If anybody likes any of the ideas here please join in the conversation and maybe create your own relative threads asking for these features too.
I’m not sure if this would work perfectly in every instance but wouldn’t it be cool and this would be a tickable option for programmers that when you paste in your items like your code and your request to GPT on the chat GPT website if it added your code that you pasted to a console like how GPT outputs your code on the website so it was easily readable and looked nice and then you know you could have in the background a gpt 4o mini… Model add your request to a a separate console that is always above the code even if you type in your request below the code and it wouldn’t be a standard console like where GPT adds the code with color code formatting it would be a similarly designed panel that is improves the look of your user request and in my idea each one of these consoles to separate them from GPT’S output would be in a slightly different color like a light gray or something so still easily readable but you could easily locate it…
the main reason why I’m saying This is the every time you paste in code and your command it really never looks nice because you have a small window to type it in or paste it into so you can never get the formatting right… And it always kinda looks just like a jumbled mess…And I’m sure everybody is voice typing anyways so there’s gonna be voice typing errors in there that subtly destroy the formatting as well
hoping people come here religiously from open AI to inspect this As this one’s pretty important…
but I was impressed at the awesome new improvement that lets you know immediately when you send a request if it’s too long… But I notice when I pasted what turned out to be a 4000 line AST in there and tried to shorten it so it would run by deleting half and then looking at the button to see if it was lit and it wasn’t so I deleted half again and repeated that one more time and then when I finally got the button to light and pressed it it processed the code and then said that the message was too long so I feel the best way would be to use a very simple python len() Every single time text is entered just as the very top cond of your code…
This is awesome idea allow to duplicate a chat session because basically I’m in one chat session right And I’m Testing code it fixed for me and it’s taken a couple minutes because I’m doing open AI stuff you know so I just go over to my window and I say hey by the way while I’m doing this you know can you look up this api and find me a new solution for something you know completely unrelated and you know then I go back to the window and I think okay you know II finished testing the original code I was testing and I have results I’d like to paste in but then my the latest data I got from the API look up for something completely different is going to get buried so this is a little tiny improvement all you have to do is just add a duplicate button you just duplicate it From the menu that allows you to archive it… and continue working in one of them and then when you’re ready to analyze your API look up you could go back to the other one and you could even have it name it appropriately based on the last output response.
1 Like
Add a memory use indicator bar to each chat:
See if you read above where I’m discussing creating new chats when the memory gets bogged down that pretty much is a very integral variable in the level of trust you could give the AI models on your site so you know it just works great you know GPT 4O model you know just works awesome all the time but do I completely trust it with my code all the time and that’s no I would if it was always like he just opened up a brand new chat but I don’t as there’s a certain gradient Concerning how long you been in the scene between the time when you first get there and the time were completely starts to fail because of too much memory and that probably not totally in and of itself but that is a huge contributing factor to the amount of trust your users can give to your models and if you just you know really work on handling the memory that it accumulates and stuff you could totally minimize or even possibly eliminate chances of failure in that area… For example if it was always a brand new scene and it worked like it was always a brand new scene every time without the memory bugging it down the models would be just way better you know all the time you could always trust them it would be really robust as far as How much you’re able to trust the output of the models.
So to sum up the main contributing factor to the trust issue above is how long I’ve chosen to Have stayed in that chat without switching to a new one So in a nutshell it appears that you know sometimes your motto will fail but it’s really my fault because I stayed in the chat too long and I know that but all your end users may not be aware so you could put a beautiful colored RGB indicator bar across the bottom of the chat… And you know the point that you decided to flash the warning message and block people out of a scene when it’s accumulating too much memory make that the red and right when you get to a scene it’s green and then um you know have an indicator bar there so you know as memory piles up it goes from Green to yellow to orange to red and then when it hits red you block people out of the chat… Then if I get a weird response I could look at my indicator bar and be aware…
1 Like
The tool tips are interesting i’d like to see where you go with that but I think it’s a vital that tool tips only contain brand new information the thing that I noticed was that copy was obvious to have a tool tip that says copy under it… And it kind of gets in the way I like just clicking the copy button without it…
I think the one over options is obvious too because everybody knows what those triple dots mean and but the messages is is empty could probably be a benefit because sometimes I look down and wonder if there’s anything in there because sometimes the webpage is scrolled in such a way that there’s recently input text that you input there that looks just like what’s in the box and sometimes have to look up or have to look through the box…
If it were my website I would just make sure each one has brand new information I would I would try and make him just a tiny bit smaller maybe even teeny tiny tool tips where you could glean the information but they don’t interrupt your beautiful website interface…
The world’s best community feedback and feature request system:
All these requests if they have requests or feature requests or feedback or anything you should have the AI automatically parse this site in your own internal system and then every time he gets a feature request that it thinks is a good idea it should let you know. And to derive whether it’s a good feature request it should scan all the information that the person that send the request has in all their topics to get perfect context about this person then analyze the feature request versus another database of you know everything you’re currently working on your system and everything you know so this could be a pretty you know involved system but then instantly every time you get a feature request that GPT thinks is good it’ll send you an email and that then you’ll have a list and then you’ll have another assistant looking over that list all the time keeping them in the proper order of what it thinks is most important and then you know you could scan that logic every now and then to make sure it’s doing the right thing… That way it would be a genius system that would practically inject community feedback as an operator in your code with just a couple more tie ins you could practically have the community As a function in your system…helping you keep your system robust.
