Working paper number: 478
Author(s): Derrick Sekgala, Zea Leon, Jane Kelly, Rebecca Maughan-Brown, Elona Toska
Unit: CSSR

Abstract: 

This working paper examines the artificial intelligence (AI) tool usage patterns, barriers, and training needs of 25 members of the Centre for Social Science Research (CSSR) at the University of Cape Town through an internal survey conducted in June to July 2025.

The study reveals widespread AI adoption, with 96% of respondents currently using AI tools in their academic work, which significantly exceeds patterns observed in broader student populations. ChatGPT dominates usage (92%), followed by Grammarly (48%) and Elicit (40%). Most respondents employ multiple tools simultaneously with sophisticated quality assurance strategies: 70.8% cross-check outputs with other sources, 66.7% use iterative prompting, and 37.5% combine outputs from different tools.

Despite high adoption rates, several barriers were reported. Accuracy concerns affect 68% of users, ethical considerations worry 64%, and lack of formal training is a major obstacle for 56%. Critically, 80% of respondents have received no formal AI training, yet 72% express strong interest in structured capacity building, particularly hands-on workshops (73%) and ethics guidelines (69.6%).

The findings expose a disconnect between widespread usage and self-assessed ‘moderate’ understanding; this disconnect suggests that adoption of AI is driven by practical necessity instead of formalized knowledge. Qualitative insights reveal users envision AI as ‘a partner in social science research’ instead of replacement technology, and anticipate gains in efficiency and creativity while maintaining concerns about potential impacts on critical thinking.

The working paper argues that CSSR represents a community highly engaged yet cautiously navigating AI integration. To transition from informal experimentation to responsible implementation, five evidence-based recommendations are proposed: developing ethical guidelines, implementing tiered training programmes, establishing peer learning networks, forming an AI working group, and leveraging institutional resources. These interventions aim to maintain academic integrity while maximizing AI's transformative potential in social science research in South African higher education contexts.