Home › Social Science › Research Article

Research Article · Social Science · Vol 1, Issue 11

Generative-AI use and study habits of college students in fully remote, hybrid and fully onsite courses: a simulated comparative study

Maya Okafor (Department of Education, Example University); Daniel Reyes (Center for Learning Analytics, Example University)

Published 2026-11-16 · DOI 10.0000/aos.2026.0001 · CC BY 4.0

Abstract

Simulated cross-sectional survey of 612 undergraduates comparing weekly generative-AI use and study habits across fully remote, hybrid and fully onsite course delivery, and their relation to final grade.

generative AIremote learninghybrid learningstudy habits

Full text

Introduction

Universities moved large parts of teaching online during the pandemic and have kept many hybrid formats since [1]. At the same time, generative AI assistants became freely available to students [2]. Remote learners work with less external structure, so they may lean more on AI tools for explanation and drafting [3]. Whether that reliance helps or harms learning is an open question [4,5]. We ask three questions: how does AI use differ across remote, hybrid and onsite delivery; how do study habits differ; and how are both related to course outcomes?

Methods

Design and participants

We simulated a cross-sectional survey in one semester. Participants were 612 undergraduates taking the same three introductory courses delivered in three formats. Group sizes are given in Table 1. Students self-reported demographics, weekly hours of AI tool use, planning habit (a mean of five items on a 1–5 scale), study sessions per week and procrastination (1–5). Final course grade was taken from the registry.

Statistical analysis

Group differences were tested with one-way analysis of variance and described with eta-squared. Pairwise differences use Welch intervals. Grade was modelled by ordinary least squares with delivery mode, AI use, AI use squared and planning habit; onsite delivery was the reference. All data were generated with a fixed random seed, and the analysis script is available from the corresponding author.

Results

Participants

Table 1 summarises the three groups.

Delivery mode n AI use, h/week Planning (1–5) Sessions/week Final grade, %
Fully remote 204 4.2 (2.2) 3.37 (0.66) 5.2 (1.4) 71.1 (7.6)
Hybrid 211 2.9 (1.7) 3.56 (0.65) 6.3 (1.4) 75.2 (7.2)
Fully onsite 197 2.1 (1.3) 3.57 (0.64) 6.0 (1.5) 74.0 (7.6)

Table 1. Participants and outcomes by delivery mode. Values are mean (standard deviation).

AI use by delivery mode

Weekly AI use differed by delivery mode (F = 67.0, eta-squared = 0.18; Figure 1). Remote students used AI 2.0 h/week more than onsite students (95% CI 1.7 to 2.4).

Box plot of weekly AI use for remote, hybrid and onsite students; remote students report the highest use.
Figure 1. Weekly generative-AI use by delivery mode. Boxes show the median and interquartile range; dots are a random sample of 60 students per group.

Study habits

Planning habit differed by delivery mode (F = 6.1, eta-squared = 0.02), with remote students lowest and hybrid and onsite students similar. Hybrid students reported the most study sessions per week, and remote students reported the most procrastination (Figure 2).

Three bar charts comparing planning habit, study sessions per week and procrastination across the three delivery modes.
Figure 2. Study habits by delivery mode: planning habit, study sessions per week and procrastination. Bars show means with 95% confidence intervals.

AI use and course grade

Grade rose slightly with AI use in the raw data, and the fitted curves for the three modes were similar in shape but offset in height (Figure 3). After adjustment the association was not distinguishable from zero.

Scatter plot of final grade against weekly AI use with a fitted curve for each delivery mode.
Figure 3. Final course grade against weekly AI use. Points are individual students; lines are quadratic fits for each delivery mode.

Table 2 and Figure 4 give the adjusted model. Relative to onsite delivery, remote students scored about three points lower, and the hybrid difference was small. The coefficients for AI use and AI use squared both had confidence intervals that included zero.

Term Coefficient 95% CI
Intercept (onsite, reference) 62.71 59.21 to 66.21
Fully remote -3.02 -4.61 to -1.44
Hybrid 0.90 -0.53 to 2.33
AI use (h/week) 0.42 -0.41 to 1.25
AI use squared -0.01 -0.10 to 0.07
Planning habit (1–5) 2.94 2.06 to 3.83

Table 2. Linear model of final grade (R-squared = 0.12).

