Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 26 of the 26 references it cites were found in a registry or an index; 3 of the 7 sources its claims cite were checked for support.

A Survey on Student Use of Generative AI Chatbots for Academic Research

Open access
Deschenes, Amy, McMahon, Meg · 2024 · Evidence Based Library and Information Practice

Abstract

Objectives – To understand how many undergraduate and graduate students use generative AI as part of their academic work, how often they use it, and for what tasks they use it. We also sought to identify how trustworthy students find generative AI and how they would feel about a locally maintained generative AI tool. Finally, we explored student interest in trainings related to using generative AI in academic work. This survey will help librarians better understand the rate at which generative AI is being adopted by university students and the need for librarians to incorporate generative AI into their work. Methods – A team of three library staff members and one student intern created, executed, and analyzed a survey of 360 undergraduate and graduate students at Harvard University. The survey was distributed via email lists and at cafes and libraries throughout campus. Data were collected and analyzed using Qualtrics. Results – We found that nearly 65% of respondents have used or plan to use generative AI chatbots for academic work, even though most respondents (65%) do not find their outputs trustworthy enough for academic work. The findings show that students actively use these tools but desire guidance around effectively using them. Conclusion – This research shows students are engaging with generative AI for academic work but do not fully trust the information that it produces. Librarians must be at the forefront of understanding the significant impact this technology will have on information-seeking behaviors and research habits. To effectively support students, librarians must know how to use these tools to advise students on how to critically evaluate AI output and effectively incorporate it into their research.

Overview

CoChat Summary

TL;DR

A survey of 360 Harvard undergraduate and graduate students found that about 64–65% had used or planned to use generative AI chatbots for academic work, even though most distrusted their outputs for academic use. Students wanted guidance on incorporating AI, prompt creation, and how the tools work, and they reported higher trust in a locally maintained AI tool.

Key findings

  • Around 64% of respondents had used or planned to use generative AI for academic work; 10% were unsure and 26% said they would not use it. ChatGPT dominated use at 81.18%, and the most common frequency was “occasionally” (39.59%).
  • Non-users’ top reasons were “Using AI chatbots feels like plagiarism” (63.6%) and “Using AI chatbots undermines my learning” (59.6%). Among all participants, concerns included factually incorrect information (88.6%), unclear sources (83.1%), and privacy (61.1%).
  • Trust was low: about 59% disagreed that AI chatbots are trustworthy enough for academic work, about 66% disagreed that their generated information is trustworthy enough, and about 83% disagreed they could use outputs with minimal editing.
  • Students reported higher trust and likely use for a locally maintained AI tool: 68.18% said they would find it generally trustworthy, compared with 58.56% who found broader generative AI tools generally untrustworthy; 67.2% said they would use it, versus the current 65%. Guidance was strongly desired: 74.3% wanted help incorporating AI chatbots into academic work, 72.8% wanted prompt-creation guidance, and 64.7% wanted information on how AI chatbots work.

Methods

  • A team of three library staff members and one student intern created and analyzed a survey of 360 undergraduate and graduate students at Harvard University, after IRB approval in August 2023 and a pilot survey. Recruitment used convenience sampling via listserv announcements and tabling at campus locations; the anonymous online survey ran from October 16 to November 6, 2023, and data were collected and analyzed in Qualtrics.
  • The survey included closed- and open-ended questions, a screening question, four demographic questions, and branching logic for three paths: no/never use, have used/will use, and unsure. All questions were optional.
  • Open-ended responses were inductively coded with thematic codes and interrater reliability meetings; descriptive statistics examined frequencies, and Qualtrics Crosstabs used chi-square tests to identify statistically significant differences between undergraduate and graduate students, with p < 0.05.

Limitations

  • The paper states that convenience sampling was used, that respondents were most likely to respond if they used or had strong feelings about generative AI, and that they are therefore not representative of the overall student population.
  • The paper notes that at publication the institution had initial guidelines but no official University-wide policy on student generative AI use, and that different institutions will have different adoption rates based on local factors.
  • The paper states the survey gives a general look at undergraduate and graduate students and that adoption and attitudes may vary by academic discipline and school; the text also makes evident that findings are self-reported, from a single institution, and based on optional survey questions.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.18438/eblip30512

Cites

26 of 26 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

2 Valid · 2 Not valid · 4 Not checked · 58 Own results and interpretations

37 more claims

Create a free account to see all 66 claims and every source verdict.

See all claims — free account
  • Probably not a substantial claim. 9
  • The quote only partly states this claim. 2
  • It may state two separate claims as one. 17

Create a free account to see these claims.

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

2 of 2

Citations

Cites

Cited by

See all 49

Metadata

Authors
Deschenes, A., & McMahon, M.
Affiliations
Harvard University (US)
Venue
Evidence Based Library and Information Practice
Published
2024-06-14
Volume
19
Issue
2
Pages
2-22
ISSN
1715-720X (ISSN-L)
OpenAlex
W4399665471
Cited by
49
References
24
Citations by year
2026: 19 · 2025: 29 · 2024: 1
FWCI
58.55
Type
journal article
Language
en
Open access
Locations
  • Evidence Based Library and Information Practice (publishedVersion, cc-by-nc-sa)
  • Érudit (Université de Montréal) (submittedVersion, cc-by-nc-sa)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Topics
  • Social Sciences › Library and Information Sciences › Library Science and Information Literacy
  • Social Sciences › Education › Artificial Intelligence in Education
  • Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
Keywords
generative AI, chatbots, academic research, student use of AI, trust in AI outputs, information seeking, research habits, librarian support, AI training for students, critical evaluation of AI outputs

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (qwen3.8-omni-flash).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.