Cited by 51Artificial Intelligence in EducationWriting and Handwriting EducationEducational Technology in Learning

Improving writing feedback for struggling writers: generative AI to the rescue?

Open access
Evmenova, Anya S., Regan, Kelley, Mergen, Reagan, Hrisseh, Roba·2024·TechTrends

Abstract

Generative AI has the potential to support teachers with writing instruction and feedback. The purpose of this study was to explore and compare feedback and data-based instructional suggestions from teachers and those generated by different AI tools. Essays from students with and without disabilities who struggled with writing and needed a technology-based writing intervention were analyzed. The essays were imported into two versions of ChatGPT using four different prompts, whereby eight sets of responses were generated. Inductive thematic analysis was used to explore the data sets. Findings indicated: (a) differences in responses between ChatGPT versions and prompts, (b) AI feedback on student writing did not reflect provided student characteristics (e.g., grade level or needs; disability; ELL status), and (c) ChatGPT’s responses to the essays aligned with teachers’ identified areas of needs and instructional decisions to some degree. Suggestions for increasing educator engagement with AI to enhance teaching writing is discussed.

Integrity signals

Registry

Crossref · 10.1007/s11528-024-00965-y

Cites

36 of 36 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Claims the paper makes

We couldn’t retrieve this paper’s full text, so its claims and self-check weren’t run. The registry and citation checks still apply.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • AI tools in academic writing fall into four functional categories: generative AI models, automated writing evaluation systems, translation tools, and multimodal platforms.

    “The tools identified fall into four functional categories: generative AI models (ChatGPT, Gemini, DeepSeek, Copilot) for higher-order tasks like brainstorming and drafting; automated writing evaluation systems (Grammarly, ProWritingAid, Hemingway Editor) for grammar and mechanical corrections; translation tools (DeepL, Google Translate) for cross-lingual support; and multimodal platforms (Canva, Q”

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Authors
Evmenova, A. S., Regan, K., Mergen, R., & Hrisseh, R.
Affiliations
George Mason University (US)
Corresponding author
Evmenova, Anya S.
Venue
TechTrends
Published
2024-05-14
Volume
68
Issue
4
Pages
790-802
ISSN
  • 1559-7075 (ISSN-L)
  • 8756-3894
OpenAlex
W4396897879
Cited by
51
References
36
Citations by year
2026: 28 · 2025: 20 · 2024: 3
FWCI
19.18
Type
journal article
Language
en
Open access
Locations
TechTrends (publishedVersion, cc-by)
Funders
U.S. Department of Education, Office of Special Education Programs (H327S180004)
Topics
  • Physical Sciences › Computer Science › Information Systems › Artificial Intelligence in Education
  • Social Sciences › Education › Writing and Handwriting Education
  • Physical Sciences › Computer Science › Information Systems › Educational Technology in Learning
Keywords
generative AI, writing feedback, struggling writers, ChatGPT, writing instruction, students with disabilities, English language learners, teacher feedback, writing intervention, prompting

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.

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