1 Like
Integrate WINMERGE functionality to chat output responses for coders:
Every time GPT outs puts code to you if you click a little check box you’d see an instant differential just as good as WINMERGE so you instantly know what it added
And keep in mind this would be completely invisible to everybody else unless you tick something in settings under what I would suggest a new coder’s tab in the preferences and then when you take it on it could still be a tiny check box that you enable and then in this case you know every time GPT outputs code essentially A WINMERGE window would appear.
1 Like
Also happy to help expand on any of the logic here if you don’t think it’s prudent or can be done ask me about it i’ll answer any questions.
Also per the display of the website for example the scrolling is somehow tied in to wiwhen GPT is generating code for you I don’t know if it’s maybe tied in with a cachet system that you’re storing on the end user’s operating system because sometimes the scrolling gets slow …if it is tied into A cachet in preferences I would expose settings for the end user to interoperate with this cachet and delete it and understand it I think that’s important for any cached system that works through Google Chrome… Especially when likely most of your end users aren’t programmers
Throw open AI at Macy’s Day style parade in New York City, as a thank you, literally…
Also I came up with this idea because the makers of open AI pushed this into the forefront I think mostly by themselves got it known and you know now we have awesome AI in our phones and everywhere else came up with an idea it only cost $80 if somebody wants to jump on this first and I know it seems kind of corny but it’s not because this is world changing technology we’re going strong into the a new technological age so I was thinking that as a thank you…
It would just cause $80 to get this set up with videos but we should make a Kickstarter, keeping in mind my funds are currently tied up but if somebody wants to jump on this i’ll send him a link of a guy who can make a Kickstarter that would raise funds to as a thank you throw Open AI seriously awesome Macy’s Day style parade in New York City as like thank you for this major boon in technology for the entire world for pervying it like they did until it got into the cultural zeitgeist and everybody adopted it…
Chime in if you agree with this but I think this is really important and just 80 bucks i’m having a Kickstarter built now for my own products so I’m aware that this guy will do the videos and everything on F I V E R R…for $80
But I think the whole world would join in and donate funds just to give the creators at openAI thank you for all they’ve done.
And yes definitely what they did is a big deal!!. They create the software for a new technological aids that is making a lot of us a lot of money.
Here’s the link of the fiverr pro Login to Fiverr
So if you see this and openAI has helped you incredibly like it’s helped me just go ahead and grab this link and this guy’ll do all the work for you and just get started giving back to openAI
If you do work it in to proclaim yourself master of ceremonies of the parade along with celebrities of course who I’m sure would be happy to join in and maybe work it in so you could pitch your own product you know as it thanks for yourself or your own enterprise…
But hurry up because this is a really good idea so as soon as I make money with Current Kickstarter I’ll be jumping on this chance…
Edit:
I am so incredibly busy with my business I haven’t had chance to implement much on this except one thing and that’s I sent a message to an editor at Forbes magazine that wrote an article about Chat G P T and asked if they would write an article trying to get public support for this…
Everybody is encouraged to do the same that reads this just think when you’re sitting there having open AI servers generate your programming code just try and remember what it used to be like before this was available and then head on over to the Forbes website and send a message with their contact links on the bottom of the page and just you know public support just to thank open AI for making this all a reality…
Give ability to expand message box where you enter text into chat GPT
Now this one is absolutely vital for when you’re generating code, what you should do is make it so you can mouse over the edge of the box where you enter all your text into chat GPT and drag it up so you can have a better ability to edit the text that’s inside of it and view it… If you just look at what they do on blender artists .com like when you’re entering a comment or something or reply to a response then you have the ability to always drag up on the box and you know have a full workspace to edit whatever you’re adding easily…
Google Chrome profiles are affecting your product
I have the Google Chrome profile that I always use but its email address is different from the email address I used to log in to open AI and get to chat GPT and And I was so happy today that I could log in to Chat GPT with my proper email address in the profile that I always use that has all my bookmarks. The previous days every time I tried to log in it tries to automatically Log me in with the Gmail address that’s associated with the Google Chrome profile that I always use… If there’s something in Google Chrome That’s making it so this happens the thing you’d wanna do here is talk to Google Chrome developers and let them know that this is a world class product, chat GPT is, that everybody uses and see if they could do some sort of work around to make it so it doesn’t happen
Another reason why showing differentials on your website for people that are generating code would be awesome…
.
