AI evidence extraction

Impact of Large Language Models on Personalized Learning, Assessment Automation, and Student Outcomes in Higher Learning Institution

Onesme Niyibizi · Journal of Technology-Assisted Learning · 2026

AI-generated evidence extraction, verified across multiple analytical personas. Not a substitute for the peer-reviewed original.

7/10

Relevance

Evidence

Citations

0.00

FWCI

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.70232/jtal.v2i1.22

Methodology & findings

Study design

Quantitative survey-based study employing Multivariate Analysis of Variance (MANOVA) to examine associations between LLM use and multiple outcome variables including perceptions of personalized learning, academic performance, online engagement, satisfaction with assessment feedback, and motivation for lifelong learning..

Sample

N = 686, 2 groups

Primary method

Multivariate Analysis of Variance (MANOVA) to examine relationships between LLM use and multiple dependent variables including personalized learning effectiveness, academic performance improvement, online engagement, satisfaction with assessment feedback, and motivation for lifelong learning

Main result

The study found that "Students reported that LLMs enhance personalized learning by providing adaptive guidance, improving academic performance through instant clarification and practice support, and increasing online engagement by offering interactive and accessible learning assistance." Additionally, "The results further showed that LLMs contribute to greater satisfaction with feedback mechanisms and stimulate motivation for continuous and self-directed learning." However, lecturers expressed "significant apprehension regarding students' overreliance on LLMs, the risks associated with inaccurate or biased outputs, and the potential erosion of academic integrity."

Reports effect sizes.

Research paradigm

Positivist/quantitative empiricism

Author conclusions

The authors conclude that "while LLMs hold transformative potential for improving learning experiences, their integration must be supported by robust institutional policies, targeted capacity-building initiatives, and ongoing research. Such measures are essential to promote equitable, ethical, and effective adoption of LLMs in higher education."

Risk of bias

Selection bias: Participants from single private institution in Rwanda may not represent broader higher education populations; Perception-based measures: Study relies on student and lecturer perceptions rather than objective outcome measures; Potential response bias: Self-reported benefits of LLMs may reflect social desirability or technology adoption enthusiasm; Lecturer concerns may reflect adoption resistance rather than evidence-based risks; Self-report bias: Student perceptions of LLM benefits may be inflated due to novelty effect or social desirability bias; Asymmetric sample composition: 658 students vs 28 lecturers creates unequal group sizes for comparison; Lack of control group: No comparison between LLM users and non-users or before-after measurement; Confounding variables: No mention of controlling for prior technology experience, digital literacy, or disciplinary differences; Temporal confound: Cross-sectional design during single academic year (2024–2025) prevents causal inference; Self-report bias: student and lecturer perceptions measured through survey; Possible response bias: lecturers may over-report concerns about ethical issues

Open questions raised

The authors identify the need for ongoing research, robust institutional policies, and targeted capacity-building initiatives to address concerns about data privacy, ethical use, algorithmic bias, academic integrity protection, and student overreliance on LLMs.

Data: not_statedCode: not_statedExtracted from: pdf

Explore related topics

Productivity Effects Doctoral Training Empirical Study