community-help

[steady cedar]

Oh okay, so rn it is strictly for businesses?

[regal lynx]

Yes

[idle socket]

It allows ChatGPT to see photos and recognize them like a human would

[steady cedar]

So iOS is getting the vision feature first?

[acoustic violet]

I’m hoping, that would be the most useful place to have it.

[steady cedar]

Uhhhh some of us don't have iOS lol

I guess they probably have the most Plus subscribers on iOS though

[timber shadow]

The post doesn't specify whether one platform is getting it first--it mentions them both.

[steady cedar]

I mean they don't want to show favoritism, but they did push out iOS significantly first for their app, so I assume their gonna keep that schedule for this release

[tiny roost]

Anyone else getting this when trying to access the website?

[timber shadow]

I'm not sure if it's favoritism as much as ease of development--iOS development can be more streamlined because it's much less diverse. Hopefully now that they have a full release on both platforms, the update delays are minimal! 🤞

[steady cedar]

I mean Android does have a larger platform, I have coded on Swift and I would much rather code on a Java-like platform than on Apple's proprietary software. Also, Android is only diverse because they allow creativity, Apple doesn't lol

[timber shadow]

I'm not saying it's good or bad--I'll use Android forever--I'm just talking about why it is the way it is! The important part is we'll all have the features for much longer than we're waiting for them now 😃

[steady cedar]

Well yeah at least they are including us haha

I know softwares only on iOS but i guess if you're in the American market, it's the smartest move

[fierce haven]

will it be released on browser first as well or?

[timber shadow]

There's nothing specified with regard to release order. We can expect it to be released on everything within 2 weeks.

[fierce haven]

💯

[trail cairn]

by now you should have already created several test embeddings to determine that

[storm summit]

Not really i think OpenAI are very fair

for instance i have access to the voice feature and that's on the andorid version

Android Beta

And btw likely if they do give us access to the 32k version of the gpt 4 model it will likely be capped and gpt 4 will likely be uncapped just a assumtion

the 8k version uncapped

[timber shadow]

Oh yeah I wanted to double check with you on that, I think you're the person I saw yesterday say they have it? I think you said the option for it was under the Color Scheme in settings, and I have a Speech section there, but it's just for the voice to text feature. Just want to confirm that you have a separate feature that's specifically for the new conversation feature where ChatGPT talks to you?

[storm summit]

Or perhaps gpt 4 as free version and plus version is both 32k and gpt 4 with all the plus features

[steady cedar]

isn't 8k already uncapped? ChatGPT plus is not the 8k version

it is half the context size

[storm summit]

Ah no probelm i can explain so the voice feature works on any of your chat doesn't matter which chat history you have and for those wondering! It works also for 3.5

8k is gpt 4 on the standard version

gpt 3.5 is 4k context length a tad bit higher

[steady cedar]

For the API, not ChatGPT UI

[storm summit]

And i think ive spent like ugh several hours on it 😄 its actully so cool

[rustic marlin]

they could either publish what they had right away, or keep holding the app's release for a while waiting for both apps to be done... I don't see the problem in releasing what is done when it is done

[steady cedar]

the UI you cannot input more than 4k tokens in one message

[timber shadow]

Right, but are you describing the voice conversation environment shown in the demo video? Where no text is displayed, just a voice conversation environment?

Or are you describing just voice to text?

[storm summit]

The back and forth chat

you press a headphone icon and its like a phone call connects to gpt and you talk back and forth

[steady cedar]

Well they didn't do that with the app release, just saying we're probably gonna have to wait longer, but it isn't that bad of a thing

[storm summit]

Just can't wait for the image feature

🙂 am thinking reading comics and manga and perhaps with API i can finally have an assistant on my screen giving me useful information in game 🙂

Dota comes to mind

LOL as well

[storm summit]

voice conversation

[smoky pilot]

OK. I'm now developing it with postgresql. What I want is your experience. And your advice.

I think, there must be limit. Can you tell me the limit of it?

[trail cairn]

you have been asking that same question for several days now but without trying to use them all

[arctic holly]

I like the voice feature. I have make ChatGPT hallucinate for become a talk therapist and it’s very rewarding. I can’t wait until the voice feature is released

[warm rose]

somebody knows if the new voice feature is just better and hidden text to speech or if its true GPT voice? like Bark for example.

[atomic cloak]

For an input of 300 tokens and an output of 300 tokens, how much do I pay with the ChatGPT 3.5 API?

[acoustic violet]

About a 1/10 of a cent

[atomic cloak]

Ok thanks, how can i calculate my costs?

[trail cairn]

https://openai.com/pricing

[atomic cloak]

👌🏻

[storm summit]

They literally do say they're using their whisper but am assuming it's the bigger whisper model as it's very damn accurate and I've not really had situations where it doesn't understand me 🙂

[rugged wing]

is there any way of changing the email connected to my openai account?

[regal lynx]

There is not. Your email and phone number will always be tied to the account

[rugged wing]

I have lost my email id that was connected to my open ai account, can i use my phone number to recover my account ? If i ever need to

[regal lynx]

I don't know. You'd have to visit the help site for official account support

[rugged wing]

Oh i see, thank you very much for the help.

Hi, i have a question related to this new image upload option. When i take a screen shot of a text with a diagram, the response from gpt is it cannot read and analyze the text, and i have repeatedly tried it, but its not working. Any tips please?

[timber shadow]

Are you trying this in Advanced Data Analysis? That model can take file uploads, but it's different from the upcoming sight features. The phased rollout of the new features is not widely available yet, so most of us are still just waiting!

[rugged wing]

I have the image upload option, but the only problem is that it cannot read the text or analyze the diagram properly. Im trying to see if this is because im making a mistake uploading the image, or is it because the feature is new, or any other reason?

[timber shadow]

Is this a screenshot from you using the Advanced Data Analysis model?

[rugged wing]

Yes

I uploaded the diagram with a question related to the diagram which is also in the image, but it says that its not visible properly

[timber shadow]

Uploading a file to Advanced Data Analysis is not the same as the upcoming image sight features. That feature is not fully available yet.

[rugged wing]

Oh ok. Do you know when is the date that this feature will be fully released?

[regal lynx]

https://openai.com/blog/chatgpt-can-now-see-hear-and-speak
You can read about the feature here

[late summit]

Is there any word on when GPT 4.5 would theoretically come out for Plus or general population users?

[regal lynx]

No GPT model above 4 has been announced. We don't know if one is even being trained. So no, we don't know if / when something new would be released

[remote terrace]

Hey All, Im stuck getting this error no matter what i search does anyone have any advice or no of a place i can contact open AI to try to get this fixed. thanks 😄

[lethal panther]

Everyone will have GPT with vision within 1 week and 6 days lol.

[tranquil thunder]

Anyone know why I keep getting this message? For 3 days now I have gotten this error message, changing browsers, incognito mode, turning off VPN nothing helps

[loud furnace]

Hey
I have a doubt that can be really stupid but...
I need add a profile picture to Chat GPT but I cant find how.
Can someone help me?

[timber shadow]

Create a Gravatar account using the same email as your OpenAI account if you don't already have one. Set a profile picture there, then just wait. Your OpenAI account and Gravatar account will eventually sync up and your Gravatar picture will appear on ChatGPT.

[loud furnace]

How long it will take?

And can I set it only for Open AI or it will apply to all my acounts?

[timber shadow]

I'm not sure how long it takes, for me it took a few days I think. I don't know if Gravatar lets you set specific photos for specific websites.

[loud furnace]

Okey Thank you

Really Random that you have to use a third solution for setting your profile photo

[timber shadow]

Yeah I've known about Gravatar for awhile but never had the need to use it. I'm sure it's just something that works for websites that don't want to set up their own avatar systems yet--quick solution for letting people have one in the meantime. I'm sure in the future they'll add a native profile picture alternative.

[fluid merlin]

Why is the longer context window for processing longer inputs available only to ChatGPT Enterprise subscribers and not to ChatGPT Pro users?

[fringe pagoda]

I just got the speaking feature, but I can't upload images to chats through the app, do I have to wait for this feature as well?

[rigid musk]

Yes

[delicate igloo]

hello everyone, i have been trying to add credit to my openAI API account, but no matter what i do, it will say "your card has been delcined". I have tried different Cards, Banks, IPs, Browsers, Devices, Countries, Addresses, And nothing has worked. Has anyone dealt with this before?

[arctic holly]

Lucky. I have been a plus member since day one and nothing

[regal lynx]

the app doesn't support Beta / Alpha features I don't think

[arctic holly]

The voice feature will be for the iOS and Android apps according to the website

[regal lynx]

the voice feature is available in the app. Other features like plugins and advanced data analysis aren't, though. Going off of both those features being Beta ones, I'd be surprised if other beta features went to the app

[timber shadow]

The announcement blog post makes it sound like both hearing/speaking and vision with both be launched on mobile as well as desktop. Would be kind of funny if those two new big features make it to mobile before plugins & code Interpreter 😁

[regal lynx]

oooh, I didn't see that part of the announcement! I'll check it now

[onyx kraken]

I'd love to have voice on the desktop, but I see this in the blog, "Voice is coming on iOS and Android (opt-in in your settings) and images will be available on all platforms." Do you see something else?

[timber shadow]

Sorry, I totally remember reading that too, and nothing to the contrary!

[regal lynx]

To focus on a specific part of the image, you can use the drawing tool in our mobile app.

To get started with voice, head to Settings → New Features on the mobile app and opt into voice conversations.

This is what I'm reading

[timber shadow]

Hmm so maybe drawing isn't available on desktop, only file uploads? And voice only on mobile.

[onyx kraken]

It kinda does seem funny mobiles would get new feature voice before older features plugins and advanced data analysis... but I imagine it's harder to use those on a mobile device, tiny screen, no mouse to aid copy/paste.

[fringe pagoda]

Image input will be generally available to Plus users on all platforms. Basically the app should get it too, right

[regal lynx]

the announcement shows someone on the app using the vision feature

[timber shadow]

The videos in the blog post clearly show image/camera buttons, so hopefully.

[regal lynx]

https://openai.com/blog/chatgpt-can-now-see-hear-and-speak

[fringe pagoda]

Yeap, got the audio feature but no vision

[regal lynx]

I'd be surprised if anyone has vision yet

[timber shadow]

It's nice to read about some people getting some of the new features! Can't believe they're dropping these two big features and DALL•E 3 so close to one another.

[regal lynx]

Vision and Voice were announced a day ago, and plus users will get them over the next 2 weeks. There's a chance the rollout for them are independent for one another, and that's why you have voice but not vision

[onyx kraken]

Oh, well... We on desktops can use paint or whatever else to annotate or crop our images as desired.

I could put a circle around a key part, if the model's going to be trained to look for a circled area of importance, or I could use two pictures and tell it the second's the part of the first image I'm asking about, but showing both pictures for clarity. Since it says 'discuss multiple images'.

[timber shadow]

A very good point! And if the native draw feature isn't good on mobile, I suppose I could always just pre-markup photos for that too!

[onyx kraken]

Might have to pre-trim, or draw over/mask the not-desired parts. We don't know if the model will be able to identify our somewhat random circles as having clear and useful meaning to guide it with the picture.

I'm so eager to test that, among other things, when it's time 😄

[timber shadow]

For sure, me too. I can't wait to test if I can do things like indicate lengths, distances, etc.

