Abstract

Artificial intelligence (AI) agents are increasingly integrated into software engineering (SE), performing coding, testing, and debugging tasks across the software development lifecycle (SDLC). While these tools improve development speed and automation, they also introduce uncertainty regarding software quality, developer dependence, and responsibility for failures. This study examines the value of AI agents through key socio-technical tensions, including productivity versus technical debt, controlled versus real-world performance, developer skill transformation, and accountability in AI-driven development. Drawing on a socio-technical perspective, it argues that AI agent value is not solely technical but organizational and governance-related, requiring new ways to understand responsibility and sustainable software development practices.

Download

Share

COinS