The Accountability Fabric: A Suite of Semantic Tools For Managing AI System Accountability and Audit

The life cycle of an AI system is a complex multi-stage undertaking that typically involves a range of human stakeholders (e.g., developers, managers, users) who can potentially be held accountable if harm is caused by the system. In this paper, we present the Accountability Fabric, a suite of semantic tools for managing the creation and audit of accountability knowledge graphs. Demo Link: https://rains-uoa.github.io/ISWC_2021_Demo/

[1]  Peter Edwards,et al.  A Semantic Framework to Support AI System Accountability and Audit , 2021, ESWC.

[2]  Wamberto Vasconcelos,et al.  Semantic Modelling of Plans and Execution Traces for Enhancing Transparency of IoT Systems , 2019, 2019 Sixth International Conference on Internet of Things: Systems, Management and Security (IOTSMS).

[3]  Inioluwa Deborah Raji,et al.  Model Cards for Model Reporting , 2018, FAT.

[4]  Timnit Gebru,et al.  Datasheets for datasets , 2018, Commun. ACM.

[5]  Paul T. Groth,et al.  The rationale of PROV , 2015, J. Web Semant..

[6]  Kush R. Varshney,et al.  Increasing Trust in AI Services through Supplier's Declarations of Conformity , 2018, IBM J. Res. Dev..