See when I use chat GPT to generate code and I tell it to fix a function or upgrade a function when I copy it and take it back into visual studio code I what I do is since I installed a special visual studio code extension to do this I right click and then what it does is it compares the new code with the selected code which is the original and opens up two windows in visual studio code where I could see a differential and see exactly what it changed when I when I do this I’m using chat GPT like a master you know I instantly know what the AI is doing to my code exactly I know if it made any changes that I don’t want and I know instantly if it made a quality change which it almost always does… I instantly know how to do this upgrade myself so in this instance I’m using ChatGPT like a master because I you know and I’m able to see the differential so I would really suggest that you add the differential system so all your end users that haven’t figured out this would have this capability. it’ll Make Chat G P T way more awesome.
You know how GPT doesn’t allow violent content or nudity you know I fully support that but you know another thing you could add to the list is stuff that just somehow would crap on a puppy or a kitten…
Why don’t they allow nudity I mean violent content is obvious?? edit:(Fixed voice typing error)
Because humans are reproductive animals and if there was nudity allowed there would be little possibility of getting any work done.
And what does always show details mean there’s a little toggle to toggle it on and off but I never see any effects I think that should be explained with the tool tip
And I just read the o1 mini doesn’t support the use of tools On the retry menu… Which is apparently why it was grayed out… I think that should be explained with the tool tip as well so people can immediately know what that means without looking up the Internet
could make it so you could drag out the right edge of the console
See when it’s printing out code a lot of times it’s printing out text that goes past the edge of the screen while it’s printing out I think it would be more efficient if I could see what it’s doing and make sure it’s doing the right thing so if I could drag it out real quick and read it and then drag it back when I’m done that would be more efficient
1 Like
I hope open AI sees this because basically in a nutshell I get so excited when I’m having code generated because this is like a machine that is just making free stuff I mean you’re getting free gifts every time you have a generate code…
I’m so grateful most of the time I can’t even look at it while it’s outputting code because you know I used to program without this stuff and I know how great of a gift this is…
Since you have the SHAPE e library that automatically generates 3D models if you could partner somehow with the makers of a 3D printer that people could Could purchase that’s a quality one but that’s a really low price it could couple with your system And you could link chat gpt to it and shape to it and essentially get closer to having a machine like in back to the future that would just materialize whatever you need so you’re sitting at your computer saying You know I want this or do this and it just creates these objects it creates code for you it creates objects it can be expanded later to where this controls home automation all throughout a chat GPT interface like maybe have robot arms that will reach in your cabinet pull out a pen and reach in your fridge and start cooking…
Maybe it could merge into a desktop app or even do this from the web interface but you could give it Google voice functionality where it could make calls for you a link to your email where it could send emails even more perpetuate the idea of it being an assistant it could you know connect to your windows computer create batch files for you to automate it and automatically pin them to your taskbar…
When you run a model have it make another pass over the entire history analyzing kind of what’s going on…
This would be for if you’re having it generate code you could have it look at your logic look at your code look at what you’re telling GPT to do look at what it’s outputting and see if that’s logical maybe you just told it like a good idea you saw it but maybe it wasn’t the most logical move you know because I was thinking maybe if you tell GPT to do something like write the code this way it could say are you sure you want to do that … You could use the OH-1 mini model to just do like a pass in the background every single time after you send GP to your request to just scan for anything of logical that may be GPT did or that you requested and if either happened you could just put a little tip to the user on the right next to the console saying what happened so you could remind GPT or figure out what you said illogically. It would be better if the model was instructed to only send you a tip if it was 100% that this was illogical… Skipping anything that has any risk of failure
You should add good tested code to your model because it’s looking all over the Internet right you should make it even better by adding a database of tested code what you could do is every time somebody makes a comment on chat GPT that says OK that worked now let’s do the next step or something then grab that code then just pay a guy to sit there and you know test the code and add it and as he does that figure out ways he can automate that process for himself and just build up from there so it’s like what you have now where it scours the Internet I think for working examples but you’re adding a database of tested code that you know works right off the bat.