Integrating GenAI into Software Work: Product and Process Quality Impact

Context: Generative Artificial Intelligence (GenAI), particularly through Large Language Model (LLM)-based tools, has rapidly emerged as a transformative force in software engineering, supporting activities such as code generation, refactoring, testing, and documentation. Despite their growing adoption, empirical evidence remains limited regarding how LLM-based tools influence established concepts of software product and process quality. Goal: This study investigates how Brazilian software practitioners perceive the effects of integrating LLM-based GenAI tools on software quality, especially human-centered factors, and development processes across the software lifecycle. We aim to provide a holistic understanding of both the opportunities and risks associated with their adoption in professional environments. Method: We conducted a survey with 121 software practitioners from 26 Brazilian states, spanning public and private sectors. The questionnaire comprised 26 closed and open-ended questions. Quantitative data were analyzed using descriptive statistics, while qualitative answers underwent content analysis applying open and axial coding principles. Results: Practitioners reported positive perceptions of LLM-based GenAI tools, emphasizing productivity gains, faster bug detection, and automation of repetitive tasks}. Open-ended responses revealed improvements in collaboration, communication, and innovation, particularly through rapid prototyping and creative ideation, alongside risks such as overreliance, diminished peer interaction, and a lack of long-term evidence on software quality. Conclusions: GenAI tools are perceived as important complements to software development, improving productivity, code quality, and team collaboration when used under human-in-the-loop oversight. Their integration demands governance mechanisms that ensure reliability, accountability, and sustainable quality improvements.