Software Development Productivity with GenAI: Perceptions, Measurements, and Their Alignment
Year: 2026
Type: article
Study design: Systematic Review
Abstract: Introduction: The rapid advancement of Generative AI (GenAI) has changed the characteristics of software development. The continuous increase in software demand forces companies to focus on improving productivity on individual, team, and organizational levels to accelerate development times. Although there are many productivity metrics, there is yet no perfect method for accurately assessing productivity. Therefore, a better understanding of software development productivity is needed to enable improved measurement techniques. To address the research gap, this thesis analyzes the different perspectives of software development productivity in GenAI contexts --- perceptions and measurements --- examining their alignment. In this thesis, alignment refers to situations in which perceived productivity gains or losses are supported by objective empirical measurements. Methods: A systematic literature review is conducted to collect data on productivity perceptions and metrics. The data is then synthesized to analyze the level of alignment between the two perspectives. Results: Among individual developers, perceived productivity impacts associated with GenAI are predominantly positive, with efficiency gains being the most reported benefit. However, drawbacks are also reported. Perceptions on the team and organizational level are less common and show a more mixed distribution of positive and negative impacts. Team level perceptions are all related to collaboration and coordination, whereas organizational level perceptions are more diverse, including aspects such as code quality and increased process overhead. Productivity measurements are less consistent and rely mainly on subjective methods such as surveys. Measurements are also strongly focused on individual developers, which limits examination of the alignment to the individual level only. The findings reveal gaps between how software development productivity is perceived and measured at the individual level, with alignment varying across different aspects of productivity. Discussion: Current research provides only a partial view of productivity in GenAI contexts, as it is more based on subjective experiences of individual developers than on objective evidence. This suggests a need for more balanced and robust measurement approaches that better reflect the multidimensional nature of productivity.
Source: Työväentutkimus Vuosikirja
Author: Puska, Sohvi
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7212412857