Nature recently asked me if AI could conduct a literature review.
I wrote 1,000-word piece to think through it.
The article that came out yesterday contains a few quotes.
Thought I'd share my detailed answers here:
Can current AI tools produce an accurate literature review?
I am not sure if the current generation of AI tools can produce accurate reviews of scientific literature autonomously without any supervision by a qualified scientist.
That said, AI tools can certainly make your review and writing processes efficient.
Let’s say, you are working on a literature review and you come across a certain article. You want to find out studies supporting and refuting the claims of that particular article.
You can do it manually, which will take a lot of time.
Or, you can do it with the help of an app called Scite. You run a paper through Scite, and it will give a breakdown of citation statements supporting or contrasting the claims of a given research paper.
Most AI tools that claim to do literature review are directed at undergraduate or college students.
They can give you a basic writeup at the level of an undergraduate student pulls an all-nighter and comes up with the main points of a few papers. That’s about it.
If you are looking for penetrating and original insights, AI tools are not there yet.