How generative artificial intelligence (AI) tools can enhance software development?
Nguyen, Tran Huy Nguyen (2026)
Kandidaatintyö
2026
School of Engineering Science, Tietotekniikka
Kaikki oikeudet pidätetään.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi-fe20260729113092
https://urn.fi/URN:NBN:fi-fe20260729113092
Tiivistelmä
Generative artificial intelligence (GenAI) tools have become a regular part of software development. However, research on their impact on developer productivity shows mixed results. Controlled experiments show speed increases of over 50%, while a randomised controlled trial with experienced developers working under real conditions found a slowdown. This thesis looks at how software engineers use GenAI tools in their daily tasks and how they view their impact on productivity.
The study used a qualitative approach. Six developers at various seniority levels from different organisations in Northern Europe were interviewed between December 2025 and July 2026. The semi-structured interviews were analysed using thematic analysis, guided by the SPACE framework for developer productivity.
The analysis produced three main themes. First, GenAI is integrated into daily work. The tools speed up routine tasks and help developers understand requirements and unfamiliar code. Second, perceived productivity gains are consistent but conditional. One participant estimated an overall saving of about a few percentages; another described much larger time savings on individual tasks. These results depended on the type of task, the developer’s experience, their familiarity with the codebase, and the rules set by their organisations. Third, the developer’s role is changing from writing code to supervising it, as some of the saved time goes into providing context and checking output.
The findings suggest that the contradictory results in previous research come from different measurement conditions rather than actual disagreements. Also, relying solely on output metrics can exaggerate what GenAI contributes to developer productivity.
The study used a qualitative approach. Six developers at various seniority levels from different organisations in Northern Europe were interviewed between December 2025 and July 2026. The semi-structured interviews were analysed using thematic analysis, guided by the SPACE framework for developer productivity.
The analysis produced three main themes. First, GenAI is integrated into daily work. The tools speed up routine tasks and help developers understand requirements and unfamiliar code. Second, perceived productivity gains are consistent but conditional. One participant estimated an overall saving of about a few percentages; another described much larger time savings on individual tasks. These results depended on the type of task, the developer’s experience, their familiarity with the codebase, and the rules set by their organisations. Third, the developer’s role is changing from writing code to supervising it, as some of the saved time goes into providing context and checking output.
The findings suggest that the contradictory results in previous research come from different measurement conditions rather than actual disagreements. Also, relying solely on output metrics can exaggerate what GenAI contributes to developer productivity.