What is it about?
Fairness matters more and more as machine learning and AI systems, such as generative AI, large language models, and agents, are used to make decisions. Fairness toolkits help practitioners find and reduce bias, but they are still used by few software practitioners. Since fairness is socially constructed, the cultural values of each person may change how they see and use these tools. In this study, we asked whether individual cultural values relate to practitioners' intention to adopt fairness toolkits and to their actual use. We combined a well-known model of technology acceptance (UTAUT2) with Hofstede's cultural dimensions, surveyed 181 software professionals, and analyzed the data with a statistical technique (PLS-SEM).
Featured Image
Photo by Tingey Injury Law Firm on Unsplash
Why is it important?
We found that three cultural values—Power Distance, Collectivism, and Long-Term Orientation—both directly affect adoption and change some key relationships in the adoption process. For example, people with a more collectivist orientation were more likely to turn their intention to use fairness tools into actual use, which points to the role of shared team goals. This means that promoting fairness tools is not only about the tools themselves: strategies that take cultural values into account can help organizations increase their adoption.
Perspectives
This paper builds on our ICSE-SEIS 2025 study “From Expectation to Habit”, where we looked at why practitioners adopt fairness toolkits. Here we added the lens I studied during my PhD: individual cultural values. Unlike in our study on LLM adoption, where culture did not change much, here cultural values played a clear role. I find this contrast interesting: fairness is a socially constructed concept, and the values of each person seem to matter more when tools touch on ethics.
Prof. Stefano Lambiase
Aalborg Universitet
Read the Original
This page is a summary of: From values to adoption: on the role of individual cultural values on fairness toolkit adoption in software development, Empirical Software Engineering, July 2026, Springer Science + Business Media,
DOI: 10.1007/s10664-026-10902-7.
You can read the full text:
Contributors
The following have contributed to this page
