CoChat

Journalism Students’ Use, Expectations, and Understanding of Generative AI Tools

Open access
Veenstra, Aaron S., Wilder, Marcy, Shimu, Farhin, Schlauder, Luke, Dousdebés, Francesca · 2025 · Journalism & Mass Communication Educator

The DOI matches its Crossref record; 21 of the 21 references it cites were found in a registry or an index; none of the 19 sources its claims cite is checked for support yet.

Abstract

This study investigates how journalism students perceive and use large language models (LLMs), such as ChatGPT, their understanding of the technology, and their ethical considerations. Findings indicate that while students see LLMs as inevitable tools in their future careers, they often misunderstand how these tools generate content, in part because they lack more fundamental knowledge about digital tools. The paper emphasizes the need for tech literacy in journalism programs among both students and faculty.

Overview

CoChat Summary

TL;DR

In interviews with 12 undergraduate journalism majors at one southeastern U.S. research university, students saw generative AI/LLMs as inevitable in their future careers but largely misunderstood how the tools produce content, often describing them as a front-end to Google. They agreed that copying generated text as one’s own is cheating but had less consensus on ethical support uses, and most expected to use LLMs professionally while worrying about job replacement.

Key findings

  • None of the interviewees knew how LLM output was actually produced; the most common belief was that LLMs act as a front-end for Google or pull from the internet. Participant 10 said “it’s pulling it from somewhere on the internet,” and Participant 11 called it “unbiased . . . because it gathers data that’s in the internet already.”
  • Usage split into minor fun/entertainment uses and more extensive serious uses. Serious uses centered on idea generation or writer’s block; some also used ChatGPT professionally for a slogan, Instagram captions, or source questions. Perceptions of others’ use ranged from knowing nobody who uses it to “every student,” with most seeing it as at least fairly common; three called others’ use “lazy,” and another four said others “just don’t want to” or “just don’t care.”
  • Students were unanimous that copying and pasting wholly generated material into a class submission is cheating, but had less consensus on support-tool use; many endorsed “generating ideas” as ethical. Classroom exposure was inconsistent: some professors never mentioned LLMs, one syllabus simply disallowed them, and others both encouraged and discouraged use.
  • Most participants expected to use LLMs in their careers, and some had already done so professionally, but they generally viewed this negatively. Eight of the 12 interviewees mentioned concern that LLMs could replace jobs; Participant 1 worried news writing jobs “might be completely replaced with AI” and reoriented toward a PR career path, while Participant 5 thought journalism’s need for one’s own ideas made it less replaceable.

Methods

  • Qualitative design using in-depth, semi-structured interviews with 12 undergraduate journalism majors at a large research university in the southeastern United States. Interviews were conducted in November and December 2023 by graduate students enrolled in a communication research methods course and were recorded for transcription and analysis.
  • Sample: 6 women and 6 men, ages 18 to 27; 3 participants were Black, 9 were white, and 5 were Hispanic (of any race). All were journalism majors in upper-division undergraduate courses; four had a minor or double major.
  • Interviews were designed around five research questions covering students’ use and perceptions of others’ use, knowledge of how the tools work, ethical assessments, what they hear in classes, and career expectations. The provided text does not name a specific coding or analytic technique.

Limitations

  • Stated in the paper: “our study is limited by its focus on students from one program and cannot be generalized.”
  • Evident from the text: the study relies on self-reported interviews with only 12 students from a single program, so it cannot support claims about journalism students generally; the paper itself notes the one-program limitation.
  • Evident from the text: the provided text does not describe a formal coding scheme, reliability check, or detailed analytic procedure for the interview data.

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

Integrity signals

Registry

Crossref · 10.1177/10776958241296474

Cites

21 of 21 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

0 Valid · 0 Not valid · 16 Not checked · 62 Own results and interpretations

30 more claims

Create a free account to see all 78 claims and to verify the 7 references still open.

See all claims — free account
  • Probably not a substantial claim. 26
  • It may state two separate claims as one. 28
  • The quote only partly states this claim. 3

Create a free account to see these claims.

Citations

Cites

Research works printed but not linked: 4

  1. Bright, A. (2018). A qualitative look at journalism programs in flux: The role of faculty in the movement toward a digital curriculum. Teaching Journalism & Mass Communication, 8(2), 1-10.
  2. Demmar, K., & Neff, T. (2023). Generative AI in journalism education: Mapping the state of an emerging space of concerns, opportunities, and strategies. Journalism Education, 12(1), 47-58.
  3. Simon, F. M. (2024). Artificial intelligence in the news: How AI retools, rationalizes, and reshapes journalism and the public Arena. Columbia Journalism Review. https://www.cjr. org/tow_center_reports/artificial-intelligence-in-the-news.php
  4. Aaron S. Veenstra (PhD, University of Wisconsin-Madison, 2009) is an associate professor and the journalism area coordinator in Florida Atlantic University's School of Communication and Multimedia Studies whose research focuses on political communication, digital media, and social identity. His co-edited volume, The Press and Democratic Backsliding: How Journalism Has Failed the Public and How It Can Revive Democracy, was published by Lexington Books in 2024.

Cited by

See all 24

Metadata

Authors
Veenstra, A. S., Wilder, M., Shimu, F., Schlauder, L., & Dousdebés, F.
Affiliations
Florida Atlantic University (US)
Corresponding author
Veenstra, Aaron S.
Venue
Journalism & Mass Communication Educator
Published
2024-11-14
Volume
80
Issue
1
Pages
115-128
ISSN
  • 0022-5517 (ISSN-L)
  • 1077-6958
  • 2161-4326
OpenAlex
W4404387482
Cited by
24
References
21
Citations by year
2026: 18 · 2025: 6
FWCI
8.95
Type
journal article
Language
en
Open access
Licence
journals.sagepub.com/page/policies/text-and-data-mining-license (tdm)
Locations
Journalism & Mass Communication Educator (publishedVersion)
Topics
  • Social Sciences › Communication › Journalism and Media Studies
  • Social Sciences › Education › Artificial Intelligence in Education
  • Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
Keywords
generative AI, large language models, journalism students, ChatGPT, technological literacy, AI ethics, student perceptions of AI, journalism education, AI content generation

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.

A language model drew the claims from the paper on 2026-11-20, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-11-20 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

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.