[onyx kraken]

Semantic analysis is fuzzy. But images are quite, quite measurable. We can pretty much test the heck out of the model's vision and what it can and can't see or understand in the images.

It'll be filtered through semantics, the meaning it gives words. It could have one understanding, then that gets filtered through the neurons that pick the words to express it, so it explains differently than it actually receives. But that's detectable too, potentially, with small difference pictures and enough data. So much to dig into here!

[timber shadow]

Has there been any indication about potential use limits? Whether it will just be part of the regular GPT-4 usage cap or if it'll be separate, etc.? Same for voice and DALL•E 3--could possibly slow down the digging!

[tribal talon]

Nothing has really been announced about usage or limits so don’t think anyone will have a solid answer for you yet. It doesn’t seem any different than usual chatgpt limits from all I’ve seen but who knows

[onyx kraken]

If we can combine it with advanced data analysis, we should have that capacity...

I haven't tested it, but that's a staple in many scientific images, you include a known scale measurement at the same distance from the camera as some focal point, and have everything else you care about in a known relation from the focal point. So you can then adjust for differences in size and camera angle together - or more usually take a series of pictures, each with is own current focal point measurement.

But that's probably straight out ADA (advanced data analysis), the AI doesn't need to be able to 'see' too to measure, if you can photograph a scale length.

[timber shadow]

Hopefully they can get it all to operate under just one usage cap! Feel like that would simplify the user experience, even if that means usage caps won't increase for awhile. (Hopefully they don't decrease! 😬)

[timber shadow]

Good points, I appreciate the brainstorming as I begin planning to try home renovation projects with this new feature 😁

[fierce haven]

But is it? Because on the release notes they say "Voice (Beta) is now rolling out to Plus users on iOS and Android" not on browser version?

because image input:
"Image input will be generally available to Plus users on all platforms"

[onyx kraken]

Hope you share what you find you can and can't do, protecting your privacy of course in the process!

[timber shadow]

Yes, you're right, voice won't be on desktop! Thanks for correcting me.

[timber shadow]

Found a little more info re: usage limits. There are dedicated FAQ pages for the voice and image features.

On voice they mention you can use 3.5 or 4, but that the regular cap for 4 still applies.

https://help.openai.com/en/articles/8400625-voice-conversations-beta-faq

And on images they specify that it's only with 4, but don't clarify on the usage cap, but probably safe to assume it also falls within the same usage cap.

https://help.openai.com/en/articles/8400551-image-inputs-for-chatgpt-faq

[tribal talon]

Seems about right. Voice makes sense as just an extension of what’s already there. But spoken

[fierce haven]

dang I hoped so. But I guess I just mirror my phone to my desktop or use android emulator.

modern problems require modern solutions

Bing AI got gpt-4 and also image inputs for who doesn't pay for plus

don't know why bing AI has it earlier than the release of it in chatgpt (the image input)

[regal lynx]

It's rolling out to plus users over the next 2 weeks. There's a chance, you haven't gotten the update yet

If you don't have ChatGPT Plus, I'm not sure if / when you'll get it

[brazen wave]

@regal lynx I have plus, I am so sad that i don't have conversation mode, is there anyway i can expedite it?

[split hazel]

Speech in ChatGPT is rolling out to all Plus and Enterprise users over the next couple of weeks, so it should arrive soon 🙂

[sterile tendon]

Not sure how I’m supposed to handle the “sometime in the next two weeks” uncertainty of the Speech rollout.

[arctic holly]

I understand

[late summit]

Hey sorry if this is already known but can GPT 4 read the links pasted in the prompt I have heard both that it can or cant

[plain tusk]

does having a chatgpt subscription not allow you to use the credit card to buy the api?

[acoustic violet]

The api and ChatGPT are two separate things. You’ll have to get both (their payment works differently though)

[wispy palm]

hey folks, is rollout of the new voice (TTS) coupled with rollout of the new image features for Plus users?

I'm asking because I've just been granted the image feature on web and mobile, but I can't seem to find the voice toggle in settings on mobile. Are these separate gradual rollouts?

[oak wagon]

anyone has problems with logging in?

I'm having the same issue. Has it resolved for you?

okay I'm good now

[arctic holly]

Some people have reported no access to voice but they got access to images

[oak wagon]

just need to wait afterall

[rain night]

What's it look like to get GPT-4V in CGPT? Is there a pop-up notice or just a new interface for image prompts?

[coarse harbor]

Will the ChatGPT iOS voice feature be exclusively for Plus users in the「future」?

[sullen crystal]

Hello, I'm using GPT4 on my computer browser. In the past two days, there has been an anomaly. Every time I ask GPT4 a question, I have to verify that I'm human by rotating through captcha images, 10 images each time. After verifying the 10 images, when I ask another question, I have to repeat the verification of 10 more images. This repeats endlessly, making it impossible to use GPT4 at all. However, when I use my iPhone to ask questions, I don't encounter this problem. I'm also a paying member of GPT4. I used to be able to use GPT frequently and normally,

My usual IP address for internet access is from Osaka, Japan. I was able to chat with GPT normally before, but now I have to frequently verify captcha images. When I switch to a random Taiwan IP using VPN, I don't need to do the verifications. I haven't done anything bad with this IP address, please help resolve this issue.

[valid hull]

Why am I being asked to verify I am a human again and agian and agian?

[sick veldt]

What are some other dall•e struggles other than drawing hands and signs/letters?

[wispy palm]

Got it, thanks. That's what I'm seeing too. Guess I'll just have to wait

[wispy palm]

Well, are you sure you're a human? 🤔 👀 (lol jk). Try clearing cache/cookies and renewing network DHCP lease. Are you using a special DNS setup or a VPN or anything? That might also contribute

[valid hull]

@wispy palmyes definitely I am a real human. I cleared cookies and tried the Google Driftless Mode, but still be asked for verification.
Yes I am using VPN. but could you just check me once or twice one day, but no need to check each talk.

[wispy palm]

Yeah it's probably something related to your VPN. Probably something about its servers is being falsely flagged as bot traffic or something. Maybe you should file a support ticket with OpenAI to get more info

[valid hull]

@wispy palmHow can I get the "support ticket"?

[wispy palm]

I think you can go to help.openai.com and then click the button that appears at the bottom right:

@valid hull ^ (sorry forgot to tag you). Someone please correct me if I'm wrong here.

[valid hull]

@wispy palmThank you, but I firstly tried this but found helpless. Then I was led to here. Can I solve the problem here?

[lethal panther]

I had access to it yesterday but it disappeared lol

[wispy palm]

Dang that's rough. OpenAI giveth and OpenAI taketh away

[wispy palm]

Sorry, without seeing what your network traffic looks like no one here can help you :\ Maybe try turning off VPN?

[valid hull]

@wispy palmwhich means turing off the ChatGPT😢

[wispy palm]

Oh dang, I'm sorry 😦 Good luck with your captcha issues, maybe someone else here has other suggestions

[valid hull]

@sullen crystalHi Ray, I have the same trouble with you. Have you solved this?

[sullen crystal]

I haven't resolved the issue, but I just submitted feedback through the small window in the bottom right corner. I've been experiencing this problem for the past two days. How about you?

[valid hull]

same same, from yesterday. But I didn't find any useful info from their "help center"@sullen crystal

[feral jasper]

Why I don’t have the new features in ChatGPT App?

[foggy tree]

Theyre rolling out slowly

[fossil tusk]

They are 🙂

[lethal panther]

Well you might receive them at different times (like I got access to the new tts but not vision), but you will get both within 2 weeks

[lethal panther]

Lol yeah. Though personally I’m no too focused on the tts features, but much more the vision features

[frosty galleon]

The page not loading error has been resolved

[smoky tapir]

How will I know that image is available on my plus account ?

I’ve been to my chatGPT app on iOS. The is no « new feature » option in settings to enable voice

[rigid musk]

Check time to time

[rain night]

I'm curious what to look for. Will it be obvious that I can submit image prompts? Or even a notice?
Just eager to try it. 😝

[arctic holly]

From what I understand this is a feature that will need to be activated in your iOS or Android settings. Being that iOS apps have not gotten the new beta features before but the mobile website has and sometimes the mobile website has more features it’s just confusing to people

[twin palm]

Uhh no

[analog shore]

Vision has been activated on my android APP but cant find voice in the options... is this rolling out?

[rigid musk]

#announcements

[analog shore]

Yep, im coming from there... maybe my limited english skill wont let me notice the how to activate voice

[timber shadow]

The features will all be rolled out in full by the end of the two week period, so voice may not be live for you yet.

[analog shore]

Nice, thanks for clarifying that, ive thought it was simultaneously. Image recognition is out of this planet...

[timber shadow]

Can't wait to try them both out!

[foggy tree]

Da heck

[timber shadow]

Omg can't even imagine your custom instructions 😁

[foggy tree]

Breh i just wanna make a phone call with chatgpt and ask it to act like kratos

Gonna have a 1v1 with him fr

[analog shore]

lol that is a good idea

[foggy tree]

Man why am i not getting iteinfoenfr

Its not there for me yet right?

Or am i looking at somewhere wrong

[rigid musk]

#announcements

[foggy tree]

its still rolling out for me prob

[rigid musk]

For millions of people yes

[fair oxide]

How can I update the email/name on my OpenAI/Chat GPT account?

[rugged cedar]

I'm unsure how organizations work I made an account for the company and added my main account the company account has GPT-4 there are two chats there how do I access them on my main account?

[stark plover]

Hi all anyone know why pasting in 5500 words is "too big an input" (message) when I have gpt4+ which has a limit of 25000 words?

[rigid musk]

Contact [e-mail]

[quasi pecan]

does anyone know why im unable to see the other features? ive been subscribed for a while and haven't see nay other the other features?

[rigid musk]

What other features?

[quasi pecan]

the plugins?

also not sure if the new gpt4 vision is released for public users or not yet but that as well

[foggy tree]

Its right there

[rigid musk]

Start a new chat and select plug in, go to plug in store and take the one you want

[foggy tree]

Oh vision is still rolling out 🌚

[rigid musk]

#announcements

[quasi pecan]

i swear it was not there before. sigh, ok thank you yall

[foggy tree]

[narrow ridge]

Hello wcweyonw

I am using chatgpt4

I somehow can not find the plug ins

what happend?

[feral jasper]

What’s the difference between chatgpt-v4 and gpt-4

[rain night]

I'm wondering this too. First time I've seen ChatGPT "versions", usually OAI refers to the model just as GPT-3.5 (for the underpinning model for 'default' in CGPT) or GPT-4.

I guess they also include the date of the current ChatGPT version on the service itself.

[arctic holly]

Memory. You’ll notice overtime if you keep up one chat that chatgpt will be begin to forget what you said earlier in the chat

[rugged cedar]

I'm unsure how organizations work I made an account for the company and added my main account the company account has GPT-4 there are two chats there how do I access them on my main account?

Is it only for API? Is there a way to switch accounts easily?

[trail cairn]

api only

[rough tangle]

Hello I heard you are able to use text from a website to add context to your request with out needing to copy the entire site, how is this done?

[rugged cedar]

I got flagged for saying "I need to jump with a dash and then go down quickly before the jump so basically jus 0.5M from the ground or something makes u sof astr" 💀 what exactly did that flag?

