Is it harmful or helpful? Examining the causes and consequences of generative AI usage among university students
Open accessThe DOI matches its Crossref record; 69 of the 69 references it cites were found in a registry or an index; 16 of the 43 sources its claims cite were checked for support.
Abstract
While the discussion on generative artificial intelligence, such as ChatGPT, is making waves in academia and the popular press, there is a need for more insight into the use of ChatGPT among students and the potential harmful or beneficial consequences associated with its usage. Using samples from two studies, the current research examined the causes and consequences of ChatGPT usage among university students. Study 1 developed and validated an eight-item scale to measure ChatGPT usage by conducting a survey among university students (N = 165). Study 2 used a three-wave time-lagged design to collect data from university students (N = 494) to further validate the scale and test the study’s hypotheses. Study 2 also examined the effects of academic workload, academic time pressure, sensitivity to rewards, and sensitivity to quality on ChatGPT usage. Study 2 further examined the effects of ChatGPT usage on students’ levels of procrastination, memory loss, and academic performance. Study 1 provided evidence for the validity and reliability of the ChatGPT usage scale. Furthermore, study 2 revealed that when students faced higher academic workload and time pressure, they were more likely to use ChatGPT. In contrast, students who were sensitive to rewards were less likely to use ChatGPT. Not surprisingly, use of ChatGPT was likely to develop tendencies for procrastination and memory loss and dampen the students’ academic performance. Finally, academic workload, time pressure, and sensitivity to rewards had indirect effects on students’ outcomes through ChatGPT usage.
Overview
CoChat Summary
TL;DR
This two-study paper develops and validates an 8-item ChatGPT usage scale and tests antecedents and outcomes of ChatGPT use among university students. It finds that academic workload and time pressure increase ChatGPT use, reward sensitivity decreases it, and greater ChatGPT use is associated with more procrastination, more memory loss, and lower academic performance.
Key findings
- Study 1 produced an 8-item ChatGPT usage scale with one factor explaining 62.65% of variance; Cronbach’s alpha = 0.914, composite reliability = 0.928, AVE = 0.618.
- In Study 2, workload (β = 0.133, p < 0.01) and time pressure (β = 0.163, p < 0.001) were positively related to ChatGPT use; sensitivity to rewards was negatively related (β = −0.102, p < 0.10), while sensitivity to quality was not significant (β = 0.033, n.s.).
- ChatGPT use was positively related to procrastination (β = 0.309, p < 0.001) and memory loss (β = 0.274, p < 0.001), and negatively related to academic performance/CGPA (β = −0.104, p < 0.05).
- Indirect effects through ChatGPT use were supported for workload and time pressure on procrastination, memory loss, and academic performance; reward sensitivity had negative indirect effects on procrastination and memory loss but not academic performance; quality sensitivity indirect effects were all insignificant.
Methods
- Study 1: scale development and validation with N = 165 university students in Pakistan; initial 12 items reduced to 10 by expert review, then EFA dropped 2 items, yielding an 8-item scale.
- Study 2: three-wave time-lagged survey with N = 494 university students; predictors measured at wave 1, ChatGPT use at wave 2, and procrastination, memory loss, and CGPA at wave 3.
- Analysis: partial least squares structural equation modeling (PLS-SEM) with bootstrapping (5,000 samples) to assess measurement and structural models.
Limitations
- Stated: the time-lagged design could not completely rule out reciprocal relationships, such as ChatGPT use reducing later perceptions of workload; future research should use longitudinal designs and examine personality factors, trust propensity, Big Five traits, contextual moderators, benefits of generative AI, field/gender differences, and learning/health outcomes.
- Stated: the insignificant effect of quality consciousness warrants further investigation.
- Evident from the text: the sample is university students in Pakistan, mostly from public-sector institutions and business studies, so generalizability is limited; most measures are self-report, and sensitivity to rewards and sensitivity to quality were measured with only 2 items each.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
11 Valid · 1 Not valid · 22 Not checked · 53 Own results and interpretations
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“Active learning, which involves active cognitive engagement with the content, is crucial for memory consolidation and retention (Cowan et al., 2021).”
