Is fine-tunig dead?

Hello,

I am leading a business creation project in AI in France (Europe more broadly). To concretize and structure this project, my partners recommend me to collect feedback from professionals in the sector, and it is in this context that I am asking for your help.

Lately, I have learned a lot about data annotation and I have seen a division of thoughts and I admit to being a little lost. Several questions come to mind, in particular is fine-tunig dead? RAG is it really better? Will we see few-shot learning gain momentum or will conventional learning with millions of data continue? And for whom?

Too many questions, which I have grouped together in a form, if you would like to help me see more clearly the data needs of the market, I suggest you answer this short form (4 minutes): https://forms.gle/ixyHnwXGyKSJsBof6. This form is more for businesses, but if you have a good vision of the sector, feel free to respond. Your answers will remain confidential and anonymous. No personal or sensitive data is requested.

This does not involve a monetary transfer.

Thank you for your valuable help. You can also express your thoughts in response to this post. If you have any questions or would like to know more about this initiative, I would be happy to discuss it.

Subnotik