RAG's future

[valdesguefa]

RAG is certainly promising, but do you think that RAG can be used by a company in general to address customer or product concerns?

[chen700564]

RAG is an effective method for incorporating external knowledge, but it faces two serious problems: (1) the retrieval results may not be correct; (2) the language model's responses may not be reliable. Therefore, for some scenarios like in a company, there is a need for more refined design and inspection, such as improving retrieval, enhancing RAG of LLMs, or using other methods to inject external knowledge (such as domain pre-training). Due to my unfamiliarity with the business scenario, I am unable to provide further suggestions.