Paper Type

ERF

Abstract

The explainability of AI is largely approached from an algorithm-centric perspective to increase model transparency. However, the organizational dynamics in which AI systems are integrated remain underestimated, even though they are also a source of opacity. This article, therefore, repositions AI explainability as an organizational design capability rather than an essentially technical property of models. From a socio-technical perspective, using a Design Science Research approach, we have designed a visual tool to operationalize explainability in AI projects: the eXplainability Design Canvas (XDC). The XDC materializes explainability as a boundary object to facilitate cross-domain mediation. Thus, within a structured framework, the XDC allows heterogeneous groups of stakeholders to formalize the plurality of explainability requirements specific to a common organizational context.

Paper Number

1757

Comments

SIG OSRA

Share

COinS
 
Aug 15th, 12:00 AM

eXplainability Design Canvas (XDC): A Sociotechnical Design Artifact to Operationalize AI Explainability in Organizations

The explainability of AI is largely approached from an algorithm-centric perspective to increase model transparency. However, the organizational dynamics in which AI systems are integrated remain underestimated, even though they are also a source of opacity. This article, therefore, repositions AI explainability as an organizational design capability rather than an essentially technical property of models. From a socio-technical perspective, using a Design Science Research approach, we have designed a visual tool to operationalize explainability in AI projects: the eXplainability Design Canvas (XDC). The XDC materializes explainability as a boundary object to facilitate cross-domain mediation. Thus, within a structured framework, the XDC allows heterogeneous groups of stakeholders to formalize the plurality of explainability requirements specific to a common organizational context.

When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.