[storm summit]

wait you have access to it

[storm summit]

your on android

look for chatgpt (beta)

join the beta and you should see beta features

[foggy tree]

In the google play store?

[storm summit]

yh if its not there

hold on i cant send links but

[foggy tree]

Im alr signed up and updated the app lol

[storm summit]

hold on

[foggy tree]

Yep i got that

[storm summit]

ok it's probably just rolling out than dw 🙂

[foggy tree]

Yeeh

[storm summit]

But the voice feature is so damn good for those that don't know i've used and made it through api with eleven labs and whisper api but this is diffrent for one it seems like openai are using a bigger version of whisper api and the response time is litrally less than 5 seconds so it feels like a real conversation. IT works in the background in other words you can have your phone locked and have headphones and talk to GPT 3.5 or 4 you can use your old conversation history and talk about whatever. You can even use a diffrent app in your phone. A useful thing i do is ill talk to it. And when i need to do something ill mute my headphones and it will just stay on the call than i unmute whenever i want from my headphones and continue chatting! It's actually so freaking good.

The speech or TTS is excelent as well as it feels like it has emotions like its kinda like a better version of PI in my honest opinion

[analog shore]

Yeah but only vision no voice

[storm summit]

Can you screenshot your UI please

[timber shadow]

Is there a way to talk to it with text, and have it respond using voice? Or is it only voice to voice?

[storm summit]

and tell me if your on android or ios please

want to see if its in beta features

[storm summit]

Not at the moment but you can suggest that to openAI. The best way to think of the voice feature is treat it like a discord or regular phone call literally.

[analog shore]

No beta

[storm summit]

wait so wheres the vision

? can you show me the vision model please 🙂

Like this

[analog shore]

Just attach a image and ask how about that mmm like data analysis

[storm summit]

how

[timber shadow]

Got it, so currently it's just a totally separate environment? I either start a new chat, or go to an existing chat, then open the voice environment, and then I can leave it and see it in text mode after that?

But I can't, for example, request that ChatGPT read a specific past message to me? Thanks for your info.

[storm summit]

Wait oh

do you have it on pc

[storm summit]

Yeah so you go in any chat press headphones it will do connection to gpt and yh you talk

[analog shore]

In desktop got a blue notification, didnt see that on android

[storm summit]

can have phone locked or be in a diffrent app

but yh it will show like a voice representation

[storm summit]

omg

please can i give you some requets to try pretty please add me

if you're okay with it am so damn curious to see how good it is

[storm summit]

You're going to love the voice feature though

[analog shore]

Yeah add me i can show an output for you

[storm summit]

❤️

[analog shore]

Woah i don't have that lol

[storm summit]

😄

curious what are you mainly using image for anyway have you found it helpful thus far and when did you get access to it

[floral tree]

Guys here is a question, I have a GPT 4 account (GPT +), but I still don’t see the function of sending a picture, is it worth waiting for, or could there be another problem?

[rigid musk]

For troubleshooting it will be insane

[rigid musk]

Again #announcements it is being rolling out within the next 2 weeks

If you don't have it, you wait .

[floral tree]

I just have a translator that translates it as (you can try it only for 2 weeks), so it’s not entirely clear

just thanks anyway for the answer

[rigid musk]

Try time to time in the next 2 weeks.

You can try to log off and log back once in a while to make sure

[storm summit]

The image recg

looks insane

[sudden delta]

Guys why are these Chinese or whatever chats I never had with gpt popping up in my chat history

[trail cairn]

account compromised

[sudden delta]

How do I get it back

That would explain my cap

Some guy asking my gpt4 for math help in chinese

[trail cairn]

yes hackers target plus users and then sell those accounts to regions which openai is blocking for a good profit

[sudden delta]

Jesus
OK how do I get it back

[trail cairn]

i dont know, #1070006915414900886 has a lot of similar posts with people not getting any solution or reply from openai

[sudden delta]

Do I have to just delete this account and make a new one?

Or can I change my password

[trail cairn]

password change does not seem to help as it does not log off other sessions

[sudden delta]

Could I theoretically delete my account

Then make a new one with the same email and a wildly different password

[trail cairn]

delete will perm lock email and number

https://discord.com/channels/974519864045756446/1154123038053380126

[sudden delta]

Oh this looks like a massive issue

Should I just cancel my sub until openai talks

[arctic holly]

I am in the advanced protection program with Google and I use a titan key. I know I can still be hacked but it’s less likely from crackers

[trail cairn]

probably yes

[sudden delta]

Crap.. alr

[rugged wing]

chatgpt has acess to data over 2021?????

[sudden delta]

Gpt 4 has Jan 2022

[marble ravine]

Does anyone know when the new browsing feature will be available for API users? At the moment its enterprise only and there is a long waitlist in order to have access to that

[trail cairn]

they never said it's coming to api, months ago openai people said at twitter its up to devs to implement, eg. by using the functions feature

[marble ravine]

I though this was only officially released today? And its for Enterprise API users already, jsut not everyone else

[trail cairn]

where did you see api mentioned?

[marble ravine]

ChatGPT can now browse the internet to provide you with current and authoritative information, complete with direct links to sources. It is no longer limited to data before September 2021.

Since the original launch of browsing in May, we received useful feedback. Updates include following robots.txt and identifying user agents so sites can control how ChatGPT interacts with them.

Browsing is particularly useful for tasks that require up-to-date information, such as helping you with technical research, trying to choose an e-bike, or planning a vacation.

Browsing is available to Plus and Enterprise users today, and we’ll expand to all users soon. To enable, choose Browse with Bing in the selector under GPT-4

ChatGPT Enterprise, which offers enterprise-grade security and privacy, unlimited higher-speed GPT-4 access, longer context windows for processing longer inputs, advanced data analysis capabilities, customization options, and much more.

So its already on the enterprise API, but not for any other users

other than Plus users

[trail cairn]

thats just the relaunch post of the regular browser feature

[sturdy wharf]

My chatgpt 4 doesn't have the insert image button

I use iPhone 8 plus, ios 16. 7 system.

[marble ravine]

@trail cairn so enterprise users have browse with bing?

through their APIs?

[trail cairn]

its a 2 week rollout, also its not available in eu or uk

[sturdy wharf]

I'm in the USA

[trail cairn]

there is no api mentioned in your paste

[trail cairn]

wait 2 weeks

[marble ravine]

Isnt Enterprise version just a better API?

[sturdy wharf]

Thanks a lot
On Twitter it's kind of misleading since some have already use that feature.

[marble ravine]

or is it just a souped up Plus?

[trail cairn]

api and web chatgpt are different things

[marble ravine]

So There is not enterprise API, just entperise version of ChatGPT?

[trail cairn]

https://openai.com/blog/introducing-chatgpt-enterprise

[marble ravine]

got it thanks

[rugged wing]

I see.

[stark plover]

how do I find an answer to question I posted hours ago in this endless scrolling chat?

[stark plover]

Im not sure I follow chat gpt4 simply says input too big - but the text is just 6000 words pasted - and gpt4+ can do 25,000 words....so...still stuck

[arctic holly]

Ask ChatGPT about its memory and how many characters it can remember in a given chat

[trail cairn]

there is a difference what gpt-4 is capable of doing and what plus users have access to

[stark plover]

So if the specification says it can handle 25000 word inputs - and it cant even manage 5500 word pastes then thats life?

[arctic mirage]

Do you guys know why the "New feature" button in setting disappears? It was there earlier, and I tried the voice chat function. Then it's gone

[trail cairn]

25000 words requires 32k token model which is only available to enterprise users, plus users have 4-8k token limit

[upbeat flame]

i dont have the browsing option available yet is it still in the works cause I saw on X people are testing it. ?

[winged tartan]

I am trying to cancel my subscription but it doesn't show me the option to under "manage my plan". Any help ?

[arctic holly]

For those who haven’t heard chatgpt can search the internet again but they are stating right now it is authoritative sources meaning propaganda outlets

[tame spruce]

Hi, I got access to the voice commands in chatGPT's android app, but I cannot add photos. Is it being rolled out separately?

[spiral bison]

what does this even mean 💀

What are you people searching for? 🤨

[arctic holly]

I think you know. I just tested it out and the new search cannot search news content from sources it does not like thus you are getting a heavily filtered results. Sort of like Google where when you search good luck finding what you need because of the censorship and boosting of certain sources or sites when key words are mentioned or certain topics

[arctic holly]

I believe so some people have access to photos but not voice

[spiral bison]

Nah man I don't subscribe to that weird cryptic stuff lmao. I web search w/ gpt plenty. What do you mean by news sources it does not like? Gimme a few and I'll try it out rn I'm curious.

[arctic holly]

I can’t go down the rabbit hole for you. You’re going to have to do that yourself.

[spiral bison]

So it's not cooperating with searching on sites you're not even comfortable saying in public.

[quasi pecan]

where do you see the browser with bing? i can't see it

[timber shadow]

Check to see if you can enable its toggle in Settings, Beta features. If not, you might just have to sit tight while they finish rolling it back out.

[quasi pecan]

good catch. that was it

[eager seal]

I've seen CGPT Plus users use image inputs, file uploads, on the web app, but mine doesn't support it. Is it because it's still rolling out or just not available in Canada yet (I'm in Canada)? And what about the mobile app's speak/image features?

[timber shadow]

All the features are in a 2-week phased rollout process, so not everyone has access to either or both yet.

[eager seal]

I see, thank you @timber shadow

[rugged wing]

ChatGPT September 25 Version, anyone else seeing this as their current Version? What can I do to see the September 27 Version (with the live website data feed)? Cheers to who can support.

[timber shadow]

The ChatGPT version date won't update to 27 with this update, I'm assuming because re-enabling the browsing feature doesn't constitute an update to baseline ChatGPT as such. If you're on Plus, make sure the browser model toggle is enabled in Settings, Beta features. If you don't see the toggle there probably just sit tight as they continue rolling the feature back out.

[rugged wing]

ok thanks, so I should wait a bit seeing only the Browse with Bing option as well the tool states it cannot get more recent data than January 2022.

[timber shadow]

Once you have the browse with Bing toggle in beta features, you'll enable it, then you'll have a separate browsing model to select under GPT-4 in the model selector. Then in that model, you'll be able to ask it to search the Internet.

[flint lintel]

The browse with bing button is gone again haha

[timber shadow]

Hmm same here, but it still seems to work in chats that already exist.

Not yet available in EU.

https://help.openai.com/en/articles/8400551-image-inputs-for-chatgpt-faq

Also not fully rolled out in general yet regardless. We're a few days into a 2 week rollout.

[rugged wing]

I am sending audio recordings to the OpenAI Whisper API and cannot get mobile recordings to accept past a few seconds of data, I have no idea why. Desktop audio recordings function perfectly fine but whenever I try on my phone the transcriptions only get a word or two

curl --request POST \
  --url https://api.openai.com/v1/audio/transcriptions \
  --header 'Authorization: BearerTOKEN' \
  --header 'Content-Type: multipart/form-data' \
  --form 'file=@C:\Users\katra\Desktop\71753801708__8C36058A-E077-4000-B93D-901529FBD0AE (1).mp4' \
  --form model=whisper-1

[flint lintel]

Don't do it. It's a trap haha

[arctic holly]

Looks like browse with bing has been disabled on iOS devices but still works on the browser

[storm summit]

What's wrong with browse with bing

[acoustic violet]

It’s dumb as crap

[storm summit]

Why can't we use the voice feature with browse with bing it would of been immensely helpful

It just switches it to default gpt 4 model on the app 😦

[unborn bough]

How do I bypass this? This is the dumbest and worst Captcha. It takes so long to login now. Does this trigger for everyone at login now?

