Yes, especially in the literature survey section. It’s optimal to perform literature survey with AI assistant. LLM tokenized and indexed a lot of papers, some are extremely hard to search correctly using a normal search and query engine like Google scholar. Instead AI allow us to query all related papers to a specific topic. By doing so, researchers can explore more relevant and maybe unexpected results from literature survey.

I do not support the use of AI to create the methodology and experiment section due to “hallucinations“.

But if using AI in literature survey is optimal, using it might be a good idea. And if you using AI to build a part of your paper, it should be part of the authors or reference as tools used.

Secondly, hiding away the use of AI is academic negligence and a sign of academic misconduct. You have to cite the information that is not yours, thus even the analysis of literature survey from AI tools is considered not your information. You will need to cite it, and in this case because it can not be cited, should be listed as authors or tools used.

Dear Ngoc, since AI models like ChatGPT or other research assistants cannot assume responsibility, accountability, or legal/ethical obligations, they do not meet the formal criteria for authorship according to most scientific publishers and academic guidelines.

They can be used, but I don't think they can be authors, because you cannot go back to the LLM and ask if the paper was generated or authored by them. There is a very specific conversation that would have led to the responses. We can't recreate the exact same conversations. Even if we do, we are not sure if we'll get the same responses. So, it's not the LLM that authors the paper, but rather the user who guides the LLM towards a certain topic or certain area of research.

But otherwise, LLM's are really good at connecting different dots effectively which humans can miss out. For example, taking references from older papers to complete parts of your paper.