The Effect of Generative AI Tools on Learning Behavior and Independent Thinking in Low-Cost Schools in Pakistan
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
The rapid expansion of generative artificial intelligence (AI) tools is reshaping how students access information, complete assignments, and engage with learning. In Pakistan, where many students attend low-cost private and government-supported schools, these tools are often adopted informally without structured institutional guidance. This study examines how generative AI use is associated with learning behavior and independent thinking in resource-constrained school contexts. Using a cross-sectional exploratory mixed-methods survey design, data were collected from a convenience sample of 100 students in Grades 6–12 across urban and peri-urban schools. The instrument assessed digital access, frequency and purposes of AI use, study habits, perceived reliance, and problem-solving independence. Quantitative data were analyzed using descriptive statistics, and open-ended responses were examined through inductive thematic analysis.Findings indicate that students frequently use AI as an informal academic support mechanism, particularly for explanations and step-by-step guidance. Over half of participants reported that AI helps them complete schoolwork more efficiently, and many described cognitively engaged use patterns, such as verifying answers and generating practice questions. However, episodic answer-copying and variability in perceived independence suggest emerging risks related to cognitive offloading. Uneven device ownership and internet affordability highlight ongoing equity constraints, while students reported mixed perceptions of linguistic accuracy and contextual relevance across English and Urdu use.The findings suggest that generative AI is already functioning as informal learning infrastructure in low-cost school environments. Its educational implications appear contingent on patterns of use, access distribution, and institutional guidance. The study contributes context-specificempirical evidence to emerging global debates on AI integration in education, particularly within developing-country and low-resource settings.