From Expectation to Habit: Why Do Software Practitioners Adopt Fairness Toolkits?

Abstract: As the adoption of ML systems continues to grow across industries, concerns about fairness and bias in these systems have taken center stage. Fairness toolkits, designed to mitigate bias in ML models, serve as critical tools for addressing these ethical concerns. However, their adoption in the context of software development remains underexplored, especially regarding the cognitive and behavioral factors driving their usage. As a deeper understanding of these factors could be pivotal in refining tool designs and promoting broader adoption, this study investigates the factors influencing the adoption of fairness toolkits from an individual perspective. Guided by the Unified Theory of Acceptance and Use of Technology (UTAUT2), we examined the factors shaping the intention to adopt and actual use of fairness toolkits. Specifically, we employed Partial Least Squares Structural Equation Modeling (PLS-SEM) to analyze data from a survey study involving practitioners in the software industry. Our findings reveal that performance expectancy and habit are the primary drivers of fairness toolkit adoption. These insights suggest that by emphasizing the effectiveness of these tools in mitigating bias and fostering habitual use, organizations can encourage wider adoption. Practical recommendations include improving toolkit usability, integrating bias mitigation processes into routine development workflows, and providing ongoing support to ensure professionals see clear benefits from regular use.

• The paper reveals that performance expectancy and habit significantly drive fairness toolkit adoption.
• It employs UTAUT2 and PLS-SEM to analyze survey data from software practitioners on mitigating bias in ML.
• The study suggests that integrating fairness toolkits into workflows can enhance ethical AI practices in industry.

Adoption of Fairness Toolkits in Software Development: Key Findings and Implications

The paper, "From Expectation to Habit: Why Do Software Practitioners Adopt Fairness Toolkits?" by Gianmario Voria et al., addresses the critical issue of fairness in ML systems by investigating the adoption of fairness toolkits among software practitioners. This research is grounded in the context of increasing adoption of ML across industries coupled with rising ethical concerns regarding fairness and bias in these systems.

Research Framework and Methodology

To explore the factors influencing the adoption of fairness toolkits, the authors employ the Unified Theory of Acceptance and Use of Technology (UTAUT2), a well-regarded framework for studying technology acceptance. The study utilizes Partial Least Squares Structural Equation Modeling (PLS-SEM) to analyze data collected from a survey of software practitioners. The paper aims to provide insights into both the intention to adopt and the actual usage behavior of fairness toolkits.

Key Findings

The findings reveal two primary drivers of fairness toolkit adoption: performance expectancy and habit. Performance expectancy, or the belief that fairness toolkits will enhance job performance by effectively mitigating bias, emerges as a significant factor influencing the practitioners' intention to adopt these tools. Habit, the tendency to perform actions automatically over time, significantly affects both the intention to use and the actual adoption of fairness toolkits.

The study finds that other UTAUT2 constructs such as effort expectancy, facilitating conditions, and social influence do not significantly contribute to the intention or actual use of fairness toolkits.

Implications

Practical Implications:

Organizations seeking to promote broader adoption of fairness toolkits should emphasize the efficacy of these tools in performance contexts and foster habitual usage by integrating them seamlessly into existing workflows. Improving usability and providing consistent support may also enhance adoption rates.

Theoretical Implications:

The findings support the assertion that cognitive factors, along with habitual behavior, play a crucial role in the adoption of technological tools aimed at addressing ethical concerns. This highlights the importance of designing fairness toolkits that align with user expectations and can be easily integrated into routine development processes.

Future Directions

The study opens up several avenues for further research. Longitudinal studies could explore how these adoption dynamics change as technologies evolve and as awareness of AI ethics becomes more prominent. Additionally, research could investigate how cultural and organizational factors might influence the adoption process across different contexts and industries.

Conclusion

This paper provides valuable insights into the drivers behind the adoption of fairness toolkits in software development. By identifying performance expectancy and habitual use as key influencers, the research highlights important considerations for both practitioners and researchers in the field of machine learning fairness. These findings underscore the need for continued development and refinement of fairness toolkits to meet the evolving demands of the software engineering community.