<![CDATA[AI-assisted development tools are widely adopted in software engineering (SE), yet their effects on team dynamics and software delivery outcomes remain poorly understood in sub-Saharan African settings. This paper investigates how AI tool integration influences team roles, collaboration, skill requirements, and software delivery outcomes among South African software development professionals. A mixed-methods design was used, combining a structured survey with thematic analysis of open-ended responses from 40 participants across developer, tester, DevOps, and team lead roles. Multiple linear regression and Spearman's rank correlation were applied to quantitative data; thematic analysis followed the six-phase approach of Braun and Clarke. Findings show that GitHub Copilot was used by 75% of respondents. Interpersonal trust was the strongest predictor of development speed (β = 0.485, p = 0.017), exceeding all AI-specific variables in the model. AI use at the adoption onset reduced development speed; frequency of use increased it. Role transformation was reported by 95% of respondents and predicted team productivity. However, causal inference is not warranted given the cross-sectional design and reliance on self-reported measures. The findings are further constrained by a purposive sample of 40 drawn from networked professional communities, which limits statistical power and generalisability. To the authors’ knowledge, no prior published study has examined AI adoption and team dynamics within a South African SE population, using this combination of methods, though a systematic literature review of that literature was beyond the scope of this study.]]>
Copyrights © 2026