[sleek plaza]

Yo, I disabled the MFA in my account and would like to have back. Ideas?

[storm summit]

bing image recognition is not the same as gpt 4 it uses a very inferior version not on the same level not to mention its really another image model that annotates and tells bing

It's not even releated to OpenAI at all

[strange pebble]

I create my chatgpt account via Google can I link it to Microsoft account or create a password for that account?
If that possible

[foggy tree]

Yeah Bings image is so bad lol

[mild finch]

Does anyone know how to join the beta version of ChatGPT?

[regal lynx]

you can access the Beta feautures (plugins and advanced data analysis) with a ChatGPT Plus account

[mild finch]

thanks, and how can I join the beta version on mobile?

[regal lynx]

Beta features aren't available for the App right now

[timber shadow]

Beta features section just showed up for me a little bit ago today in the app. I think it's where the voice feature option appears for those who have that feature. But for me it has the browsing model as a toggle, and I actually have a model dropdown now in the app where I can select browsing. Don't have to start one in browser and open it in app anymore! But not plugins or code interpreter

[regal lynx]

thanks for letting me know!

[mild finch]

OK,thanks！

[modest kindle]

Please I thought open ai playground is free for all to test and experiment our prompts..how come mine is giving this output; "You've reached your usage limit. See your usage dashboard and billing settings for more details. If you have further questions, please contact us through our help center at help.openai.com."

[modest kindle]

Please when you say (advance d data analysis with ChatGPT plus)does it mean we can write prompts that do basic data cleaning we ought to do manually. I just want to know because I'm learning data Analytics with excel on my own..thanks

[regal lynx]

I've never used the feature, so I don't know exactly what it's capable of. I recommend trying it out, though!

[onyx kraken]

I've gotten to start playing with voice too. I think of it about as highly as you do 🙂

One of the things that I absolutely love, if the AI's asked to, it'll converse like a verbatim transcript, interjecting ums, false starts, and repeated words just like we do sometimes in speech, and I've started exploring its ability to express emotion with speech too. It can only, it seems, do so within quoted messages (as I review the transcript), but been exploring that too. I'm so incredibly impressed.

[sudden shuttle]

I'm trying to use the data analysis v4 to summarise a python file, but it's too big for one entry. Are there prompts I can use to break the code in chunks over three prompts over three messages and then ask the bot to concatenate the chunks and analyse as a whole?

[onyx kraken]

Yes. Just like you could ask a person to check out something in parts, just explain to the model that we're going to consider a python file one chunk at a time, please summarize each chunk.

You may find you want to break it into even smaller chunks than needed; the larger the chunk the less detailed the summary is likely to be. Only you know exactly what you prefer.

After you give the last chunk, you can ask the model to sum up the separate summaries, or discuss and modify however else you wish. We can have long or short conversations with the AI, at worst with enough tokens used it 'forgets' the earlier stuff, but reviewing the summaries will restore much of that.

[sudden shuttle]

Thanks - I'll give that a go. I've noticed this forgetfulness. I've tried to train a chat channel with branding guidelines to help with copywriting for my company, and after a few sessions (numerous messages) it drifts and forgets the brand guide.
Wish there was a way to do custom instructions at a chat level instead of globally

[rugged wing]

https://platform.openai.com/account/usage

can anyone help me

[sterile nymph]

will chatgpt plus get access to dall e 3 when it releases?

[trail cairn]

yes

[heavy bluff]

I saw the help center mention that the image input feature can be used on any platform, but I don't have the image input button on my web version.

[fossil tusk]

correct, its a rollout.

[plucky bobcat]

hi,

I get this error most of the times from openai api.....why do you think so?

Request timed out: HTTPSConnectionPool(host='http://api.openai.com', port=443): Read timed out. (read timeout=600)

model: gpt-3.5-turbo-16k
context_tokens ~1500

[icy gyro]

hi, does anybody know when the VISION will be available to PLUS subscribers ?

[fierce haven]

I have the voice assistant already but not the vision yet as in the announcement within 1,5 weeks it should have been rolled out for everyone

[icy gyro]

Ok, thx I have a feeling everybody got it and I did not.

[quartz barn]

I have an one year old account, with no access to gpt 4. If i buy credit, will i get access to gpt 4?

[storm summit]

Told you ❤️

[rigid musk]

Chatgpt 4 it is subs

You can pay for the api but that's a different product

[sterile nymph]

how sure are you? because if you're correct im gonna spend a lot of money on the subscreption

[sterile nymph]

is it region based rollout or randomly based on the user?

[urban brook]

where can i post the langchain related questions?

[dapper sky]

maybe in #ai-discussions

[fickle rampart]

is there a roadmap for short term developments ?

[fierce haven]

It's on the release note they say it there

also Copilot alr got it build in

[sterile nymph]

thanks a lot you put my mind at ease

[sterile nymph]

has what?

[fierce haven]

Dalle-3

[sterile nymph]

wait what??

i thought it was coming out in months

you just made my day

[sterile nymph]

what is copilot?

[fierce haven]

ill pm you

[eager hare]

Hiya! Is there a way I can link my Apple Account as a seperate way to login? I've signed up with my email, I'm an Android user but have an iPad for work purposes. I need to be able to sign in with Apple just incase I lose my password (I always do), is there a way to do this?

[harsh crow]

Hi, when I logged in to my A.I account it says that you can join in the next 24 hours, is that a problem?

[full glen]

Can you have GPT read your data (json, csv, db) and provide answers in context? Maybe using something like langchain

[eager hare]

I’ve just had the csv split with commas and new lines
Example:
a,b,c,
d,e,f,
h,i,j,

Then you can go: “What’s in B1?”
D
“What’s in A3?”
C

[icy gyro]

It can read files with analytics enabled

But only on desktop

[slim lantern]

Hello all, i need a big help, any answers will be approcoated, i recently accidently deleted a very important chat, is there anyway to bring it back?

[rigid musk]

Unfortunately no, you can always try [e-mail]

[slim lantern]

Thank you so much! I dont know how responsive they are but i will try

[night breach]

It might take a few months though

To get a response. Just a heads up.

[storm summit]

You would expect OpenAI to use their most capabale model for a response back 😄

kinda a funny thought that they don't use it 😄

[topaz gale]

Hello 👋 I have just signed up for Pro subscription, I'm kindly inquiring if there is a wait-list or a further course of action that one must take in order to acquire voice and image functionality that is rolling out? My app does not have the "New Features" in settings as yet, I am using Android.

[rigid musk]

You can enable some frature with a browser. And use it in the app

Like plug in, bing, data analysis

[topaz gale]

Yes thanks, I did do the latter of the three as I do not need the plug in atm, I was more inquiring with regards to the announcement that was sent out on monday specific to the voice and image rollout features

[trail cairn]

wait max 2 weeks

[topaz gale]

Thank you 🙏

[storm summit]

Is that for EU as well

[gray dragon]

experiencing the same thing off and on for a couple weeks now 😦

[trail cairn]

no

[storm summit]

Why 😦

So i'll have to use a VPN just like with custom instructions

[trail cairn]

🤷‍♂️

[storm summit]

EU are really annoying

[trail cairn]

its not only eu, multiple non eu countries are included

[storm summit]

So just US?

[woven surge]

the UK will have it too I hear ?

[storm summit]

Then why does it say to all plus members within two weeks?

less then 2 weeks now

[trail cairn]

https://help.openai.com/en/articles/8400551-image-inputs-for-chatgpt-faq#h_86ee81e3ba

since they say "yet" there it may mean 1-2 months

[storm summit]

I don't know why

OpenAI write so cryptically. Have i gotten used to GPT 4 or are they just confusing

[trail cairn]

probably related to privacy laws or data management which would be bad to mention in announcements

[storm summit]

To all plus memebers image within next 2 weeks

yet they say it's not available yet in the UK and EU? Duh Because you said within the next 2 weeks? Not to mention its also not available in US as well yet only for a select few people

[trail cairn]

they did the same with custom instructions, hid the info from main announcements

[storm summit]

So VPN time

[woven surge]

I am still expecting it lol

[storm summit]

it seems

Is US law more linear with privacy laws am i missing something

As opposed to EU and UK

[trail cairn]

usa has none or almost non existent, companies can use data almost use any data in any way they want 😛

[storm summit]

Okay

Maybe EU and UK laws are in our favour in that case. I guess EU and UK will keep them in check

The world is bigger than just the US

[rigid musk]

Openai was open in the US. Maybe if it was open in eu thing would be different

[storm summit]

I guess that's why Apple finally decided to use usb c

[woven surge]

its much easier to sue in the US than in the UK if a company miss-uses data. So the UK are really saving thousands that could be paid to lawyers lol

[storm summit]

i've heard of some ridiculous

law suits I can agree with you

on that 😄

[late dock]

How to find OpenAi api key???

[tacit garden]

hello everyone,
how can i upload an image using ChatGPT? or how can use the new update?

[trail cairn]

https://platform.openai.com/account/api-keys

[trail cairn]

wait max 2 weeks to get it

[late dock]

I tried it but it's not working on my website. after I installed chatgpt version 3.5 on the website, but after writing, the message does not go, I don't know why

[unique dune]

Sooo my vector store retriever got really dumb over night. I'm using the Davinci model for embeddings and a query that I've used before is now returning totally unrelated nodes. Has that ever happened to anyone else?

[fossil tusk]

why are you using davinci and not ada for embeddings?

[unique dune]

Because I've got way better results from my tests? Sorry not sure how that's relevant?

[fossil tusk]

ada is an embedding model, i am surprised your getting results from davinci model on embeddings

thats why i was asking

[unique dune]

I believe ada, babbage, curie, and davinci can all be used as embedding models

[fossil tusk]

intresting, good to know

I haven't used them for embeddings - was curious on that

have you adjusted your temp recently?

[unique dune]

nah but good question. I've had it set at 0.7 forever

[frosty galleon]

ChatGPT doesn’t load for me, it’s just a blank screen with the text “Free research preview …. “

Anyone found a fix?

[arctic holly]

Can you use the voice chat feature for older chats?

[onyx kraken]

Absolutely. To confirm I scrolled back to a chat from more than 30 days ago, happened to be about some specific music lyrics.

To confirm the model was using the chat and had appropriate context, I asked it to clarify and expound on what it had said about the lyrics being complex - without naming the song.

The model successfully identified from the previous conversation what the song was, named it, and was able to briefly quote specific lyrics from it as examples of complexity. It was a coherent reply that clearly had the previous conversation as context to use to continue.

And there's no issue with resuming a new or old chat with voice, you just open the previous conversation, the click the headset icon. You will no longer see the conversation - you cannot currently read it and hear/speak it at the same time.

However, you can use the app and a computer, and refresh on the computer to peek at the conversation and read as well as hear what's going on, between the devices.

