Continuing
from my previous post on October 5 titled “TU Delft - How to use
ChatGPT-4 effectively to assess student work”, (link above) it's worth noting
that Stanford University researchers have recently introduced an
innovative tool that leverages the capabilities of the ChatGPT-4 language
model to provide valuable feedback on scientific manuscripts. This
innovation aims to address the pervasive issue of a scarcity of
qualified peer reviewers available to evaluate scientific studies.
The Stanford researchers
conducted a comprehensive analysis by comparing the feedback
generated by ChatGPT-4 with comments from human peer reviewers, utilizing
a dataset comprising thousands
of previously published papers. Their investigation revealed a
substantial convergence between the feedback provided by ChatGPT-4 and
human reviewers, showcasing a high level of agreement.
https://hai.stanford.edu/news/researchers-use-gpt-4-generate-feedback-scientific-manuscripts
