The DOI matches its Crossref record; 68 of the 68 references it cites were found in a registry or an index; 4 of the 19 sources its claims cite were checked for support.
Incorporating generative AI into a writing-intensive undergraduate course without off-loading learning
Open accessAbstract
As generative AI becomes ubiquitous, writers must decide if, when, and how to incorporate generative AI into their writing process. Educators must sort through their role in preparing students to make these decisions in a quickly evolving technological landscape. We created an AI-enabled writing tool that provides scaffolded use of a large language model as part of a research study on integrating generative AI into an upper division STEM writing-intensive course. Drawing on decades of research on integrating digital tools into instruction and writing research, we discuss the framework that drove our initial design considerations and instructional resources. We then share our findings from a year of design-based implementation research during the 2023–2024 academic year. Our original instruction framework identified the need for students to understand, access, prompt, corroborate, and incorporate the generative AI use effectively. In this paper, we explain the need for students to think first, before using AI, move through good enough prompting to agentic iterative prompting, and reflect on their use at the end. We also provide emerging best practices for instructors, beginning with identifying learning objectives, determining the appropriate AI role, revising the content, reflecting on the revised curriculum, and reintroducing learning as needed. We end with an indication of our future directions.
Overview
CoChat Summary
TL;DR
This paper reports first-year design-based implementation research integrating a scaffolded LLM writing tool, PapyrusAI, into an upper-division engineering writing-intensive course. It identifies emerging student practices—iterative prompting, corroboration, thinking first, and reflection—and instructor practices for infusing AI while keeping learning objectives central.
Key findings
- For students, the most promising uses involved prompting iteratively from a “good enough” prompt, “talking back” to interrogate AI output, corroborating accuracy, thinking first before using AI, and reflecting on AI use; one-and-done prompting “leads to little if any learning.”
- For instructors, emerging practices included identifying student learning objectives, using backward design, determining the appropriate AI role, revising or reducing content to make time, reflecting on the revised curriculum, and reintroducing learning as needed; AI should not overtake course goals.
- Implementation occurred in Engineering 190W at UC Irvine during the 2023–2024 academic year across 3 quarters: Fall 2023 used ChatGPT’s free version, Winter 2024 piloted the MVP with GPT-4.0, and Spring 2024 added 3 instructors, each teaching at least two sections of approximately 20 students; the course requires 4,000 words or more of finished work.
- Concerns emerged that weaker writers may benefit less: one instructor said “poor writers [are] poor writers with prompts,” and the team noted the risk of the “Matthew Effect” widening achievement gaps; one instructor reported about 50% of students thought the tool was useful after brainstorming with it once.
Methods
- Design-based implementation research during the 2023–2024 academic year in an upper-division STEM writing-intensive course, using qualitative methods borrowed from phenomenological and design-based practices.
- Data included class observations, instructor interviews, student focus interviews, student writing artifacts, instructional artifacts, and student surveys; participants included the lead instructor across 3 quarters and, in Spring 2024, 3 additional instructors teaching at least two sections of approximately 20 students each.
- Analysis used flexible coding, memo-writing, coding by the first author with checks from the second and fourth authors, participant checks with the instructional team, and searches for disconfirming evidence.
Limitations
- Stated: the study is limited to a single institution and single course, so it is not generalizable to other institutions, different student demographics, or different content areas; instructors had taught the course for multiple years, and new teachers would face additional burdens; the article focuses on the instructor side, with student voices planned for future work.
- Evident: the qualitative design and small instructor sample limit broad claims; the paper reports no student writing-quality outcomes, and the authors note that bias-interrogation materials were underdeveloped and that it remains unclear whether the tool’s friction is productive effort or unnecessary friction.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
4 Valid · 3 Not valid · 11 Not checked · 75 Own results and interpretations
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“All of our instructors in some way have the students reflect at key points in the writing process on the role of AI and its impact on both the writing process and product.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Instructors also noted that the use of AI sometimes disrupted the reflection that normally occurs while writing.”
hedgedstates a possible cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Nonetheless, some students still did not seem to benefit from the AI as significantly as we had expected.”
hedged- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Instructors are also learning that generative AI can be especially helpful in teaching students certain aspects of writing that have long been challenging, such as audience awareness and perspective taking.”
plainstates a possible cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“One thing our participating instructors reported noticing is that the activity of reviewing their course through the lens of where and how to incorporate generative AI caused them to externalize and be explicit about some of the learning objectives and curriculum that had previously been hidden.”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
15 more claims
Create a free account to see all 93 claims and every source verdict.
- It may state two separate claims as one. 31
- Probably not a substantial claim. 59
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ References · p. 15
“In: Proceedings of the 2022 ACM conference on fairness, accountability, and transparency; 2002. pp.173-184.”
§ References · p. 15
“Proceedings of the 2022 ACM conference on fairness, accountability, and transparency”
Citations
Cites
Basics of Qualitative Research (3rd ed.): Techniques and Procedures for Developing Grounded Theory
Corbin, J., & Strauss, A. (2008) · SAGE · doi:10.4135/9781452230153 · Cited by 12,136
Does Active Learning Work? A Review of the Research
Prince, M. (2004) · Journal of Engineering Education · doi:10.1002/j.2168-9830.2004.tb00809.x · Cited by 7,334
The Matthew Effect in Science: The reward and communication systems of science are considered.