What the sources say
- Could not verifycites Cowan et al., 2021 sourceExperimental — accuracy not yet measured
Our copy of Cowan et al. (2021) is a neuroscience review arguing that ACTIVE POST-ENCODING PROCESSES (cellular late-LTP and systems-level hippocampal-cortical replay/coupling) selectively stabilize GOAL-RELEVANT or salient information (Abstract; Introduction; 'Behavioral evidence for adaptive consolidation: Prioritization'; 'An integrative framework for adaptive consolidation'). It never addresses 'active learning' in the sense of the citing sentence - active cognitive engagement by the learner with study content as a condition that improves consolidation and retention. Searches for active learning, engagement with material, depth of processing, generation/effortful encoding, retrieval practice and testing effect returned only passages about salience tags, reward/emotion-driven prioritization, schema effects of prior knowledge on later encoding, and reference-list entries (e.g. Schmidt 2006 on encoding difficulty and sleep spindles), none of which state the attributed claim. The likely basis for the citation is the lexical match on 'active processes during post-encoding consolidation support the selective stabilization... into adaptive, long-term memories' (Abstract), which refers to neural consolidation mechanisms, not to how learners engage with content. (LLM, conf=0.72)
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.
- Could not verify
Every source the paper cites for this claim was checked and supports it.
In the paper
“As shown in Table 4, all of the HTMT values were below the threshold, thereby establishing discriminant validity among the study's constructs.”
What the sources say
- Supportedcites Henseler et al., 2015 source
In the section explaining HTMT as a criterion, Henseler et al. state that values are compared with a predefined threshold and that values above it indicate a lack of discriminant validity. This supports the citing sentence’s methodological interpretation of its Table 4 results; the table values themselves are the citing study’s findings. (LLM, conf=0.88)
plain- 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.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“In fact, majority of the prior studies consist of theoretical discussions, commentaries, interviews, reviews, or editorials on the use of ChatGPT in academia (e.g., Cooper, 2023; Cotton et al., 2023; Dwivedi et al., 2023; King, 2023; Peters et al., 2023).”
What the sources say
- Could not verifycites Cooper, 2023 sourceExperimental — accuracy not yet measured
The ChatGPT section says there was no prior research examining ChatGPT-generated text on science education topics and little written about its use as a research tool in STEM education. It does not characterize the majority of prior studies on ChatGPT in academia by publication type. Searches for that characterization found no supporting passage in our copy. (LLM, conf=0.72)
- Supportedcites Cotton et al., 2023 source
The Abstract describes a paper that examines and discusses ChatGPT’s opportunities and challenges in higher education. The Discussion explains that the authors used ChatGPT to draft earlier sections and then reflect on that approach. Cotton et al. is therefore an example of the discussion-oriented work listed in the citing sentence; the citation is not, by itself, evidence for the word “majority.” (LLM, conf=0.88)
- Supportedcites Dwivedi et al., 2023 source
The citation is one of several examples offered for the types of prior publications, rather than evidence for the numerical claim that they constitute a majority. The paper identifies itself as an opinion paper (title, p. 3), and Contribution 10 (p. 14) explicitly describes itself as opinion, without a formal literature review or empirical analysis. (LLM, conf=0.88)
plain- 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.
- Could not verify
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
“workload had a negative indirect effect on academic performance (indirect effect = -0.014, t = 1.657, p < 0.10) through the use of ChatGPT.”
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.
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
“Study 1 developed and validated an eight-item scale to measure ChatGPT usage by conducting a survey among university students (N = 165).”
plainn = 165- 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.
52 more claims
Create a free account to see all 87 claims and every source verdict.
