Abstract
Models used in geospatial data science are often built and optimized for a specific local context, such as a particular location at a point in time. However, upon publication, these models may be generalized beyond this context, reused in research simulating or predicting other times and places. Without sufficient information or documentation, bias embedded in these models can in turn result in bias in the reuser’s research outputs. Drawing on a long-term qualitative case study of aging dams researchers and developers of models used by these researchers, we find significant documentation gaps. We combine a literature-based genealogy with interviews with model developers and reusers to how assumptions rooted in localized case studies are obscured as models are adapted, coupled, and generalized, producing accountability and transparency tradeoffs and increasing the risk of harm to marginalized communities. These steps can help platforms like I-GUIDE enable reuse while mitigating hidden bias.
Keywords
Responsible AI, AI bias, AI accountability, AI transparency, AI explainability, data ethics, AI ethics
Document Type
Paper
Start Date
15-10-2024 10:30 AM
End Date
15-10-2024 11:05 AM
DOI
10.5703/1288284318498
Recommended Citation
Abrams, Kyra M. and Darch, Peter T., "Tradeoffs of Generalization" (2024). I-GUIDE Forum. 12.
https://docs.lib.purdue.edu/iguide/2024/presentations/12
Included in
Applied Ethics Commons, Artificial Intelligence and Robotics Commons, Data Science Commons, Geographic Information Sciences Commons, Library and Information Science Commons
Tradeoffs of Generalization
Models used in geospatial data science are often built and optimized for a specific local context, such as a particular location at a point in time. However, upon publication, these models may be generalized beyond this context, reused in research simulating or predicting other times and places. Without sufficient information or documentation, bias embedded in these models can in turn result in bias in the reuser’s research outputs. Drawing on a long-term qualitative case study of aging dams researchers and developers of models used by these researchers, we find significant documentation gaps. We combine a literature-based genealogy with interviews with model developers and reusers to how assumptions rooted in localized case studies are obscured as models are adapted, coupled, and generalized, producing accountability and transparency tradeoffs and increasing the risk of harm to marginalized communities. These steps can help platforms like I-GUIDE enable reuse while mitigating hidden bias.