A Librarian tries ChatGPT

Margaret DeLacy's picture


The Scholarly Kitchen blog has posted a guest column  entitled "The Efficacy of ChatGPT: Is it Time for the Librarians to Go Home?" by Curtis Kendrick describing his experience taking ChatGPT out for a test run on a research question.

His results suggest that the answer to the question in the title is "No."  The software returned a reasonable-sounding reply written in prose that resembled that of a  "good Wikipedia article" (not sure whether that was intended as a back-handed compliment) but completely failed when it came to providing citations. 

The column drew several comments that are of interest in themselves.  One author noted the tendency of Curtis, and possibly many others, to anthropomorphize what is really just a sequence of code.

You can find the entire article and the comments here: