The Productivity-Quality Paradox: A Study of AI Code Generators in Modern Software Engineering
Author :
Saanvi GuptaJourna Name:
International Journal of Science, Engineering and Technology Volume:
14 issue:1 Year:Volume-14-issue-1 Views : 500
Abstract:
As of 2026, the integration of Artificial Intelligence (AI) into the Software Development Life Cycle (SDLC) has shifted from assistive autocomplete to autonomous \"Agentic AI.\" This paper investigates the \"Productivity-Quality Paradox,\" examining how the transition to AI-driven development impacts long-term system sustainability, architectural integrity, and technical debt. The study synthesizes longitudinal data from 2020–2025, available on open access databases and formed the structure of this research article. Results indicate a significant bimodal effect. While AI accelerates Minimum Viable Product (MVP) development by 40–60% and improves automated test case accuracy to nearly 98%, it has simultaneously triggered a sustainability crisis. Key metrics reveal a 4x increase in code duplication (violating DRY principles) and a doubling of code churn compared to 2021 baselines. Furthermore, a \"Verification Bottleneck\" has emerged: despite perceived speed gains, experienced developers spend 19% more time \"chaperoning\" and debugging AI-generated logic. Security remains a critical failure point, with over 51% of AI-authored code containing vulnerabilities. The research introduces the concept of \"Agentic Debt\"—the hidden cost of autonomous, repository-wide modifications without human contextual oversight. To mitigate systemic decay, the paper proposes a transition to the SPACE productivity framework and the implementation of AI-aware CI/CD pipelines. The study concludes that while AI is an unmatched force multiplier for speed, human-in-the-loop (HITL) verification remains the only safeguard against long-term technical bankruptcy.