Grad students using AI for literature reviews, where’s the line for supervisors?
I supervise two masters students right now and both have admitted to using ChatGPT extensively to summarize papers for their literature reviews. Neither used it to write actual analysis, but the summarization step used to take them weeks and now takes days. I am supportive of this in principle since the summarization work was never really the intellectual contribution I was assessing, but I want a clear line I can communicate to future students. Where do other supervisors draw the boundary between acceptable use for efficiency and use that undermines the actual research training a graduate program is supposed to provide. Curious if any universities have published formal graduate level guidance yet, because mine has not.
5 Replies
Join the discussion.
Log In to Reply
this seems workable for grad students who already have strong foundational reading skills, im less sure it scales down to undergrad thesis students who might use the summarization step to skip actually engaging with the source material in the first place.
@Samantha Kim thats a great distinction actually, i think the graduate context assumes a baseline of research literacy that undergrads havent built yet. might need a different standard for each level.
spent way too much on a paid summarization tool two years ago that turned out to be worse than just asking chatgpt directly. lesson learned the expensive way, wish id known then what everyone here already knows now.
the line i've found useful in practice is asking students to submit their own one paragraph summary of a paper before they see any ai generated version of it. if their summary and the ai summary roughly match, the tool saved them time without replacing the thinking. if their own summary is thin or missing, thats a sign they leaned on the tool instead of engaging first. i ran a version of this myself last term using proofademic to spot check some of the phrasing differences between drafts, mostly out of curiosity, but the paragraph first method did most of the actual work.
my university published guidance last spring that draws the line at analysis and argumentation, summarization and organization tools are explicitly permitted as long as the student verifies accuracy against the original source. seems like a reasonable framework and one i've adopted informally even without a formal policy on my own campus.