Does Generative AI Reduce Mental Effort at The Expense of Deep Thinking? an Experimental Investigation of Cognitive Load and Knowledge Retention

Authors

  • Zahid Abbas Lecturer, Department of Business Administration, Air University, Multan Campus, Pakistan Author
  • Laiba Rehman PhD Scholar, Department of Commerce, Bahauddin Zakariya University, Multan Author

DOI:

https://doi.org/10.71317/kjard.2.11.2026.1054

Keywords:

Generative Artificial Intelligence; Cognitive Load; Deep Thinking; Knowledge Retention; Cognitive Offloading; Experimental Study; Higher Education.

Abstract

The rapid integration of Generative Artificial Intelligence (GenAI) into higher education has transformed learning practices, raising critical concerns regarding its influence on cognitive engagement, analytical reasoning, and knowledge retention. This study experimentally examines whether GenAI-assisted learning reduces cognitive load at the expense of deep thinking and sustained knowledge acquisition. A randomized controlled experimental design was employed involving 120 undergraduate students, randomly assigned to an experimental group utilizing ChatGPT (n = 60) and a control group completing equivalent learning activities without AI assistance (n = 60). Participants undertook analytical reading, conceptual problem-solving, and critical reasoning tasks under standardized experimental conditions. Cognitive load was measured using a mental effort rating scale, deep thinking was assessed through rubric-based analytical reasoning evaluations, and knowledge retention was examined using immediate and delayed post-tests administered two weeks apart. Results revealed significantly lower perceived cognitive load among AI-assisted learners (M = 3.42, SD = 1.11) compared with the control group (M = 5.76, SD = 1.24), t(118) = 10.88, p < .001. Conversely, students learning without AI demonstrated superior deep-thinking performance (M = 78.40, SD = 9.65) relative to AI-assisted participants (M = 70.25, SD = 10.82), p < .001. Immediate knowledge retention scores showed comparatively modest differences between the experimental and control groups (82.30% versus 80.10%, respectively). However, delayed retention assessments revealed a substantial performance advantage for the control group (74.60%) over AI-assisted learners (65.20%), indicating greater knowledge deterioration following AI-supported learning. These findings suggest that although GenAI effectively reduces perceived mental effort during academic tasks, such efficiency gains may coincide with diminished analytical engagement and weaker long-term knowledge retention. The study extends cognitive load theory and cognitive offloading perspectives by identifying a potential trade-off between immediate learning efficiency and sustained cognitive development. The findings underscore the importance of pedagogically structured GenAI integration that encourages independent reasoning, active knowledge construction, and meaningful cognitive engagement rather than passive dependence on AI-generated responses. Implications for instructional design, assessment practices, and responsible AI adoption in higher education are discussed.

Downloads

Published

2026-11-16

How to Cite

Zahid Abbas, & Laiba Rehman. (2026). Does Generative AI Reduce Mental Effort at The Expense of Deep Thinking? an Experimental Investigation of Cognitive Load and Knowledge Retention. Kashmir Journal of Academic Research and Development, 2(11), 125-151. https://doi.org/10.71317/kjard.2.11.2026.1054