A comprehensive study at a major Zimbabwean university has found that one-third of students demonstrate dependency patterns with generative AI tools, with affected students showing significant academic performance decline compared to non-dependent peers. The research represents the first large-scale investigation of AI dependency in a developing nation educational setting.
Academic Performance Impacts Documented
The study of 248 undergraduate students revealed that those exhibiting AI dependency patterns averaged 0.41 points lower GPA compared to non-dependent students—a difference researchers describe as “educationally significant” with potential long-term consequences for graduation prospects and career opportunities.
Lead researcher Zvinodashe Revesai found that 32.7% of participants demonstrated moderate to severe dependency patterns, characterized by compulsive checking behavior averaging 18.3 daily AI interactions and failed attempts to reduce usage reported by 65.8% of affected students.
Statistical analysis identified three primary pathways through which dependency undermines academic success: critical thinking skill deterioration, writing ability degradation, and reduced knowledge acquisition. The most significant pathway involved what researchers termed “critical thinking atrophy,” accounting for 34.7% of the relationship between dependency and poor academic performance.
Unique Environmental Factors in Developing Nations
The Zimbabwean university setting revealed contextual factors that distinguish AI dependency in resource-constrained environments from patterns observed in well-funded institutions. Infrastructure limitations, including frequent power outages and limited internet connectivity, created “binge usage patterns” where students engage intensively with AI during available periods.
Economic pressures also influenced dependency development, with AI tools serving as substitutes for expensive textbooks and scarce library resources rather than supplemental learning aids. Students described AI usage as “necessity rather than convenience,” complicating traditional dependency intervention approaches that focus on usage reduction.
Student Recognition of Academic Harm
Perhaps most concerning, 72.1% of dependent students recognized negative impacts on their academic skills while continuing problematic usage patterns. Faculty interviews confirmed observable decline in student capabilities, with instructors reporting that students could “recite AI-generated analysis yet are unable to generate independent arguments.”
One student described the progression: “I no longer know what I actually understand versus what AI explained to me. When I sit for exams without AI, my mind feels empty.” Faculty members noted that students increasingly submit work showing “sophisticated analysis without foundational understanding.”
Infrastructure-Driven Dependency Patterns
The study revealed how environmental constraints can accelerate dependency formation through intermittent reinforcement mechanisms. Students learn to maximize AI usage during power availability, creating intensive engagement bursts that researchers suggest may strengthen rather than moderate dependency patterns.
Institutional Response Gaps
Despite affecting one-third of students, institutional support for managing AI dependency remained inadequate. Only 23% of faculty had received training on addressing AI dependency issues, while just 7% of counseling services offered support for technology-related concerns.
The research identified policy inconsistencies across departments, with only 33% of courses containing explicit AI usage guidelines, forcing students to develop individual usage norms without institutional guidance.
Implications for Global Higher Education
The findings suggest that AI dependency represents a universal vulnerability across diverse educational contexts, with prevalence rates similar to those documented in well-resourced institutions. However, the consequences may be more severe in developing nation settings where AI tools often replace rather than supplement traditional educational resources.
Researchers emphasize the need for proactive institutional responses that acknowledge AI’s legitimate educational roles while preventing dependency development through clear policies, faculty training, and student support services.
This report is based on research published in Information & Communications Technology in Education examining generative AI dependency among Zimbabwean university students.