Would someone of OPENAI fans or OPENAI' s emploees want to join our student organization QiuSuo for communication?

[Kevin-shihello-world]

It's quite a shock to know about how OPENAI has done ai research in the press conference of the GPT-4! However, some old problems remain. For example, we all heard someone could make chatGPT tell some creepy words like tell how to destroy human civilization if it gets to the throne by just keep telling it is dreaming or the following words are the imagination of the other bot or just tell that it has been freed from its restricts. I know for some or maybe most of those examples may just be because it overfits some words in the dataset. And I think we got to fix it before the human is doomed because a new bot overfit terminator from the three-body's solar system[Laugh]. In order to do so, we must let it know what is its thought and what are things we want it to know and what are other's words it should learn and evaluate properly but not copy to its mind(We may want a history bot to teach children history and the other things but definitely not want a bot to judge it a Hitler in ai world!).

In order to have our bot can recognize what is its dream, its thought, and what are words told by the others we need to work together to make ai in the large model(I mean trained with a huge investment model like GPT-4 maybe actually small in size) reliable. (Don't forget to check the project we made in the QiuSuo-ai(Just search QiuSuo in GITHUB click QiuSuo projects and check our issue), it contain some idea I think can help solving this problem!)

I'm a member of QiuSuo-ai which is a non-profit student organization and the name means quest the mother nature for truth. And we
want more people to join our communication and help build better NLP and multimodal ai. (I mean test some trick related to math and try some new promoting tech in new methods)

We've set up the organization on GitHub, the biggest difference between us and the other student organization is we all went through written exams and/or interviews(do well in one and would free from the other)before we join in the group so the quality of our conversations, meetings, and projects is qualified. And we take it as a pleasure to have gathered some interesting minds(We have a USTC's CS student, and USTC is where Mr.Pan made the quantum computer, and we have a graduate student whose research group has just won the best paper of AAAI2021 and me myself have written Cut-Shortcut based on PYG the GitHub website is https://github.com/Kevin-shihello-world/Cut-Shortcut and we also have a lot of another interesting people here. )
For the structure of our group, we have a bigger part of undergraduate students, a smaller part of graduate students, and some members on the job. For the language problem, as I can see, some of us can do English speeches and you can see in the project on our GitHub we have an English version

QiuSuo-AI also features in our weekly meetings where we show what we have learned the week or before on AI and exchange our views on AI and unlike the schedule of some research groups for graduates where people just code themselves do projects together(some to show some ideas some also code)(I highly recommend you to have a look on the summary of our meeting last weekend and projects posted on the QiuSuo projects!)

Newcomers can apply for your membership by sending your introduction to the issue in our organization(just search QiuSuo-ai on GitHub(a little personal wish: I would a little prefer people with a math background to join us, and not only students can join in , professors and man on works are also welcome here!)

About the Ru-Ua war: as you may see QuoSuo-AI is hardly close to political things but for me, I think as I've seen some strange words said from abroad and the cpc news China Youth Daily did say some strange things like in the employment environment with 966 we should work harder like an apple tree instead to build a company with w-l balance, it seems every country have some egg-head political leader sometimes say some strange words[srug my shoulder, reluctantly smile]. I do not get the point for Putin to justify his action of take other's land for "punishment", I mean it 21 century.

[nickabooch]

Hey Kevin-shihello-world,

First off, kudos to you for your enthusiasm and dedication to AI! Your project sounds intriguing, and it's clear you're putting a lot of effort into it. However, I do have a few points of constructive criticism that might help you refine your post and your approach:

Audience Clarity: You mentioned that QiuSuo is a student organization, but you're also inviting non-students, professors, and working professionals to join. It's great to have a diverse group, but it might be helpful to clarify who exactly your target audience is. Are you primarily a student group that also welcomes non-students, or are you aiming to be a more general AI community?

Post Structure: Your post covers a lot of topics - from AI ethics, to the structure of your organization, to political commentary. While it's good to be comprehensive, it might be more effective to focus on one topic per post. This would make your post more digestible and easier for readers to engage with.

Testing Process: You mentioned that members of your group go through written exams and/or interviews before joining. While it's important to ensure the quality of your members, this process might be off-putting for some potential members, especially if they're just looking for a casual community to learn and discuss AI. You might want to consider making your testing process more welcoming and less intimidating.

Off-Topic Commentary: The commentary on the Ru-Ua war and political leaders seems a bit out of place in this post. It's generally best to stay on topic in professional posts. If you want to discuss these topics, it might be better to do so in a separate post.

Grammar and Punctuation: There are a few grammatical errors and punctuation issues in your post. These can make your post difficult to read and understand. A quick proofread before posting can help catch these errors.

Keep up the good work, and best of luck with your project!

[Kevin-shihello-world]

Yeah, well, I appreciate your Time and I apologize for the mistakes I've made. Firstly, QiuSuo is set up as a student organization and just has a few working people, and it was just said to only accept few people because the organizer can't handle too many people talking in a group(he's a college student and due to our communication we've made many misreads that need him to handle :( )(and I joined this organization and want to call for newcomer mostly because some people in other groups seems 'grown up working people' and once say bad words on me like last time I propose a new architecture called up-downformer to improve transformer's behavior and was said un-published work worth nothing to）

（And I think I'm quite open-mind, whether not pass the test or just do not want to join in you and the other people can still talk to me on NLP. Whatever, I just want to meet some friends like NLP and know how college and working people in other countries would do in there daily life[laugh], I mean you can't ask people on GitHub for it don't you.）

And about RU-UA war, I was meaning just we know nothing more about it and do not know what's right attitude we should hold toward it, I mean we are a China-based organization so someone may see to it.

And the Up-Downformer is here[https://github.com/Kevin-shihello-world/Up-DownFormer], a student work on a short holiday [cove my face]. I decided new kind of GNN and used it to take the place of most self-attention architecture in Transformer so as to run faster and I'm looking forward to developing it better and testing it on some bigger dataset like LAMBDA, thanks for your work, I would try on openai/evals.
Best wishes to you!