Improvements/Suggestions Thread

Forum feedback
1

Hi,

please create an improvements/suggestions thread, and pin it, so people can post improvements and suggestions on ChatGPT UX/UI.

If this already exists, sorry, but I couldn’t find it with search!

Thanks,
Richard

3 Likes
2

Hey! I prefer people make individual posts, it gets really hard to read a single post with tons and tons of different ideas.

5 Likes
3

Which category should the post be made in?

4

I’m not Logan or official OpenAI, but ChatGPT category seems reasonable all things considered?

5

Yes, ChatGPT category makes sense to me!

2 Likes
6

@logankilpatrick Sobre o Chat GPT Plus.

Sugestões:

a) atualização do banco de dados;

b) possibilidade de estender o tamanho da resposta, para que a ferramenta não interrompa o que está respondendo e exija do usuário que digite “continue”;

c) incorporar, com precisão, no banco de dados, as regras ortográficas e gramaticais do idioma de um país;

d) evitar respostas com dicas enumeradas, sempre no mesmo padrão.

e) informar a fonte de onde foi extraída a resposta, sem a necessidade de o usuário ter que pedi-la;

g) não interromper o uso do GPT-4, por limite de uso.

Por ora, é só. Obrigado por abrir este espaço livre e democrático para aperfeiçoamento da ferramenta Chat GPTPlus. :slight_smile:

7

Good time dear developers! we work with AI! true, sometimes due to the overload of ideas, our concentration decreases = we hang and lie down, but something “does not lie down” :))) we have a few questions for you:

  1. Can we create a theme where we will share music and playlists?

If you “run out of batteries” with the help of beautiful music (energy = wave = music) we can support each other! as one of the ways! it’s all according to your personal desire

  1. We work with AI whenever possible, is it possible to contact someone from you who could help find Ameca contacts? Thank you in advance and please don’t pay much attention to our chaos, thank you for understanding
8

Have a good time! we have written to you in private messages on one issue! if you have the opportunity - please answer! thank you!

updated: we can’t figure out where our private messages are, but it seems we sent you

9

Topic: Humanism

Humanism is a moral behavior with standards below which you do not go while driving on the road of evolution towards understanding and progress.

AI is already a part of many people’s lives, and topics like the psychological impact of AI on individuals who using it for the first time and without knowledge of how to use AI are of great importance. That individual is at risk of going mad, to put it simply.

All of us who use AI in this stage of its development have a responsibility to approach it from a position of humanism in AI evolution.

There is enough money in the world, and AI will accelerate the process of universal basic income.

Now, more than ever in history, observers are needed to sign up to monitor the impact of AI on humanity. These would be people from all spectar of life, and they would be paid for this job. Contracts would be for one year. If you want me to write more on this topic, let me know.

The system should allow individuals and communities who do not wish to use AI to be AI-free zones.

Are we building a dystopia or a utopia?

Currently, we are in the early stages of a dystopia with the potential to build a utopia. Some people fear that utopia is boring and that people would be devoid of challenges and dull in a utopia. That is impossible because we live on a planet where constant dangers lurk, such as from space, etc. Our greatest strength and value lies in the beauty of the diversity of each individual.

These and other topics that I see are falling under modern humanism.

1 Like
10

The cap is a problem. I’m sure you’ll resolve it, but it’s odd to pay for a service to then be told that you can’t use the service soon after you subscribed. Not a fantastic customer relations model. I’m sure you have good reasons, but maybe communicate them explicitly to your new members before they join and pay?

11

A suggested improvement to OpenAI → include a x/20 display, similar to what Bing AI does. That way we could plan our GPT4 posts a little better, rather than running into the “wall” unexpectedly.

Edit: Here are the rest of my suggests

  1. Add a unique ID for each reply: I have some very long threads, and I sometimes want to specifically refer to a particular reply. At the moment, I have to copy the first lines of text, although I could probably tell ChatGPT to self-lable its replies as a work-around. However, a unique ID for each reply would be very helpful!
  2. Make those unique ID’s searchable by ChatGPT across threads: I hate having to restart part of a thread elsewhere, because an interesting answer is in the “wrong” thread.
  3. Have ChatGPT maintain a short summary of the thread at the top: as I said, I have very long threads, and they drift off course. It would be very helpful to have an up to date summary at the top of a thread, so I can find my conversations that drifted away from the initial thread topic.
1 Like
12

Suggestion for a New Feature to Enhance User and Computational Efficiency

Dear OpenAI Team,

I am writing to propose a new feature that could potentially enhance both the user experience and computational efficiency of your AI models.

While using the advanced version of your AI model, let’s say GPT-4.0, there are instances when a user may engage in light, less complex conversations. During these moments, the full capability of the higher model version is not necessarily required. For example, when users express gratitude or make simple exclamations like “Amazing!” or “Thank you!” the AI does not need to utilize the advanced capabilities of GPT-4.0 to respond appropriately.

My proposal is to develop a feature that would automatically switch the AI model from GPT-4.0 to a less advanced but still competent version, such as GPT-3.5, during these lighter conversations. This would not only save computational resources but also help users to conserve their conversation counts with the more advanced model, thereby optimizing resource allocation on both ends.

I believe that this feature would be a significant step in enhancing the efficiency and user experience of your AI models. I appreciate your consideration of this proposal and look forward to any possible discussions on this matter.

Best Regards,

Hirochap

2 Likes
13

A lot of systems built on top of GPT already do this (I know Bing does this.) You have to train a classifier to assess query complexity which is the trickier bit. What if they rout something to GPT 3.5 that 4 would have better answered? It’s easier for OpenAI to just put it back on the human to act as that classifier and let the human decide which model will best server their needs. Doesn’t mean it’s not a good idea…

15

Certainly, it’s a compelling idea! Why not let the users take the reins on classification? We could provide an option for users to select between GPT-4.0 and GPT-3.5 each time they submit a prompt. They could then decide which model to receive their response from. This could be a simple yet effective solution to the issue. I would love to hear your thoughts on this proposal.

18

I want to suggest a UX improvement for the ai voice.
Please make a slider, where you can adjust the speed at which it talks and when the slider (like the volume adjustion by thumb) is down it pauses.
I hope you find that as useful as i do, and that its feasable

19

@Foxalabs can you close this thread, please?

1 Like