Forest plot of adjusted grade differences for remote and hybrid delivery compared with onsite delivery.
Figure 4. Adjusted difference in final grade for remote and hybrid delivery compared with onsite delivery, with 95% confidence intervals.

Discussion

In this simulated sample, remote delivery went with heavier AI use, weaker planning and more procrastination, while hybrid students studied in the most sessions. The lack of a clear adjusted association between AI use and grade is consistent with the view that the benefit of AI depends on how it is used alongside regular study rather than on how much it is used [4,5]. Because the study is cross-sectional and the data are simulated, no causal claim is made. A real study would need randomisation or longitudinal follow-up, validated habit scales and pre-registered analyses [6].

Limitations

The data are simulated, so the effects reflect the assumptions used to generate them. Self-reported AI use is subject to recall error, and students were not randomised to delivery mode.

Data availability

All data are simulated. The generating script and a CSV of the simulated dataset are available from the corresponding author on request.

Ethics statement

No human participants were involved. All records are synthetic.

References

  1. Bozkurt A, Sharma RC. Emergency remote teaching in a time of global crisis due to coronavirus pandemic. Asian Journal of Distance Education. 2020;15(1):i-vi.
  2. Kasneci E, Sessler K, Kuchemann S, et al. ChatGPT for good? On opportunities and challenges of large language models for education. Learning and Individual Differences. 2023;103:102274. doi:10.1016/j.lindif.2023.102274
  3. Broadbent J, Poon WL. Self-regulated learning strategies and academic achievement in online higher education learning environments: a systematic review. The Internet and Higher Education. 2015;27:1-13. doi:10.1016/j.iheduc.2015.04.007
  4. Zawacki-Richter O, Marin VI, Bond M, Gouverneur F. Systematic review of research on artificial intelligence applications in higher education. International Journal of Educational Technology in Higher Education. 2019;16:39. doi:10.1186/s41239-019-0171-0
  5. Means B, Toyama Y, Murphy R, Bakia M, Jones K. Evaluation of evidence-based practices in online learning: a meta-analysis and review of online learning studies. US Department of Education; 2010.
  6. Nosek BA, Ebersole CR, DeHaven AC, Mellor DT. The preregistration revolution. Proceedings of the National Academy of Sciences. 2018;115(11):2600-2606. doi:10.1073/pnas.1708274114

Download references

Data & code

All data are simulated; the generating script is available on request.

AI-use disclosure

None

Competing interests and funding

None Funding: None

Open peer review

Editor decision: accept (2026-11-16)

Accepted with thanks.

Reviewer Rex Reviewer — accept (round 1)

Careful, reproducible study. Careful, reproducible study. Careful, reproducible study. Careful, reproducible study.

Competing interests: None

Reviewer Rae Reviewer — accept (round 1)

Careful, reproducible study. Careful, reproducible study. Careful, reproducible study. Careful, reproducible study.

Competing interests: None

Metrics & integrity

Views
12
Downloads
5
Integrity audit
Passed integrity audit (pass), signed off by a human integrity editor on 2026-11-16
Reviews published
2

View and download counts are simple daily totals. We do not store IP addresses or track readers.

Cite & share

APA

Okafor, M., & Reyes, D. (2026). Generative-AI use and study habits of college students in fully remote, hybrid and fully onsite courses: a simulated comparative study. Annals of Open Science, 1(11). https://doi.org/10.0000/aos.2026.0001

BibTeX

@article{aos20261, title = {Generative-AI use and study habits of college students in fully remote, hybrid and fully onsite courses: a simulated comparative study}, author = {Maya Okafor and Daniel Reyes}, journal = {Annals of Open Science}, year = {2026}, volume = {1}, number = {11}, doi = {10.0000/aos.2026.0001}, url = {https://doi.org/10.0000/aos.2026.0001} }

Download .bib Download .ris

Share: Email · DOI link · JATS XML · Crossref XML