Paper Number
1736
Paper Type
Complete Research Paper
Abstract
Artificial Intelligence (AI) has been transformative in recent decades, but its black box nature and the lack of suitable documentation guidelines make regulatory compliance difficult, thereby hindering broader adoption in practice. This paper addresses the need for effective AI documentation in organizations for auditing, transparency, and regulation adherence. Based on design science research a documentation guideline for AI applications is designed and evaluated in two use cases, involving seven expert practitioners. The resulting guideline, validated by an AI auditor, provides a comprehensive structure for documenting AI systems, emphasizing legal compliance and responsible implementation. The guideline covers AI's development process, AI’s integration into business and its governance, facilitating risk detection and management of AI. The paper also introduces four design principles for AI documentation: adaptability to context, re-use of existing evidence, clarity in documentation processes, and linkage of documentation to system evidence. This research offers significant contributions to academia and practice.
Recommended Citation
Koenigstorfer, Florian; Krieger, Felix Friedrich Anton; and Thalmann, Stefan, "Bridging AI Development and Compliance: Design Principles for the Documentation of AI" (2024). ECIS 2024 Proceedings. 13.
https://aisel.aisnet.org/ecis2024/track04_impactai/track04_impactai/13
Bridging AI Development and Compliance: Design Principles for the Documentation of AI
Artificial Intelligence (AI) has been transformative in recent decades, but its black box nature and the lack of suitable documentation guidelines make regulatory compliance difficult, thereby hindering broader adoption in practice. This paper addresses the need for effective AI documentation in organizations for auditing, transparency, and regulation adherence. Based on design science research a documentation guideline for AI applications is designed and evaluated in two use cases, involving seven expert practitioners. The resulting guideline, validated by an AI auditor, provides a comprehensive structure for documenting AI systems, emphasizing legal compliance and responsible implementation. The guideline covers AI's development process, AI’s integration into business and its governance, facilitating risk detection and management of AI. The paper also introduces four design principles for AI documentation: adaptability to context, re-use of existing evidence, clarity in documentation processes, and linkage of documentation to system evidence. This research offers significant contributions to academia and practice.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.