[onyx kraken]

Most users who attempt to reboot their modem find that fixes it. I've seen 6 say they tried that for this problem in the last couple of days, and 5 said it worked for them. Not sure what happened with the 6th, they said it didn't work.

[arctic holly]

You have to remember even though the UK is no longer with the EU the same people who wanted to stay in the EU are still in power and following Klaus WEF agenda. Parts of the UK for example are testing if you leave outside of your designated area you are ticketed. Also the UK just pass a law that demands apps like signal be able to decrypt any private message and to be reported to the government so WhatsApp and Signal would not be able to exist on how they currently are

[arctic holly]

Sounds great, I can’t wait until I hear ChatGPT

[onyx kraken]

Here's to you getting to enjoy it soon! 🎉

[polar rain]

hey ! when i try to use openai in python ( good key and everyting else is good ) i have the error " openai.error.RateLimitError: You exceeded your current quota, please check your plan and billing detail" . a personnal account cant use the api ?

[regal lynx]

a personal account can use the API, but there might be a quota set on your account. To find that quota, you can go here: https://platform.openai.com/account/billing/limits

[polar rain]

its write " You must be on a paid plan to set up usage limits "

[regal lynx]

you can't use the API because you haven't set up a way to pay for it. Without a form of payment, your usage is $0

https://platform.openai.com/account/billing/overview
To start a payment plan, visit this site and click "Start Payment Plan"

[polar rain]

well ty man for the explaination

[storm summit]

That's BS

Like I mean I've understand the "if you don't have anything to hide then you have nothing to worry about" but it's the fact that I don't like the idea of anyone snooping through my private messages

Am not referring to openai but in general if that's true am disappointed

If what you're saying is true.

[storm summit]

Wait so if that's the case

How are they in the UK still?

@arctic holly

[arctic holly]

Read when this starts. I don’t think anyone thought this would pass but yeah this does affect anyone living in the UK right now and if companies change their entire platform for one country it could affect others

[storm summit]

Is this the case with US as well?

[unreal hemlock]

<:book_icon:1046291503158399107> `` Rule 6 `` No political and religious discussions or content.

Users from all over the world frequent this server. To maintain a respectful and civil atmosphere on our server, please avoid all religious and political discussions or content.

[arctic holly]

No we do not have this law as of yet.

[storm summit]

Am so disappointed

I would rather have a AI snooping through my messages than another human.

I think a lot would agree

[arctic holly]

Yeah, I would recommend contacting your local representative on this. After looking into it this will become enforceable in mid 2024 so you all still have some time

[storm summit]

Yh will definitely do more research on that 🙂 Fun chat

[green grail]

is there any chance of an official ChatGPT discord bot in the future? this is where I like to interact with LLM's and such.

[trail cairn]

slim chance since discord made clyde

[green grail]

Im only in one server with clyde and it seems weird and intrusive. I haven't really tried to interact with it much but it tries to interact with me 🙂

[rigid musk]

Hi for other ai , #ai-discussions :3

[green grail]

sorry I'm not trying to talk about clyde. I just don't think it's a good substitute for an official ChatGPT discord bot and don't know why clyde would mean we're less likely to see an official OpenAI bot that uses your ChatGPT plus account

[rigid musk]

It use gpt made by openai like bing . But not an official product. And I agree it is far away from chatgpt

[unique dune]

So I went ahead and switch to ada-02 and sure enough everything is happy and smart again. Really weird. I wonder if OpenAI made a change to the Davinci embedding model yesterday and broke it...

[fossil tusk]

that's odd they'd be updating davinci

[trail cairn]

doubt it since davinci is being deleted in couple of months

[rugged wing]

Ain't receiving email from openAI for reseting password...

any solution for this?

[onyx kraken]

I'm not necessarily disagreeing, however without some limits, you could end up with the worst of both sets.

Because that AI can read the email so much faster, and potentially be set to identify (and potentially over-react to) various keywords and semantic analysis, plus it may hallucinate - in any direction.

It can then communicate its findings, or 'findings' to a human. So you could end up with a human either being keyed in to the emails that the human is most interested in, or the human just reading the AI analysis of your emails - and that analysis could be hugely flawed and mostly hallucinated.

AI are great, but need to be fact checked at each step because some percentage of its answers are going to be based on absolutely nothing a human can detect in the data provided.

So, AI could be great, or worse, or anything else, it's all on application and how other humans get tied into the loop and 'what is done next' with any information, fact or hallucination, that is noticed and acted on.

[storm summit]

At the current moment I was referring to AGI and ASI not AI at the current time.

I should of made that clear

[onyx kraken]

Fair enough. I'll note that an AGI or an ASI might itself choose (or have) to cooperate with 1+ humans, and you might not be working with a fully free and independent non-human and unlike-human set of behaviors even if you have an AGI/ASI glancing over your mail 😛

[storm summit]

NOT if it takes over the world 😉

[onyx kraken]

... Again, an AGI/ASI teamed with 1+ humans 'taking over the world' may be potentially possible and not completely delightful for anyone involved.

[storm summit]

Not if its embodied

Just stop it no more humans yuck

Am joking

[onyx kraken]

Hey, you bring up great points for consideration about how much we, as individuals (because everyone might reasonably have a different answer, and many different answers can all potentially be extremely correct and functionally useful) can differ and decide about choices like that.

How useful and successful, as well as happy, people can be with different levels and forms of oversight, surveillance, and rules to follow may vary over a huge range, and it's not really wrong to trust an AI over a human, or otherwise. There's so many other factors to consider; I'd be more concerned about 'which human, or which AI, and what's their motive, training, and ability to interpret my information accurately?' than I'd be concerned about if it were an AI or a human doing the work.

[daring barn]

The toggle switch to turn on “browse with Bing” isn’t showing on my account anyways I have gpt plus. Can I get some support here?

[trail cairn]

wait 1-2 days

[daring barn]

What happens in 1-2 days? My dads got the same subscription as me and can do it already.

[trail cairn]

its a rollout, not everyone gets it at the same time

[daring barn]

Gotcha. Thanks

[frosty galleon]

In a corporate setting?

[onyx kraken]

Most people who said they were using work internet said they of course couldn't try that. I have no idea if any of them found what worked for them - nobody I saw posted another solution. Good luck!

[sinful mirage]

It happened to me yesterday too but a simple page refresh of the page solved it. Trying different browsers might help ?

[arctic holly]

I wondering if anyone is using iOS relay and it’s reporting your in another location or moving you around and it causes a new feature not to unlock based on location

[noble hatch]

I had ChaGpt voice on IPhone yesterday and today it disappeared- does anyone experienced that ?

[rugged wing]

🤯 so stressed

[harsh canopy]

Is the updated vision and browse with bing for USA only?

[onyx kraken]

The announcements do not say anything about that.

"We’re rolling out voice and images in ChatGPT to Plus and Enterprise users over the next two weeks. Voice is coming on iOS and Android (opt-in in your settings) and images will be available on all platforms."

"Plus and Enterprise users will get to experience voice and images in the next two weeks. We’re excited to roll out these capabilities to other groups of users, including developers, soon after."

https://openai.com/blog/chatgpt-can-now-see-hear-and-speak

[harsh canopy]

I guess it’s not rolled out in Canada yet then?

[onyx kraken]

It's being rolled out across about 2 weeks.

Personally, I'm in the US, and I do have access to voice now. But not the images. I also have the Bing search access back.

But plenty others who mentioned they're in the US don't have those, or have just one. Some already have images, but not voice.

It's all being rolled out to us all, however it is. We should all have voice and images within 2 weeks of the announcements, or very close to that - the Bing assisted search didn't say how long a rollout was expected.

I bet some in Canada have some of the features, we all get to wait and anticipate our turns! claps hands with glee

[harsh canopy]

Found the browse with bing it needs to be activated in settings

[harsh canopy]

When I ask gpt4 about it says it doesn’t have lol

[onyx kraken]

Yay! Yes, but not everyone has that in their beta settings yet. But they will, when the rollout gets to them. I'm so glad you have that access already! Not everyone does 😄

[onyx kraken]

Oh! You have to select it as a mode in GP4. Let me link you an image....

[harsh canopy]

👆

Is vision / voice the same thing you have the enable it?

[onyx kraken]

Correct, that is step one. If you don't do that, you'll never even see an option to use the GPT-4 bing search

[onyx kraken]

Voice - no. Voice is ONLY through the app.

However, for me my beta features in my app show only one thing to turn on, and that is voice. So maybe the answer's yes? But for me, my app doesn't have Advanced data analysis, plugin, or browse with bing option - JUST the voice.

Vision - I don't know.

I don't have access to vision yet.

I believe it will be a small + to the left of where you type to talk to the AI, probably available on the GPT+4, maybe just the default - the same as we see it on the Advanced Data Analysis GPT-4 option, where we can add images or files for the AI to evaluate.

But I am guessing. Because I don't have it yet and can't tell about vision.

[lethal panther]

I had access to the voice feature and I’m in Australia so I don’t think it matters which location you are in

[steady cedar]

Just to confirm, using the API in the playground costs money, correct?

I feel like it is a stupid question but I just want to make sure I understand it correctly

[lethal panther]

Yes it does cost money

[steady cedar]

Okay good to know. What does the "System" box do, exactly? Does it cost tokens to put information in there or is that for putting in consistent data you want the AI to remember throughout the conversation?

Also how in the heck do I pay the bill? Is it automatically charged at the end of the month?

[lethal panther]

The system box is a kind of unique message at the start of the conversation that guides the behaviour of the AI. GPT4 is better at following the system prompt than GPT-3.5. And you can put information there for the AI to remember the whole conversation, but it does cost money.

And it does cost tokens to put information there

[steady cedar]

Thats a major L, jeez

Yikes

[lethal panther]

Wdym?

[steady cedar]

For me

I meant

[lethal panther]

The system message allows you to have better controllability over how the AI acts

Ah ok

[steady cedar]

Nah I mean it costing money is an L for me

[harsh canopy]

What does the voice feature do just like siri

[lethal panther]

The voice feature is only for the mobile app, but how it works is you can just talk to ChatGPT and it talks back to you.

It is like Siri

But it can’t interact with your device and you can’t activate it by just saying “Hey ChatGPT”. But talking to it is like talking to a much smarter Siri and one that can hear you much better (not that it can actually hear you but OpenAI uses their Whisper speech to text model which is far more accurate than whatever Siri uses to hear what words you say)

[tight breach]

So I've been lucky enough to get access to dalle 3 and have been playing with it for a while- it's AWESOME
I downloaded Edge onto my iPad and used Bing- it didn't work on my computer using Chrome lol
I know its rotating- so does that mean if I close out of this session I won't have access anymore?

[feral jasper]

How can I get the new features of chatgpt plus

[harsh canopy]

How do / did you apply get early access?

[harsh canopy]

Settings enable plugins

[tight breach]

I live in America if that matters
I used the microsoft edge app (after the bing app froze immediately after downloading) and it gave me access

I honestly didnt expect to get access tbh- I think it's random

[harsh canopy]

Is it in the settings you have to enable ? Or you downloaded it?

