A Bias Network Approach to Encourage Ethical Reflection Among AI Developers
Author Information
Author(s): Gabriela Arriagada-Bruneau, Claudia López, Alexandra Davidoff
Primary Institution: Pontificia Universidad Católica de Chile
Hypothesis
Can a sociotechnical method help AI developers identify and relate biases in the AI development process?
Conclusion
The Bias Network Approach (BNA) aids AI developers in visualizing interconnected biases, promoting ethical reflection and transparency in decision-making.
Supporting Evidence
- The BNA helps visualize interconnected biases and their impacts.
- It promotes transparency in decision-making throughout AI development.
- Developers recognized the influence of professional biases on their work.
Takeaway
This study shows a new way for AI developers to think about biases in their work, helping them see how different problems are connected and how to make better decisions.
Methodology
The study used semi-structured interviews and thematic analysis to explore the experiences of AI developers applying the BNA in a healthcare project.
Potential Biases
The developers' professional biases and material limitations influenced their decision-making and the emergence of additional biases.
Limitations
The study's findings are based on a single case study, which may limit generalizability.
Participant Demographics
The participant team consisted of three members with backgrounds in applied mathematics, computer science, and medical informatics.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website