Authors:
Antonello Calabrò
1
;
Said Daoudagh
1
;
Eda Marchetti
1
;
Oum-El-kheir Aktouf
2
and
Annabelle Mercier
2
Affiliations:
1
Institute of Information Science and Technologies “A. Faedo”, National Research Council of Italy (CNR), Pisa, Italy
;
2
Grenoble INP, LCIS - Université Grenoble Alpes, Valence, France
Keyword(s):
AI, Agile, Cybersecurity, DevOps, Holistic, Human-Centric, Lifecycle, Privacy, Trustworthiness.
Abstract:
Ai’s potential economic growth necessitates ethical and socially responsible AI systems. Increasing human awareness and the adoption of human-centric solutions that incorporate, combine, and assure by design the most critical properties (such as security, safety, trust, transparency, and privacy) during the development will be a challenge to mitigate and effectively prevent issues in the era of AI. In that view, this paper proposes a human-centric Dev-X-Ops process (DXO4AI) for trustworthiness in AI-based systems. DXO4AI leverages existing solutions, focusing on the AI development lifecycle with a by-design solution for multiple desired properties. It integrates multidisciplinary knowledge and stakeholder focus.