Gardado en:
Detalles Bibliográficos
Autor Principal: Mahmud, Iftikhar
Formato: Recurso digital
Idioma:inglés
Publicado: Zenodo 2026
Subjects:
Acceso en liña:https://doi.org/10.5281/zenodo.18863771
Tags: Engadir etiqueta
Sen Etiquetas, Sexa o primeiro en etiquetar este rexistro!
Table of Contents:
  • <p>This paper examines a structural gap in contemporary AI governance: the separation between <strong>human oversight and identifiable authorship of consequential institutional decisions</strong>. While many regulatory frameworks emphasize oversight mechanisms, they often fail to ensure that decisional authority remains clearly attributable to accountable human actors.</p> <p>The paper introduces the concept of <strong>High-Impact Algorithmic Systems (HIAS)</strong> to identify algorithmic systems whose outputs materially structure consequential outcomes in domains such as public administration, credit allocation, employment, and regulatory enforcement. Unlike conventional “high-risk AI” classifications, the HIAS framework focuses on <strong>institutional function rather than technological label</strong>, highlighting how consequential algorithmic authority can exist even where systems are not formally categorized as artificial intelligence.</p> <p>Through a comparative analysis of ASEAN governance instruments and the <strong>EU AI Act</strong>, the study demonstrates that many governance frameworks institutionalize oversight primarily as a risk-management mechanism while leaving the condition of identifiable authorship comparatively under-articulated.</p> <p>The paper argues that anchoring governance triggers to <strong>consequential institutional authority</strong> rather than system classification would strengthen accountability in algorithmically structured decision systems without requiring major structural redesign of existing regulatory frameworks.</p>