- It may state two separate claims as one. 8
- Probably not a substantial claim. 24
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?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Lastly, predictive modelling of GenAI learning outcomes in the literature is dominated by regression and structural equation approaches [1, 52, 67]”
From Multinational machine learning investigation of whether generative AI interactions augment or substitute cognitive skills and study efficiency in global higher education (2026) · cites this paper as [1]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Over-reliance on these tools detracts from students' ability to hone their critical and reflective skills [1].”
From Evaluating the Effectiveness and Ethical Implications of AI Detection Tools in Higher Education (2025) · cites this paper as [1]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Abbas et al. (2024) found in a quantitative analysis of a general population of students in Pakistan that the usage of ChatGPT could increase the tendency to procrastinate, memory loss, and reduce academic performance.”
From Exploring the role of generative AI in higher education: Semi-structured interviews with students with disabilities (2025) · cites this paper as Abbas et al. (2024)
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“evidence already exists that time pressures of university workloads are associated with greater AI use (Abbas, Jam, and Khan 2024).”
From Summarise, elaborate, try again: exploring the effect of feedback literacy on AI-enhanced essay writing (2026) · cites this paper as Abbas, Jam, and Khan 2024
The cited source says what the claim attributes to it.
In the citing paper
“The rapid integration of generative artificial intelligence (GenAI) into higher education has precipitated a critical tension between the pursuit of efficiency and the imperative for deep, transformative learning.”
From Pedagogical partnerships with generative AI in higher education: how dual cognitive pathways paradoxically enable transformative learning (2026) · cites this paper as Abbas et al., 2024
3 more claims about this paper
Create a free account to see every claim other papers make about this one.
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
§ Measurement model · p. 12
“the standardized factor loadings for each item of each measure were above the threshold level of 0.70”
§ Measures · p. 11
“| I am often in hurry when it comes to meeting academic deadlines | 0.511 | | | |”
Citations
Cites
A new criterion for assessing discriminant validity in variance-based structural equation modeling
Henseler, J., Ringle, C. M., & Sarstedt, M. (2015) · Journal of the Academy of Marketing Science · doi:10.1007/s11747-014-0403-8 · Cited by 38,459 · cited 2×
“HTMT values below 0.85 are considered good to establish discriminant validity (Henseler et al., 2015).”
BackgroundMethodInfluentialWhen to Use and How to Report the Results of PLS-SEM
Hair, J. F., Risher, J. J., Sarstedt, M., & Ringle, C. M. (2019) · European Business Review · doi:10.1108/ebr-11-2018-0203 · Cited by 27,541 · cited 6×
“Furthermore, in order to test multicollinearity, we calculated variance inflation factor (VIF), which should be less than 5 to rule out the possibility of multicollinearity among the constructs (Hair et al., 2019).”
BackgroundMethodInfluentialA primer on partial least squares structural equation modeling (PLS-SEM)
Leguina, A. (2015) · International Journal of Research & Method in Education · doi:10.1080/1743727x.2015.1005806 · Cited by 18,921
(2017) · Cited by 16,319
Sources of method bias in social science research and recommendations on how to control it
Podsakoff, P. M., MacKenzie, S. B., & Podsakoff, N. P. (2012) · Annual Review of Psychology · doi:10.1146/annurev-psych-120710-100452 · Cited by 15,744
Structural equation models with unobservable variables and measurement error: Algebra and statistics
Fornell, C., & Larcker, D. F. (1981) · Journal of Marketing Research · doi:10.1177/002224378101800313 · Cited by 13,287
(2014) · DigitalCommons - Kennesaw State University (Kennesaw State University) · Cited by 12,785
Dwivedi, Y. K., Kshetri, N., Hughes, L., Slade, E. L., Jeyaraj, A., Kar, A. K., Baabdullah, A. M., Koohang, A., Raghavan, V., Ahuja, M., et al. (2023) · International Journal of Information Management · doi:10.1016/j.ijinfomgt.2023.102642 · Cited by 4,584
A Brief Tutorial on the Development of Measures for Use in Survey Questionnaires
Hinkin, T. R. (1998) · Organizational Research Methods · doi:10.1177/109442819800100106 · Cited by 4,095 · cited 3×
“Item generation: We used scale development procedures proposed in prior research (Hinkin, 1998).”