Merton, R. K. (1968) · Science · doi:10.1126/science.159.3810.56 · Cited by 6,720
What is AI Literacy? Competencies and Design Considerations
Long, D., & Magerko, B. (2020) · Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3313831.3376727 · Cited by 3,241
MacArthur, C. A., Graham, S. W., & Fitzgerald, J. (2005) · Cited by 2,327
Conceptualizing AI literacy: An exploratory review
Ng, D. T. K., Leung, J. K. L., Chu, S. K. W., & Qiao, M. S. (2021) · Computers and Education: Artificial Intelligence · doi:10.1016/j.caeai.2021.100041 · Cited by 2,205
Technology and Social Inclusion: Rethinking the Digital Divide
Warschauer, M. (2003) · doi:10.7551/mitpress/6699.001.0001 · Cited by 1,825
Self-citationFlexible Coding of In-depth Interviews: A Twenty-first-century Approach
Deterding, N. M., & Waters, M. C. (2021) · Sociological Methods & Research · doi:10.1177/0049124118799377 · Cited by 1,626 · cited 1×
“Coding was based on the flexible coding methods of Deterding & Waters [17], which are optimized for qualitative data analysis software and large amounts of data.”
Voices of the Mind: A Sociocultural Approach to Mediated Action
Baker-Sennett, J., Rogoff, B., Bell, N., & Wertsch, J. V. (1992) · The American Journal of Psychology · doi:10.2307/1423207 · Cited by 1,505 · cited 1×
“Writing is culturally situated and mediated by tools [15, 61].”
The Power of Feedback Revisited: A Meta-Analysis of Educational Feedback Research
Wisniewski, B., Zierer, K., & Hattie, J. (2020) · Frontiers in Psychology · doi:10.3389/fpsyg.2019.03087 · Cited by 1,199 · cited 1×
“The embedded prompts are focused on pedagogically effective interactions, based on tutoring [25], writing [20, 21, 64], and HCI research.”
Research works printed but not linked: 1
- Weisz, Justin D., Jessica He, Michael Muller, Gabriela Hoefer, Rachel Miles, and Werner Geyer. “Design Principles for Generative AI Applications.” In Proceedings of the CHI Conference on Human Factors in Computing Systems, 1–22. Honolulu HI USA: ACM, 2024. https://doi.org/10.1145/3613904.3642466. · DOI · Paper page
Cited by
Agentic AI in Education: State of the Art and Future Directions
Kostopoulos, G., Gkamas, V., Rigou, M., & Kotsiantis, S. (2025) · IEEE Access · doi:10.1109/access.2025.3620473 · Cited by 69
Qiao, M. S., Gu, M. M., & Lu, C. (2025) · International Journal of Applied Linguistics · doi:10.1111/ijal.70034 · Cited by 37
Vairagade, V. S. (2025) · Discover Computing · doi:10.1007/s10791-025-09868-9 · Cited by 12
Hou, Y. (2025) · Higher Education · doi:10.1007/s10734-025-01603-z · Cited by 24
From Wandering to Collaboration: Discourse Patterns in Middle School Generative AI Use
Ritchie, D., Nixon, N., Tate, T. P., & Warschauer, M. (2026) · International Conference on Learning Analytics and Knowledge (LAK) · doi:10.1145/3785022.3785094 · Cited by 2
Self-citation
Metadata
- Authors
- Tate, T. P., Harnick-Shapiro, B., Ritchie, D. R., Tseng, W., Dennin, M., & Warschauer, M.
- Affiliations
- University of California, Irvine (US)
- Corresponding author
- Tate, Tamara P.
- Venue
- Discover Computing
- Published
- 2025-05-08
- Volume
- 28
- Issue
- 1
- ISSN
- 2948-2992 (ISSN-L)
- OpenAlex
- W4410189969
- Cited by
- 14
- References
- 63
- Citations by year
- 2026: 9 · 2025: 5
- FWCI
- 14.56
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- cc-by
- creativecommons.org/licenses/by/4.0 (tdm)
- creativecommons.org/licenses/by/4.0 (vor)
- Locations
- Discover Computing (publishedVersion, cc-by)
- eScholarship (California Digital Library) (submittedVersion, cc-by-nc)
- Funders
- National Science Foundation (23152984)
- Topics
- Social Sciences › Education › Artificial Intelligence in Education
- Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
- Social Sciences › Education › Educational Technology in Learning
- Keywords
- generative AI, large language models, writing-intensive courses, scaffolding, prompting, iterative prompting, student reflection, instructor role, curriculum development, writing process
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).
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
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Incorporating generative AI into a writing-intensive undergraduate course without off-loading learning. https://app.cochat.ai/papers/10.1007/s10791-025-09563-9
- The paper
- Tate, T. P., Harnick-Shapiro, B., Ritchie, D. R., Tseng, W., Dennin, M., & Warschauer, M. (2025). Incorporating generative AI into a writing-intensive undergraduate course without off-loading learning. Discover Computing. https://doi.org/10.1007/s10791-025-09563-9
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.