CoChat

Participatory Co-Design and Evaluation of a Novel Approach to Generative AI-Integrated Coursework Assessment in Higher Education

Open access

The DOI matches its Crossref record; 25 of the 25 references it cites were found in a registry or an index; 15 of the 34 sources its claims cite were checked for support.

Abstract

Generative AI tools offer opportunities for enhancing learning and assessment, but raise concerns about equity, academic integrity, and the ability to critically engage with AI-generated content. This study explores these issues within a psychology-oriented postgraduate programme at a UK university. We co-designed and evaluated a novel AI-integrated assessment aimed at improving critical AI literacy among students and teaching staff (pre-registration: osf.io/jqpce). Students were randomly allocated to two groups: the ‘compliant’ group used AI tools to assist with writing a blog and critically reflected on the outputs, while the ‘unrestricted’ group had free rein to use AI to produce the assessment. Teaching staff, blinded to group allocation, marked the blogs using an adapted rubric. Focus groups, interviews, and workshops were conducted to assess the feasibility, acceptability, and perceived integrity of the approach. Findings suggest that, when carefully scaffolded, integrating AI into assessments can promote both technical fluency and ethical reflection. A key contribution of this study is its participatory co-design and evaluation method, which was effective and transferable, and is presented as a practical toolkit for educators. This approach supports growing calls for authentic assessment that mirrors real-world tasks, while highlighting the ongoing need to balance academic integrity with skill development.

Overview

CoChat Summary

TL;DR

This study co-designed and evaluated an AI-integrated postgraduate assessment in which students used generative AI to draft and critique a blog, with random allocation to “compliant” or “unrestricted” AI-use groups. Findings suggest the approach was feasible and supported critical AI literacy, but markers could not reliably distinguish compliant from unrestricted AI-generated submissions.

Key findings

  • Students and teaching staff generally found the co-designed AI-integrated assessment feasible and acceptable, with students valuing critical thinking development but reporting that AI did not reduce workload because outputs required substantial checking and refinement.
  • Markers correctly identified 6 out of 14 students in the compliant group (42.9%) versus 3 out of 6 in the unrestricted group (50.0%); Fisher’s exact test gave an odds ratio of 0.75, p = 1.00, indicating no meaningful difference in marker accuracy.
  • Assessment marks ranged from 35 (fail) to 78 (distinction); most markers’ scores fell within a ten-point range, but one assessment ranged from 58 to 78.
  • Teaching staff initially found marking time-intensive due to evaluating both AI and student contributions, though efficiency improved with familiarity; most endorsed transparent, pedagogically grounded AI use despite integrity concerns.

Methods

  • Participatory co-design and evaluation within a psychology-oriented postgraduate programme at a UK university (King’s College London), using the 2023–24 cohort of approximately 30 students; 12 participants took part: 8 students and 4 teaching staff.
  • The adapted assessment required two AI-generated blog drafts, a final blog combining AI outputs with student revisions, and a critical commentary on AI-generated content; students were randomly allocated to a “compliant” group following the brief or an “unrestricted” group free to use AI by any means, and teaching staff marked blinded using an adapted rubric.
  • Qualitative data came from semi-structured focus groups and interviews (45–60 min), workshop feedback, and co-design workshops using think-aloud methods; analysis used Braun and Clarke’s thematic analysis separately for students and staff, plus Fisher’s exact test for marker accuracy.

Limitations

  • The paper states that students were not involved in the initial design phase, falling short of authentic co-production; qualitative feedback was collected after the initial pilot rather than a full module-wide rollout; and the single-institution, small-cohort, small-sample design limits generalisability.
  • The paper also notes limitations of participatory approaches generally, including potential power imbalances, risks of tokenism, and over-reliance on stakeholder input at the expense of expert judgement.
  • Evident from the text: the very small sample and short-term pilot limit statistical inference and long-term conclusions; reported group counts for marker identification (14 compliant and 6 unrestricted) do not align clearly with the stated 8 student participants, suggesting possible reporting or allocation ambiguity.

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

Integrity signals

Registry

Crossref · 10.3390/bs15060808

Cites

25 of 25 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

11 Valid · 1 Not valid · 11 Not checked · 66 Own results and interpretations

24 more claims

Create a free account to see all 89 claims and to verify the 1 reference still open.

See all claims — free account
  • Probably not a substantial claim. 40
  • It may state two separate claims as one. 28

Create a free account to see these claims.

Citations

Cites

Research works printed but not linked: 3

  1. Cook-Sather, A., Bovill, C., and Felten, P. (2014). Engaging students as partners in learning and teaching: A guide for faculty, John Wiley & Sons.
  2. Healey, 2019, Student engagement through partnership: A guide and update to the AdvanceHE framework, Advance HE, 12, 1
  3. Someren, M. W. v., Barnard, Y. F., and Sandberg, J. (1994). The think aloud method: A practical guide to modelling cognitive processes, Academic Press.

Cited by

See all 16

Metadata

Authors
Martin, A. F., Tubaltseva, S., Harrison, A., & Rubin, G. J.
Affiliations
  • King's College London (GB)
  • School of Liberal Arts, Richmond American University London, London W4 5AN, UK
Corresponding author
Martin, Alex F.
Venue
Behavioral Sciences
Publisher
MDPI AG
Published
2025-06-12
Volume
15
Issue
6
Pages
808
ISSN
2076-328X (ISSN-L)
OpenAlex
W4411284328
Cited by
16
References
25
Citations by year
2026: 15 · 2025: 2
FWCI
67.38
Type
journal article
Language
en
Open access
Locations
  • Behavioral Sciences (publishedVersion, cc-by)
  • Preprints.org (acceptedVersion, cc-by)
  • PubMed (publishedVersion)
  • PubMed Central (submittedVersion, cc-by)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • Research Portal (King's College London) (submittedVersion, cc-by)
Funders
  • Institute of Psychiatry, Psychology and Neuroscience at King’s College London’s Strategic College Teaching Fund (NIHR200890, NIHR207394)
  • National Institute for Health and Care Research (NIHR200890, NIHR207394)
Topics
  • Social Sciences › Education › Higher Education Practises and Engagement
  • Social Sciences › Education › Innovative Education and Learning Practices
  • Social Sciences › Education › Higher Education Learning Practices
Keywords
generative AI, coursework assessment, academic integrity, critical AI literacy, participatory co-design, authentic assessment, AI-assisted writing, ethical reflection

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 (gpt-6-sol).

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.