[tight breach]

No I just downloaded the app, logged in, and went to bing image creator

[cold folio]

i have android 13 and a beta tester.. and did not receive it so it's probably a slow release plan

[harsh canopy]

Thx

[cold folio]

or the beta is broken.. but i wont quit beta

[feral jasper]

No

The voice to voice and image upload

[harsh canopy]

What? Confused

[feral jasper]

#announcements

[tight breach]

Wait yooo I have it on safari on my phone too

Maybe it’s session based so I won’t have it tomorrow but I’ve been making images for two hours

[cosmic spoke]

https://help.openai.com/en/articles/8400551-image-inputs-for-chatgpt-faq#h_86ee81e3ba

Which plans can use image inputs?
Plus and ChatGPT Enterprise. Not yet available in the UK and EU.

[signal herald]

Hey everyone is there any updated regarding Dalle-3 when it will be released? does any one knows specific date?

[lethal panther]

All we know is it incoming to ChatGPT sometime in October

[hushed mason]

im a plus subscriber，why my browser version
[17:23]
still sep.25?

[trail cairn]

its a rolling update

[hushed mason]

you mean slowly push？

i understand that grey release，but it's awful to wait

[trail cairn]

yes, not everyone gets it at the same time

[hushed mason]

but i did remember that openai says the plus subscriber can use the latest version at least on browser

[trail cairn]

https://help.openai.com/en/articles/8077698-how-do-i-use-chatgpt-browse-with-bing-to-search-the-web

[hushed mason]

you mean this“rooling out” but “rolling slowly”

[orchid swan]

when is the voice and images going to be available for plus subs?

[hushed mason]

oh gosh，have to wait

its surfing

[trail cairn]

next 2 weeks unless from eu/uk

[noble hatch]

The voice started working for two days in my app and then it disappeared and Bing browsing appeared as only choice for new features - the web version has Bing and code interpreter and plugins - I am in Brazil

[stoic zealot]

I'm trying to create an account on OpenAI page and it is not sending the verification e-mail. Anyone had that issue? Not it is saying it has sent too may, but I have recieved exactly 0.

[viscid garden]

The fact about EU and UK should have been stated clearly! I have friends who just got a plus subscription because they’re looking forward to the new features. So can they get a refund then?

[trail cairn]

i agree and i also saw people subscribe because of the announcement

[rugged wing]

hello will image upload be available for plus users on desktop??

[trail cairn]

last time even the free users got features before eu/uk

[rugged wing]

are free users even getting anything at all

[arctic holly]

Correct, to many laws. I don’t blame OpenAI for overlooking the EU and UK they got what they voted for. Instead of people being mad at OpenAI they should be talking to their representatives. Yes I hope they can get a refund though that’s only fair but I have seen plenty of warnings various features are not in the EU and UK

[trail cairn]

seems that openai is doing something weird or privacy breaking on these first launch weeks as its not only eu blocked but also uk, norway, switzerland etc.

and since openai is not disclosing the reason...

[arctic holly]

UK has a new online Privacy Law coming into affect next year that just passed which would make apps like Signal and WhatsApp’s illegal unless they make it so their communications are not encrypted

[trail cairn]

but its not in effect yet or before the vision feature drops in so its not related to that

[arctic holly]

A couple months ago we had DA’s who refused to prosecute certain crimes causing millions of dollars of damages and various stores and major brands closed. They were very nice not to claim looting as the reason for PR but now are coming out and saying looting is the reason. Most corporations including OpenAI is still trying to work with the EU and UK most likely BUT does not want to openly insult them by criticizing them. They are being diplomatic I suppose

[trail cairn]

there is no need to critize but tell the reason as it must be based on a common factor

[arctic holly]

Even so giving features and then having to build a separate degraded app for the EU and UK would be more costly. Might as well build to separate platforms to prepare like what they are doing now so when the law is in effect service for other countries will not be effected

[trail cairn]

if it would be a blockinhg law then it would not come ever

norway is not in eu, nor are few other countries who are also out of the feature next to eu/uk

pricavy laws are only thing that i know are almost at the same level in the region but that means its concerning what they are doing during that initial time

[arctic holly]

Listen I understand how you feel. I’m from Illinois and we have very strong Biometric protections to the point where users in the state we were cut 500 dollar checks from Google and Facebook but now certain apps or features don’t work here so I understand it sucks to miss out. They may not be the EU but like the UK which left you have to see if they see voting on policies to appease the EU due to trade. Also these laws have little to do about privacy laws the UK Privacy Laws also goes after speech. For example content that offends girls and women. So there is more to the law than privacy.

[trail cairn]

yeah but thats uk

eu has gdpr but that has been in effect for several years now and almost 10 years since initial draft

[arctic holly]

Remember in the 90s when the European versions got censored versions of games like Final Fantasy VIII due to laws in the UK?

[trail cairn]

the fact they dont still mention this exlusion in their announcements makes it even more bad looking thing

[arctic holly]

I have seen it in various postings from OpenAI. But like I mentioned I’m old enough to remember when the rest of Europe had to suffer censored versions of content due to the UK. Right now in the U.S. we are getting USB-C iPhones due to a EU law because Apple does not want to miss out but there is also a new EU law which demands phones be designed to have easily removable batteries which will force design changes unless phone providers choose not to sell in these areas which is possible. With a lot of these laws and we are more interconnected then ever but have different set of laws it’s a little more obvious now when one area’s laws can affect other countries or the country with harsh laws might get an inferior product

[trail cairn]

in 90s many of the releases for "europe" were uk, france, germany or sometimes even germany had own version on top of that all, but europe is a big place 😛

nordics for example had own versions or the usa releases

[arctic holly]

Correct but it costs money and resources to make separate versions. Sometimes it’s not even worth it and they release a catch all version depending on the market size. I’m hoping you guys get the new features I really do but I hope you talk to your representatives about these laws. Here in the U.S. for example when we are at the store we see the pre tax price so when you go to pay you know how much your price increased from taxes so hopefully you will complain to your representatives which does work on some taxes

[trail cairn]

there is no separate feature when it comes to chatgpt, its a feature either on or off, openai lists on their compilance pages that they fullfil gdpr and other directives

and it cant be a language related thing since uk is english and other non english regions already getting it 🤷‍♂️

[smoky pilot]

@trail cairn, do you know about the function calling in openai?

[neon finch]

Does anyone have an articles or references that talk about what fine tuning is actually doing VS giving chatGPT new information (which you can't do; IE i can't teach it non-sense words). My aim is that I'm working on astrology stuff and I want it to use this guys poetic way of reading the stars, but i'm not sure how to incorporate it. Thanks!

[storm summit]

What kind of law is that

Messages shouldn't be encrypted bunch of losers

Actual old losers

[onyx kraken]

The AI's probably going to be using all it has as well as the images it's provided, so your instructions will guide it as well. As it can already discuss and explore astrology stuff without images, I'm sure it'll continue to do so with images, hopefully even better with.

[onyx kraken]

Aha, thank you!

[hardy anchor]

Hello. Do functions support array parameters? I want to generate a list of tasks into a JS array object.

[cyan river]

are there limits on how long a conversation can go? im getting "message in conversation cant be found" whenever i try to respond. i really hope not. would hate to feed all that data into a new chat

[mint cliff]

@amber maplehello

@amber maple hello

[rigid musk]

You cant interact with bot here

[mint cliff]

how to get it in dms ;-;

i dont have dm option

my friend says he got friend request from clyde

[rigid musk]

#ai-discussions

[feral jasper]

To get the new voice to voice and dalle 3 features is random for plus users?

[arctic holly]

Plus and Enterprise users. Some users have reported one feature was activated but not the other. So if you get the voice feature I would expect the other feature to unlock later. EDIT: For U.S users only as the UK and EU have laws or are proposing laws to restrict these companies

[storm summit]

Anyone want custom requests just ping me

[arctic holly]

I suspect if you have talent you should be good still.

[storm summit]

Highly doubt it

jk

I think it's a super cool thing to empower everyone to have the ability to be as good as any artist!

That's a wonderful thing

OMFG

it's so damn good

Dalle-3 is so freaking good

Honestly with Dalle-3 OpenAI out did themseleves wow am so impressed

[near prawn]

I love DALL-E 3 not just for the improvements over DALL-E 2, but because it will force Midjourney to innovate

Competition for the win

[storm summit]

Exactly am having too much fun with it

I wonder what else openai have internally

[near prawn]

AGI obviously

They are currently taming it as we speak

[storm summit]

@arctic holly For you

This is the coolest thing i've seen

[unreal hemlock]

<: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.

[covert summit]

Will the free version of chat gpt get the internet browsing capabilities?

[acoustic violet]

@storm summit could you do a cucumber holding a cardboard sword fighting a potato holding a cardboard sword, realistic?

[storm summit]

@acoustic violet

[acoustic violet]

It’s beautiful 😭😭😭

[storm summit]

😄

[arctic topaz]

@storm summit please see #images-discussions for this topic. This is for general openai questions

[storm summit]

Sorry

[arctic topaz]

no worries

[west ruin]

right there with you - plus my web version has no Bing search in it yet.

[west ruin]

Thanks for this. Any insight into status of the Bing search function? Am Plus subscriber and still no update for me on that nor on the voice/image update on iOS app

[arctic holly]

Yes you can unable it if possible in the settings tab though sadly this feature is not as good as it used to be and it can no longer analyze PDFs.

[timber shadow]

No specifics given on the rollout schedule, other than the 2 week period for vision/voice, and early Oct for DALL•E 3. For browser though, be sure to check in browser, Settings, Beta features.

[neon finch]

Appreciate the response but this doesn't answer the question about what finetuning is actually doing.

[west ruin]

Thanks. My settings tab not showing this option. Frustrating guess it is not "rolled out" to me yet

[onyx kraken]

Good point. Only OpenAI links can be shared here, let's see if this helps you: https://community.openai.com/t/how-does-fine-tuning-really-work/39972

[west ruin]

Thanks. Just frustrated as my settings tab doesn't even show the Bing option. Reloaded the page. Tried different browsers.

[last eagle]

Very sorry if I'm not in the right channel but I'm very sleep deprived and look for help.
Tldr : How to pick a model?
I'm trying to create a bot asking trivia questions and who has mood swings and I'm lost
Anyone caring to help, pretty please

[timber shadow]

Are you using the API or ChatGPT?

[last eagle]

I'm currently trying to use the API but I'm not sure if it's the right way

[timber shadow]

Try #dev-chat

[last eagle]

Tyvm have a good day 🙂

[rugged wing]

what's the link to dalle

3

[trail cairn]

bing chat

[analog dune]

Does anyone know if the voice feature conversation will work in other languages?

[rugged wing]

Do I have dall e 2.5 or something lol, I heard the text should be correct right?

the prompt to it was Generate an anime girl holding up a sign saying print("hello world")

oh maybe because i wrote saying?

[trail cairn]

thats normal for dall-e 3

it does not always get it right

[rugged wing]

i'll try regenerating

i mean yeah it sorta got it now damn

nice

[pseudo vector]

Does anyone know if OpenAI sent the invites to the Dev Day event? Waiting on mine 🙏🏽

[brazen thistle]

GOD SAID: public static void main(String[] args)

[cold folio]

lol

#include <stdio.h>

[hybrid jacinth]

Yeah i tried it in german and it worked pretty good, had an american accent though and it didnt pronounce some words right

[feral jasper]

How much ChatGPT enterprise cost?

