CoChat

Challenges and Opportunities of Generative AI for Higher Education as Explained by ChatGPT

Open access

The DOI matches its Crossref record; 42 of the 43 references it cites were found in a registry or an index, and 1 are not checked yet; 13 of the 50 sources its claims cite were checked for support.

Abstract

ChatGPT is revolutionizing the field of higher education by leveraging deep learning models to generate human-like content. However, its integration into academic settings raises concerns regarding academic integrity, plagiarism detection, and the potential impact on critical thinking skills. This article presents a study that adopts a thing ethnography approach to understand ChatGPT’s perspective on the challenges and opportunities it represents for higher education. The research explores the potential benefits and limitations of ChatGPT, as well as mitigation strategies for addressing the identified challenges. Findings emphasize the urgent need for clear policies, guidelines, and frameworks to responsibly integrate ChatGPT in higher education. It also highlights the need for empirical research to understand user experiences and perceptions. The findings provide insights that can guide future research efforts in understanding the implications of ChatGPT and similar Artificial Intelligence (AI) systems in higher education. The study concludes by highlighting the importance of thing ethnography as an innovative approach for engaging with intelligent AI systems and calls for further research to explore best practices and strategies in utilizing Generative AI for educational purposes.

Verified 2026-10-05 · 1 reference not checked yet

Overview

CoChat Summary

TL;DR

This paper applies “thing ethnography” to interview ChatGPT itself about the opportunities, challenges, barriers, and mitigation strategies for generative AI in higher education. It finds that ChatGPT’s responses largely align with existing literature while adding less-documented concerns about personalized learning and communication/collaboration, and it calls for clear policies, training, and empirical research.

Key findings

  • ChatGPT identified seven opportunities, of which five emphasize students: 24/7 support and accessibility, personalized learning and tutoring, supplemental learning resource, language learning and communication skills, and innovative learning experiences; the other two focus on instructors/teaching assistants and research/data analysis.
  • ChatGPT identified challenges including academic integrity/misuse, quality control, limits to personalized learning, lack of domain-specific expertise and authority, and weaker communication/collaboration than human instructors or peers.
  • Barriers to adoption included lack of awareness and understanding, technological barriers, resistance to change, ethical and privacy concerns, academic rigor and quality, resource constraints, legal and regulatory considerations, and lack of interdisciplinary collaboration.
  • Mitigation priorities included policy development, education and training, collaboration and interdisciplinary efforts, research and development, ethical review processes, and continuous monitoring and evaluation; the paper also cites that out of 142 HEIs surveyed in May 2023, only one explicitly prohibited the use of AI.

Methods

  • The study adopts a thing ethnography approach, treating ChatGPT as a subject rather than an object, and engages it through a semi-structured interview conducted as a written conversation.
  • An interview guide with themes and key questions was prepared, with probes used for clarification, completion, and elaboration.
  • Data were analyzed using Braun and Clarke’s six-phase thematic analysis; dependability was supported through verbatim quotes linked to codes, and transferability was addressed by comparing findings with existing literature. The full interview transcript is included in the paper, and no additional data were created.

Limitations

  • The paper states that the study focuses strongly on ChatGPT, so findings may not represent other GenAI chatbots such as Bard or Bing Chat; future work could compare different chatbots.
  • The paper states that replicability is unclear because ChatGPT may not provide the same answers, and improved versions may require replication; it also notes potential research bias during analysis.
  • The paper states that ethnographic studies are often immersive and lengthy, though thing ethnography does not necessarily require a longitudinal design; the authors present this as a first step in a broader study. The text also makes evident that the findings are based on ChatGPT’s self-report and literature comparison rather than empirical user outcomes, and the authors repeatedly call for more empirical research.

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

Integrity signals

Registry

Crossref · 10.3390/educsci13090856

Cites

42 of 43 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

9 Valid · 7 Not valid · 24 Not checked · 81 Own results and interpretations

5 more claims

Create a free account to see all 121 claims and to verify the 37 references still open.

See all claims — free account
  • Probably not a substantial claim. 108
  • It may state two separate claims as one. 52

Create a free account to see these claims.

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?

4 more claims about this paper

Create a free account to see every claim other papers make about this one.

See all claims — free account

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.

5 of 7 · the rest aren’t loaded

2 more findings

Create a free account to see every finding of this check.

See all findings — free account

Citations

Cites

Research works printed but not linked: 2

  1. Saunders, M., Lewis, P., and Thornhill, A. (2016). Research Methods for Business Students, Pearson Education. [8th ed.].
  2. Panke, S. (2023, January 10). Workshop 6: AI in Education: Course Design Institute. Proceedings of the EdMedia + Innovate Learning, Vienna, Austria. Available online: https://www.learntechlib.org/primary/p/222656/.

Cited by

See all 882

Metadata

Authors
Michel-Villarreal, R., Vilalta-Perdomo, E., Salinas-Navarro, D. E., Thierry-Aguilera, R., & Gerardou, F. S.
Affiliations
  • Sustainability Research Institute, School of Earth and Environment, University of Leeds, Leeds LS2 9JT, UK
  • Aston University (GB)
  • Tecnológico de Monterrey (MX)
  • Falmouth University (GB)
Venue
Education Sciences
Publisher
MDPI AG
Published
2023-08-23
Volume
13
Issue
9
Pages
856
ISSN
2227-7102 (ISSN-L)
OpenAlex
W4386098991
Cited by
877
References
38
Citations by year
2026: 246 · 2025: 351 · 2024: 264 · 2023: 15
FWCI
205.22
Type
journal article
Language
en
Open access
Locations
  • Education Sciences (publishedVersion, cc-by)
  • White Rose Research Online (University of Leeds, The University of Sheffield, University of York) (acceptedVersion)
  • Aston Publications Explorer (Aston University) (acceptedVersion)
  • Falmouth University Research Repository (FURR) (Falmouth University) (acceptedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
Writing Lab, Institute for the Future of Education, Tecnologico de Monterrey, Mexico
Topics
  • Social Sciences › Education › Artificial Intelligence in Education
  • Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
  • Social Sciences › Education › Academic integrity and plagiarism
Keywords
generative AI, ChatGPT, higher education, academic integrity, plagiarism detection, critical thinking, artificial intelligence in education, responsible AI adoption

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 2026-10-05, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-05 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (qwen3.8-omni-flash and 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.