NON-TECHNICAL DEBT IN AI-ENABLED SOFTWARE SYSTEMS: A PROCESS-CENTRIC MAPPING TO LIFECYCLE STANDARDS

Date

2026-6-15

Author

Kapiyasheva, Diana

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

69
views

0
downloads

Cite This

Non-Technical Debt (NTD), arising from organizational, cultural, and people-related inefficiencies, significantly compromises the quality and sustainability of artificial intelligence (AI)-enabled software systems. While international standards address technical requirements, organizations frequently struggle to operationalize these frameworks to mitigate NTD during development. This thesis addresses this gap by mapping real-world NTD instances to established lifecycle processes, providing a structured foundation for process-level diagnosis and intervention. Employing a Design Science Research (DSR) methodology, a conceptual artifact was developed by analyzing 107 validated NTD instances originating from 18 unique industrial machine learning (ML) projects. These instances span six categories, including People, Team, Requirement, Resource Management, Project Management, and Privacy and Compliance Debt, across ML domains such as Computer Vision, Natural Language Processing, and Time Series Analysis. The resulting artifact aligns mitigating processes primarily with ISO/IEC 5338:2023. The People Capability Maturity Model (P-CMM) serves as a complementary reference for socio-technical aspects extending beyond the ISO standard’s scope, specifically within Training and Development, Communication and Coordination, and Participatory Culture process areas. The artifact was rigorously validated using a hybrid approach combining Large Language Model (LLM)-assisted verification with independent expert review. Quantitative analyses, including frequency distribution, co-occurrence analysis, association rule mining, and Jaccard similarity networks, reveal that NTD in AI projects is frequently associated with systemic deficiencies in Project Planning and Human Resource Management processes. This work provides project stakeholders with a standards-aligned perspective on where NTD manifests in AI-enabled systems, highlighting the lifecycle processes that systematically contribute to NTD.

Subject Keywords

Artificial Intelligence, Non-Technical Debt, Software Process, ISO/IEC 5338, Process Mapping

URI

https://hdl.handle.net/11511/119696

Collections

Graduate School of Informatics, Thesis