to Dall-E 3?”
parentheses around your question are basically a shortcut out of the pre instruct
It's almost 5:30 people! Get ready!
😂
what time zone mate??
7:45am here
Not true. 22:45 here
Oh, I think Eastern in the USA. I'm not even sure where open AI is. Maybe California?
ahhh ye
so end of business day is the phase rollout?
interesting.. is that consistent? id have assumed a random trickle at all hours
is this just hype... mr KB????
😢
Yeah that expert mode could get annoying We just want a simple answer. I don't know I'll have to try it out.
roll out finishes on Tuesday
Thanks Rob
Well I'm hoping. It happened one time anyway. 🙂
Ive waited my whole life for this.. i guess could wait even a few more weeks
😂
get ready for some mindblowing feature from another company to be announced within days of the end of this rollout
this gravy train aint slowing down
Hell yeah let's keep this buzz going
do plugins work with the new voice chat thing on phone?
no
is voice chat still rolling out i got vision but no voice
i would talk to it all day :D
oh, I got voice but now vision XD
i got voice but no vision.. going BLIND over here! 😂
I'd love vision too :D
tbh i took the voice for granted..
i prefer typing. but everyone i show voice to is freaked out.. i thought voice detec and synth was simple.
it doesn't sometimes understand me because of my "not english" accent
does it understand other languages by the way?
oh god so good my parents dont know english i would like them to try it with their mother tongue
oh haha
Speaks Icelandic even. Weird accent but works fine.