GSOC 2024 introduction + queries

Hello everyone, I am Benjamin Lui. I'm a final-year student at the National University of Singapore (NUS). I am interested in the project on using AI to generate content from a title and summary. I had a previous project done during a hackathon that used AI for context-aware translations.

I have a couple of questions regarding this project idea. The questions are split into frontend and backend.

Will multiple components require this AI feature to generate content?

What is the best practice for storing the OpenAi API key? Have the users use a key we provide or have the user provide their key to us?

Sorry for the long post and hope to hear from yall soon!

Hey @rnunez, I have completed a draft proposal for this. I was wondering what is the best way i can get feedback from you so that I can improve the quality of the proposal/idea?

if you mentioned my username I will see it, the backend the control-panel will be the ideal place to put it so it can be used by a backend process that all users can shared without sharing the key

@rnunez thanks for your prompt reply! I agree that the key should be on the backend side and will modify the proposal accordingly.

There is an OpenAIStore that directly interacts with the backend service. Frontend components that want to use AI have to talk to this store. The idea is that OpenAIStore handles requests from the components and the response from the database backend.

The components that talk to the OpenAIStore could be a tab that allows users to trigger the AI functionality. This tab will house a form (or some fields), that the user can fill in with the title and summary.

One other thing to note on the backend will be prompt engineering. I was thinking of spending some time before and during the program to pick it up.

Let me know your thoughts on the architecture (it is very bare and simple and for some reason I cannot embed links/images in posts...) and the overall thought process of this.