Grading exams using large language models: A comparison between human and AI grading of exams in higher education using ChatGPT
Open accessThe DOI matches its Crossref record; 40 of the 40 references it cites were found in a registry or an index.
Abstract
This study compares how the generative AI (GenAI) large language model (LLM) ChatGPT performs in grading university exams compared to human teachers. Aspects investigated include consistency, large discrepancies and length of answer. Implications for higher education, including the role of teachers and ethics, are also discussed. Three Master's‐level exams were scored using ChatGPT 3.5, and the results were compared with the teachers' scoring and the grading teachers were interviewed. In total, 463 exam responses were graded. With each response being graded at least three times, a total of 1389 gradings were conducted. For the final exam scores, 70% of ChatGPT's gradings were within 10% of the teachers' gradings and 31% within 5%. ChatGPT tended to give marginally higher scores. The agreement on grades is 30%, but 45% of the exams received an adjacent grade. On individual questions, ChatGPT is more inclined to avoid very high or very low scores. ChatGPT struggles to correctly score questions closely related to the course lectures but performs better on more general questions. The AI can generate plausible scores on university exams that, at first glance, look similar to a human grader. There are differences but it is not unlikely that two different human graders could result in similar discrepancies. During the interviews, teachers expressed their surprise at how well ChatGPT's grading matched their own. Increased use of AI can lead to ethical challenges as exams are entrusted to a machine whose decision‐making criteria are not fully understood, especially concerning potential bias in training data.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Language Models are Few-Shot Learners
Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al. (2020) · arXiv · doi:10.48550/arxiv.2005.14165 · Cited by 23,372
I.—COMPUTING MACHINERY AND INTELLIGENCE
TURING, A. M. (1950) · Mind · doi:10.1093/mind/lix.236.433 · Cited by 10,290
Weighted kappa: nominal scale agreement provision for scaled disagreement or partial credit
Cohen, J. (1968) · Psychological Bulletin · doi:10.1037/h0026256 · Cited by 8,736
Measuring nominal scale agreement among many raters.
Fleiss, J. L. (1971) · Psychological Bulletin · doi:10.1037/h0031619 · Cited by 8,733
Zawacki-Richter, O., Marín, V. I., Bond, M., & Gouverneur, F. (2019) · International Journal of Educational Technology in Higher Education · doi:10.1186/s41239-019-0171-0 · Cited by 7,242
Kung, T. H., Cheatham, M., Medenilla, A., Sillos, C., De Leon, L., Elepaño, C., Madriaga, M., Aggabao, R., Diaz-Candido, G., Maningo, J., et al. (2023) · PLOS Digital Health · doi:10.1371/journal.pdig.0000198 · Cited by 4,115
Kung, T. H., Cheatham, M., ChatGPT, Medenilla, A., Sillos, C., De Leon, L., Elepaño, C., Madriaga, M., Aggabao, R., Diaz-Candido, G., et al. (2022) · doi:10.1101/2022.12.19.22283643 · Cited by 4,115
Students' voices on generative ai: perceptions, benefits, and challenges in higher education
Chan, C. K. Y., & Hu, W. (2023) · arXiv · doi:10.48550/arxiv.2305.00290 · Cited by 2,504
Ray, P. P. (2023) · Internet of Things and Cyber-Physical Systems · doi:10.1016/j.iotcps.2023.04.003 · Cited by 2,463
Students’ voices on generative AI: perceptions, benefits, and challenges in higher education
Chan, C. K. Y., & Hu, W. (2023) · International Journal of Educational Technology in Higher Education · doi:10.1186/s41239-023-00411-8 · Cited by 2,301
Cited by
Does ChatGPT enhance student learning? A systematic review and meta-analysis of experimental studies
Deng, R., Jiang, M., Yu, X., Lu, Y., & Liu, S. (2025) · Computers & Education · doi:10.1016/j.compedu.2024.105224 · Cited by 531
Leong, Y. H., Nambiar, L., Tay, V. Y. J., Lie, S. A., & Yuhe, K. (2025) · JMIR Formative Research · doi:10.2196/77580 · Cited by 5
Fedoruk, B. D. (2025) · Journal of Educational Informatics · doi:10.51357/jei.v6i1.314 · Cited by 3
Metadata
- Authors
- Flodén, J.
- Venue
- British Educational Research Journal
- Publisher
- Wiley
- Published
- 2024-09-16
- Volume
- 51
- Issue
- 1
- Pages
- 201-224
- ISSN
- 0141-1926 (ISSN-L)
- 1469-3518
- OpenAlex
- W4402819599
- Cited by
- 82
- References
- 40
- FWCI
- 31.38
- Type
- journal article
- Language
- en
- Open access
- hybrid
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
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.