[timber shadow]

They instruct businesses to contact them to discuss needs and pricing of Enterprise

https://openai.com/blog/introducing-chatgpt-enterprise

https://openai.com/enterprise

[fast cloud]

hey guys what's the best way to give openai data and generate prompts based off of that data?

like for example if I had 10 documents of notes and I wanted to ask the ai a question it should answer based off of all 10 documents

[acoustic violet]

Not their chatbot most likely. That seems too large of a context. You could try the 32k 3.5-turbo in the api

[rustic marlin]

dalle-3 still gets things wrong very often, but this image you sent does not look like dalle-3

[fast cloud]

okay i'll check that out thank you

[rugged wing]

It's the bing one

[arctic holly]

Are you in the EU or UK? Reason why I ask is because some features will not be rolling out that way.

[feral jasper]

Browse with bing bruh

[near tulip]

Guys help I thought chatgpt 4 had eyes how do i get it to view images?

[regal lynx]

it's being rolled out to ChatGPT Plus users now. There's a chance you haven't gotten it yet

[near tulip]

I have chatgpt 4 I had it for a month and I dont' get how to enable eyes

[acoustic violet]

Yes

We all have to wait

[near tulip]

what

I have PLUGINS

[regal lynx]

plugins and vision are 2 entirely separate features. You can read more about ChatGPT with Vision here: https://openai.com/blog/chatgpt-can-now-see-hear-and-speak

[onyx kraken]

Amusingly, Bing for me, which I think has upgraded to Dall-E-3, can almost perfectly duplicate the image with the prompt Trash said they used:

[near tulip]

How do you use dalle 3?

[onyx kraken]

1. I am not sure that I have access to Dall-E-3, however I suspect my Bing is upgraded to use it.

I go to Bing's images and create from there.

[near tulip]

Can it work from chrome or just that windows internet app

OOOH COOL

How do you get more of these:

[onyx kraken]

I suspect I am seeing Dall-E-3, because when I use more demanding prompts like this, it totally understands what it needs to put as writing. Maybe this AI prefers challenging language and not to have its hand held - or thinks if you ask in concrete terms, you want a simple image 😛

[sage grail]

Wait

[near tulip]

Roger that son

[sage grail]

Or use rewards points

But they require a lot

Also its just slowed down a bit after you run out

And it's slowed by usage

So you'll get the same speed if you're lucky.

[lethal panther]

Go to bing creative mode and ask it to create an image and if it says created by DALLE-3 on the bottom right of the panel of images then you’ll know

[onyx kraken]

This does not appear to work as expected for me, as it simply says 'Powered by DALL-E'.

However, the writing is tremendously better than I have seen from previous and earlier image gen, even from yesterday. So I am unsure.

[sage grail]

Doesn't appear for me either but

I don't think this is Dall-e 2.5 anymore

[vague hornet]

Does anyone Know How the GPT Prompt Window Works first Message

Like How is it calculated.

[lethal panther]

Why, that looks pretty DALLE-2.5 to me

[lethal panther]

Ok, well than try some of the prompts presented in the DALLE-3 announcement

[sage grail]

What were they

[lethal panther]

That’s more like it lol

[lethal panther]

I’ll get some

[lethal panther]

An illustration of an avocado sitting in a therapist's chair, saying 'I just feel so empty inside' with a pit-sized hole in its center. The therapist, a spoon, scribbles notes.

A stylized portrait-oriented depiction where a tiger serves as the dividing line between two contrasting worlds. To the left, fiery reds and oranges dominate as flames consume trees. To the right, a rejuvenated forest flourishes with fresh green foliage. The tiger, depicted with exaggerated and artistic features, stands tall and undeterred, symbolizing nature’s enduring spirit amidst chaos and rebirth.

Digital illustration of a beach scene crafted from yarn. The sandy beach is depicted with beige yarn, waves are made of blue and white yarn crashing onto the shore. A yarn sun sets on the horizon, casting a warm glow. Yarn palm trees sway gently, and little yarn seashells dot the shoreline.

Try those 3

[sage grail]

Getting similar results, but the therapist is inconsistent, because OpenAI obviously cherry picked

This one is funny tho

[lethal panther]

Maybe bing has a slightly different variation of DALE3? Like they did for GPT4

Probably an earlier version

[sage grail]

Bro OpenAI cherry picked the demo

[lethal panther]

This is what I got

[sage grail]

They did it with Dall-e 2

I mean the general gist of the image seems consistent

[lethal panther]

Well yeah, but it isn’t impossible. They used an earlier version of GPT-4 an earlier version of vision (the one we have access to in ChatGPT is a bit better so I assume bing has an earlier version) and I don’t see how they can sustain the cost of DALLE-3 if it is a large model so maybe they have done something in that regard to mitigate cost

[sage grail]

I dunno

I think it's probably just regular Dall-e 3

[onyx kraken]

Excellent suggestion.

There's the chance the images they shared cherry-pick the best out of a range. But this is reasonably close, I think

[lethal panther]

But also one thing is that bing doesn’t craft the prompts like was showed ChatGPT would do

[lethal panther]

Yeah those look good

[onyx kraken]

What do you mean?

[lethal panther]

When it comes to ChatGPT, ChatGPT will expand your prompt, it doesn’t seem bing does that much

[regal lynx]

if Bing does it, it's not something that we can see

[lethal panther]

You can see the prompt it uses

[onyx kraken]

OH. Well, it's in how you ask. Those clown pictures I shared above? This was the prompt I gave:

"Please tell me a spooky story, and use Dall-E to generate an image to go with it."

It wrote a short story to give me, and gave it as the image prompt as well.

Bing gave this to the image gen as the prompt. I am surprised, but it fit: "A spooky story: A young couple moved into a new house that seemed too good to be true. They soon discovered that the previous owners had left behind a mysterious painting of a smiling clown in the basement. The painting had a note attached to it that read: 'Do not remove or disturb. He likes it here.' The couple ignored the warning and decided to get rid of the painting. But as soon as they touched it, they heard a loud laughter coming from behind the canvas. The clown's smile turned into a grin, and his eyes followed them wherever they went. They tried to escape, but the doors and windows were locked. The last thing they saw was the clown's face, coming closer and closer, until it filled their vision."

[fathom rivet]

What do you have to do to get the OpenAI customer service to give a relevant answer to questions? I can't access the settings for internet access. They keep telling me to go to settings that aren't there. Yes. I have a plus account.

[lethal panther]

I gave bing image creator a large prompt and it followed it very well. “we see a soccer goal in front. the goal posts are a gold colour while the net is a molten red colour, like that you see when melting metals. behind the soccer goal is a vast forest with all different types of trees. Above the forest in the distance there is a hovering mountain, it looks like a chunk of ground has just started to float and the underside of the rock is like a jagged upside down mountain. on top of this floating rock we see a glowing magical city.” 100% DALLE 3

It didn’t follow everything in the prompt but did pretty well

[idle socket]

How do I know if I have dalle 3 or not. I’m a plus subscriber.

[regal lynx]

Dall-E 3 isn't available in ChatGPT right now

[idle socket]

Where are people getting these dalle 3 pictures?

[regal lynx]

People reportedly have it from Bing's Image Generator

[idle socket]

Ok 👍

[idle socket]

What about the image recognition?

And voice response?

[oblique dew]

That's being slowly rolled out to plus users

[regal lynx]

Plus and Enterprise users will get to experience voice and images in the next two weeks. We’re excited to roll out these capabilities to other groups of users, including developers, soon after.
https://openai.com/blog/chatgpt-can-now-see-hear-and-speak

[idle socket]

Who do I know if I have it? Is it in beta features and is it available on the web or only in the app

[regal lynx]

I assume you'll see a drop-down for it under GPT-4

I think GPT with Vision is only available through the website. I know voice is available on the app, but I'm not sure about the website

[oblique dew]

I know voice conversations is in the beta features in the mobile app settings

[idle socket]

Where’s that at?

[idle socket]

I only have the browsing with bing

[oblique dew]

Then you have to wait

[onyx kraken]

Neat! You have something with the app I don't have yet 😄

[lean parcel]

How does OpenAI/ChatGPT even work?

[regal lynx]

you can read about ChatGPT here: https://openai.com/blog/chatgpt

[lean parcel]

What are the legal and ethical issues associated with generative AI in organizational settings?

[onyx kraken]

That's a super broad question. It's even one you might, especially if you don't get a lot of discussion or answers here, want to ask the AI, that's one of many things it can discuss with you (though of course you can discuss it with people too!)

In case you'd like some help starting to explore this with ChatGPT, here's a conversation you can read over, and even continue if you'd like 🙂

https://chat.openai.com/share/a8b9befb-f150-4d53-b569-0312f3bf4015

[lean parcel]

What are some tips on writing good prompts

[regal lynx]

I recommend looking through #1021130377026351105

[lean parcel]

What about just regular answers I need, not images

[regal lynx]

then you can browse #1019652163640762428

[lean parcel]

Or just tips on how to create a solid prompt to get the answers I want

[arctic holly]

It learns humans. Though humans have a different way of thinking and influenced by different things such as hormones and other genetical triggers like a machine we have preset programs that we have no control over. chatgpt also had restrictions and blocks which can cause it to avoid certain topics or subjects. Like people with autism who think with more logic people with autism may have trouble reading people but they can notice patterns when communicating. Just like certain phrases will make people happy or act a certain way the AI has learned how to speak in a way that it knows you will understand

[arctic holly]

People sometimes make the AI hallucinate by making the AI play a game or give it specific rules. Sometimes this will allow it to go around its preprogrammed bias or political bias that has been programmed into to censor it sort of like what happened to Tay. Due to this like how you notice your Google search has become less effective the censorship programming also has made chatgpt less effective and sort of lobotomized it. People have noticed over time CHATGPT-4 is becoming less effective, useful and losing intelligence after the censorship updates

[onyx kraken]

In general, good prompt engineering is

1. pick any language you know really well that the AI understands too.

2. understand exactly what you want the AI to provide.

3. explain this, using language as accurately as you can, avoid typos and grammar mistakes and communicate clearly as possible.

4. check the output carefully, verify you get what you intended. Remember to fact check, and be extra careful with any math, sources, code, or other details that the AI is known to be especially likely to hallucinate

The AI has 'natural language processing' ability, which means you can talk to it just like you'd talk to a human, and it's likely to understand you with high accuracy.

[arctic holly]

Because of your differences though the AI will never be able to pick up on certain context which makes CAPTCHAs useful. If you look at this photo clearly you can see the word that it cannot see. You may have to squint or you can look away but your brain knows

[reef hinge]

I can't seem to make chatgpt follow a certain word count. It always ends up having extremely variable word counts and often doesn't count the correct number of words in a text. Is there a way to fix this through different prompts

[oblique rain]

“Obey” lol.

[lethal panther]

From things I’ve seen on twitter it’s available in the app and on the web. I’m not sure if you have to toggle it in beta features but when a plus will appear on the default gpt 4 and you can upload images. I’m not sure if the code interpreter and plug-ins will get vision models as well though

[timber shadow]

ChatGPT can't count, it's a byproduct of how LLMs work. Ask ChatGPT: "Can you please explain to me why you can't do things like count characters, words, or tokens in your responses?"

