Generative AI for trustworthy systems - Towards a health check model

AbstractThe adoption of generative AI in software-intensive systems is pro- ceeding rapidly across diverse industrial contexts, but the analytical instru- ments currently used to characterize that adoption — principally unidimen- sional maturity models — compress important configurational variation into a single progressive axis. Drawing on an inductive interview study of eigh- teen senior practitioners across telecommunications, automotive, defence, avi- ation, banking, energy, government, and enterprise software services contexts, this paper presents the Trustworthy Autonomy Health Check Model: a struc- tured, multidimensional instrument for characterizing how an organization establishes trust in GenAI-assisted software engineering. The model organizes eight empirically grounded dimensions into a system layer (Scope of Agent Au- thority, Assurance Mechanisms, Data Trustworthiness, Architectural Contain- ment, Traceability & Comprehensibility) and an organizational layer (Governance, Human Oversight Posture, Workforce Capability Sustainability), each expressed on a five-level ordinal scale. A cross-cutting overlay of four trust paradigms — operational, engineering, statistical, and containment-based — captureshowtrust is established, complementing the dimensions that capture whatmust be trustworthy. The model is diagnostic rather than prescriptive: it supports cross-organizational comparison, surfaces configurational trade-offs, and locates an organization in a shared space without imposing a single pro- gression path. Crucially, higher levels are not inherently better; the goal is J. Bosch Chalmers University of Technology, Sweden E-mail: the email address R. Kazman University of Hawai’i, USA H. Muccini University of L’Aquila, Italy H. H. Olsson Malm ̈ o University, Sweden

Authors: Jan Bosch, Rick Kazman, Henry Muccini, Helena Holmström Olsson

Published in: arXiv

Publication date: 2026-09-07

Read the paper: https://doi.org/10.48550/arXiv.2609.10595

The authors and publisher do not sponsor or endorse this recording.

Source license: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/).

This audio adaptation adds an introduction and omits references and other narration distractions.