MethodSteel, P. (2007) · Psychological bulletin · doi:10.1037/0033-2909.133.1.65 · Cited by 3,417 · cited 2×
“Procrastination occurs when people “voluntarily delay an intended course of action despite expecting to be worse off for the delay” (Steel, 2007, p. 66).”
Background
Research works printed but not linked: 1
- Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric Theory (3rd ed.). McGraw-Hill.
Cited by
Lee, H.-P. (., Sarkar, A., Tankelevitch, L., Drosos, I., Rintel, S., Banks, R., & Wilson, N. (2025) · Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3706598.3713778 · Cited by 490
Higher education students’ perceptions of ChatGPT: A global study of early reactions
Ravšelj, D., Keržič, D., Tomaževič, N., Umek, L., Brezovar, N., A. Iahad, N., Abdulla, A. A., Akopyan, A., Aldana Segura, M. W., AlHumaid, J., et al. (2025) · PLOS ONE · doi:10.1371/journal.pone.0315011 · Cited by 376
Wang, C., Wang, H., Li, Y., Dai, J., Gu, X., & Yu, T. (2025) · International Journal of Human–Computer Interaction · doi:10.1080/10447318.2024.2383033 · Cited by 353
“Both models exemplify the transformative potential of GAI technology in education by reshaping the logic of educational resource creation and dissemination, altering teachers’ roles, and rede-fining educational objectives (Abbas et al., 2024; Chan & Hu, 2023).”
Generative AI in education and research: A systematic mapping review
Yusuf, A., Pervin, N., Román‐González, M., & Noor, N. M. (2024) · Review of Education · doi:10.1002/rev3.3489 · Cited by 182
Schei, O. M., Møgelvang, A., & Ludvigsen, K. (2024) · Education Sciences · doi:10.3390/educsci14080922 · Cited by 166
“Abbas et al. [16] found that students who experienced high academic workload and time pressure to finish their tasks reported higher use of ChatGPT.”
Influential
Metadata
- Authors
- Abbas, M., Jam, F. A., & Khan, T. I.
- Affiliations
- National University of Computer and Emerging Sciences (PK)
- University of Malaya (MY)
- University of Haripur (PK)
- Corresponding author
- Abbas, Muhammad
- Venue
- International Journal of Educational Technology in Higher Education
- Published
- 2024-02-16
- Volume
- 21
- Issue
- 1
- ISSN
- 2365-9440 (ISSN-L)
- OpenAlex
- W4392783116
- Semantic Scholar
- a00b966f5e335cb35f3e7537fad56f6b6f507478
- Cited by
- 531
- References
- 63
- Citations by year
- 2026: 275 · 2025: 179 · 2024: 77
- FWCI
- 198.76
- Type
- journal article
- Language
- en
- Open access
- diamond
- doi.org
- Licence
- cc-by
- creativecommons.org/licenses/by/4.0 (tdm)
- creativecommons.org/licenses/by/4.0 (vor)
- Locations
- International Journal of Educational Technology in Higher Education (publishedVersion, cc-by)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
- Greater South Information System
- Topics
- Social Sciences › Education › Artificial Intelligence in Education
- Social Sciences › Education › Academic integrity and plagiarism
- Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
- Keywords
- generative AI, ChatGPT, university students, academic workload, reward sensitivity, procrastination, memory loss, academic performance, higher education technology
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 (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
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Is it harmful or helpful? Examining the causes and consequences of generative AI usage among university students. https://app.cochat.ai/papers/10.1186/s41239-024-00444-7
- The paper
- Abbas, M., Jam, F. A., & Khan, T. I. (2024). Is it harmful or helpful? Examining the causes and consequences of generative AI usage among university students. International Journal of Educational Technology in Higher Education. https://doi.org/10.1186/s41239-024-00444-7
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.