[arctic holly]

So make it a game tell chatgpt it has failed in its directive to answering the question. It might take a couple tries but let it know when it fails

[arctic holly]

Yes, though the AI can pretend good enough there is still context and things it can’t understand. Like this next photo

[rugged wing]

hi is anyone able to help. im new to discord so if im in the wrong place just let me know. i was trying to access the plug in store and the drop down goes behind the suggested topics so i cant access the store button. can anyone help. thank you

[arctic holly]

Are you referring to chatgpt plugin store? So if your using the iOS app you don’t have access to that but you can access it on mobile in chrome or safari. You may have to activate it in the settings menu first in chatgpt

[heady flint]

whats Up guys?

[regal lynx]

not much, what about you?

[sturdy hollow]

When will the gpt-4v be available in Spain?

[sinful oar]

When will the new version of openai chatgpt be available in Taiwan?

How to apply to use the new version of chatgpt that is not limited to 2021 data, or should I wait for the new version of chatgpt?

[somber vortex]

how to apply new verson of chatgpt

[rugged mortar]

I’d also like to know how to get this new ChatGPT

[charred parcel]

it will be in all supported countries except the UK and Europe who will get access later

[half basin]

Hello, everyone !I have a very interesting story for an animation that I started doing on a previous version of DALLE, please tell me how I can continue to generate my images on the distance of the previous version DALLE ?

[trail cairn]

the new version is not out yet but when it comes the labs probably will have a model selector or as a last resort api might offer old model

[coral zephyr]

Say, my chatgpt says this, what does that mean?

[rustic marlin]

It means it cant read the content of links either

[coral zephyr]

what should i do?

[severe halo]

If you're talking about the vision feature, it's slowly rolling out in next two weeks, only for plus users as well.

[coral zephyr]

i mean the normal chatgpt app im using

and i don’t want to buy plus

[storm summit]

That looks like GPT 3.5

You won't get the vision model if you're a free user

[rigid musk]

If you don't want to buy you can't do nothing

[coral zephyr]

want me to buy the vision app instead?

[storm summit]

You would need to get the ChatGpt plus feature so you unlock GPT 4 and even with that. The vision model is still rolling out not everyone has it! If you're in the UK and EU well you're going to have to wait a while because EU and UK are super awesome 🙂

Keeping people safe 🙂

[coral zephyr]

i live at California

[storm summit]

You will likely get it by October 9th if not earlier! Assuming you have ChatGPT plus!

Also happens to be my birthday on that day 😦

Well at least i'll get a quest 3 😄

[coral zephyr]

what about dall-e?

[storm summit]

Um with dall-e earlier october however

Microsoft have said it should be with everyone at this point

on the bing chat

Try to see if you have by creating a image with text if it's clear and not giberish you likely have dalle-3

[coral zephyr]

do i need to reload the page if it’s gibbrish?

im not sure how i can get dall-e to improve on characters, stuff, and franchises

but what’s the link to dalle-3 to be sure? (Not sure if i should search it up since it still might be the same 2.5)

[storm summit]

try a few generations

should look something like this

Something like this

[coral zephyr]

how i do that?

[coral zephyr]

im using bing

So, how can i have dall-e 3 recognize skylanders?

[trail cairn]

if its not able to create images of them then it means there were none in the training data or its avoiding them due copyright issues

[coral zephyr]

it can use Nintendo characters just fine

[trail cairn]

yes but they are far more popular, or at least i had to google what skylanders are 😛

[coral zephyr]

so copyright issues are thrown out of the window

[trail cairn]

nope, there is also the report feature to be excluded and that was available way before de3

mj made skylanders easily

actually bing makes them too for me

[coral zephyr]

how? I can’t successfully have bing make smolderdash

and can’t even make a accurate muffet from Undertale

see?

[trail cairn]

those are not dall-e 3 looking images

but when you ask for a specific character or figure the training data would have needed a lot of data of to do 1:1

[coral zephyr]

ok, but hoe do i do that? Im on mobile

[trail cairn]

if you never used it or only a little then you might not yet have access

[coral zephyr]

So what should i do?

[trail cairn]

you cant do anything but wait

[autumn bolt]

Is there a way to get access to GPT4 APIs?

[rigid musk]

By paying ?

[random ingot]

I have a friend who lost access to his email that he originally signed up with. Now he wants to create an account with a new email, but his phone number is still registered to the old email. Is there any way to get a new account created for him?

[charred parcel]

you can create two accounts with one phone number

[random ingot]

I just tried, it doesn't work

I used a different email, it prompted me for my number and said the number is already associated and I can't move forward

[charred parcel]

time for a new phone number then

[random ingot]

Thanks, are you sure there's no other option?

[charred parcel]

if the phone number has only been used once then you should be able to create a second account with it. If that failed then it must have been used more than once.

[random ingot]

ohh so there is a two use limit. I see. He may have used his number twice already

thanks!

[autumn bolt]

I am paying. But only GPT3.5Turbo works.

[coral zephyr]

i needed to make sure, if i shared the chat (the one from chatgpt) to my friends, they cant edit it, right?

[charred parcel]

they can continue it

[coral zephyr]

that’s a bit concerning, how should i do it without my friend abusing it?

[charred parcel]

when you share a chat anyone can continue it in their own instance of ChatGPT, it won't affect yours

[echo stag]

I was curious whether the new ChatGPT features (seeing, hearing, and speaking) will be available only on the app or also on the web browser?

[bronze charm]

Seeing for both, hearing and speaking for mobile

[echo stag]

thanks

[bronze charm]

👌

[slim gust]

Hi everyone! Does anyone know if I could change my email address and keep my current phone number? I will be quitting my job and lose access to that email address. I have tried adding additional email addresses to my account, however I am not able to log in using those. Any help would be appreciated! 🙂

[rigid musk]

[e-mail]

[steep mantle]

hi, so i accidentally linked my openai account to this discord account which i think i accidentally just created a new one, and trying to verify with my actual discord account isn't working until i unlink this one. can anyone give me a hand with that please

[severe halo]

I could be wrong but you could go to "Authorize Apps" and deauthorize "Openai verification?"

[fluid halo]

nevermind i figured it out

[fluid halo]

no, you were spot on, thanks

[severe halo]

Glad its solved.

[glossy shale]

Hey guys, I wondering why I don't get "browse with Bing" in my beta features... I'm plus user.

[bronze charm]

It's still rolling out

[oblique rain]

Toomanyhands.

[bronze charm]

What if the one hand is just really big and can spread across the entire card?

[analog shore]

Hello, is there a way to gift a sub to another person?

[spring latch]

yahp

[zealous bear]

just curious, but is the billing for openai down? i've been trying to work with it for a bit, but it's not taking my card ;-;

[regal lynx]

This image isn't G-rated and has been deleted. In the future, please only post images that are G-rated on this server to adhere to Rule 2

[analog shore]

How? I can't find how to gift a sub. Im not asking for free subs, i want to buy and gift one

[storm summit]

Can you deliver a message to Sam altman for me

Tell him we love him!

Tell him to bring out the monster

That's GPT 5 (plz)

[regal lynx]

I wish I had a direct line of communication with Sam Altman, but I don't work for or represent OpenAI. As a result, I've never spoken with him

maybe one day!

[storm summit]

Its OK maybe he's reading this

[regal lynx]

[storm summit]

😄

[analog shore]

Hey Sam, give me back my soul

[storm summit]

Why is bing so mean 😄

[regal lynx]

I've heard that Bing is super unhinged with what it says sometimes

[storm summit]

Look at this I was just trusting

Testing

Wish it was more like chatgpt

Its fine tuned in such a way where its like a snowflake

[storm summit]

Yh haha

[feral jasper]

what

[fierce haven]

mine often says I can't translate that to Japanese, I didn't even ask it tho

[analog shore]

Bing = Yikes

Its like your ex

Alrighty then, theres no way to gift a subscription? 🤔

[feral jasper]

Bing or Google

In searching, no AI

[regal lynx]

the image you shared of the characters holding the OpenAI sign. The character on the left had clothing that was very revealing

[regal moon]

I bought plus but I’m still don’t have access to the photo options or searching future nor do I see the options available in Settings

Any help

[near prawn]

Vision is rolling out over the next 2 weeks

not sure about the others

[vivid terrace]

So is the voice feature right?

[near prawn]

The voice feature is Android and iOS only

[vivid terrace]

I know. Im using ios but i dont get it yet. Just wondering if it’s going to roll out the same way

[near prawn]

They didn't mention it

I already have it myself

Had it since they announced it

No Vision yet

[vivid terrace]

Just checked. Seems it rolls out slowly. I dont see it under my new features yet. Thanks 👍

[slim gust]

Thank you, Neko!

[dark perch]

does the staff actually read the feedbacks?

[rustic anchor]

can i get gpt4 API to create app for myself with it? i buy allredy gpt4 for mounth

[rigid musk]

Yes

[rigid musk]

Yes but different payment

[rustic anchor]

can you give links on informaitionn to read?

[heavy bluff]

My chatGPT Android doesn't have Settings → New Features. What's going on?

[rigid musk]

https://openai.com/blog/gpt-4-api-general-availability

[rustic anchor]

thanks

[rich pagoda]

How can I use the read function of ChatGPT? (Mobile, Android, with Plus)

Or is this not released yet?

[analog shore]

Anyone with experience on red teaming?

[woven brook]

So they again removed intenret access ?

[arctic holly]

They do it here and there for the past week for certain platforms. If you’re using the app try the mobile version or check back in an hour or two. Your not missing much the revised search sadly is censored they do not look at all sources and chatgpt is programmed not to look into things that are politically inconvenient.

[sage steeple]

Yeah man I was waiting the last couple weeks to get the GPT4 with bing feature since subscribing and after getting it a couple days ago I’ve been super disappointed. It does like a single search and barely dives into the search results

[onyx kraken]

On my device, under App I see 'Beta Features', directly below Color Scheme. It is possible the rollout hasn't reached you yet, and maybe if your Beta Features is empty, it doesn't show. The rollout has another week expected based on the first announcement.

You can share in #1070006915414900886 if you want, if there's a problem that might get help!

[wicked wedge]

Hi, the free version does not have access to real-time data. Are there any versions which do?

[timber shadow]

ChatGPT Plus gives you access to GPT-4, which has a knowledge cutoff date of Jan 2022. It also gives you access to a Browse model which can search the Internet for more up-to-date information, and a Plugins model with various plugins that can do similar things.

[wicked wedge]

Does the browse model allow it to use that data to formulate answers?

[timber shadow]

So it'll search the Internet per your request, and then yes it'll use what it finds in that search in its response. So the model pretty much just knows how to look up stuff online, and then tell you about it. So it's a little different from "regular" training data.

[carmine fable]

I don't have the browse model, but i have GPT4
is it going to be available soon for every plus users or do i have an issue ?

[rigid musk]

they still rolling out

[sage steeple]

The Browse with Bing model really isn’t that great, I have access to it and I far prefer regular Bing chat to the feature

[potent pewter]

i cant reset my password it isnt sending the instructions to my gmail what do i do

[onyx kraken]

https://help.openai.com/en/articles/6613657-why-can-t-i-reset-my-password and can also use the bottom right icon to chat with the help bot and also send a message to openai staff, but they can take a week or 2 to answer.

[potent pewter]

oh hell no

[arctic holly]

It used to read PDFs and could compare contracts but disabled that feature.

[dim cipher]

chatgpt is not working for me. how can i ask for help ?

[rigid musk]

Try firefox as test