An Ontology for Standardising Trustworthy AI

Abstract

Worldwide, there are a multiplicity of parallel activities being undertaken in developing international standards, regulations and individual organisational policies related to AI and its trustworthiness characteristics. The current lack of mappings between these activities presents the danger of a highly fragmented global landscape emerging in AI trustworthiness. This could present society, government and industry with competing standards, regulations and organisational practices that will then serve to undermine rather than build trust in AI. This chapter presents a simple ontology that can be used for checking the consistency and overlap of concepts from different standards, regulations and policies. The concepts in this ontology are grounded in an overview of AI standardisation currently being undertaken in ISO/IEC JTC 1/SC 42 and identifies its project to define an AI management system standard (AIMS or ISO/IEC WD 42001) as the starting point for establishing conceptual mapping between different initiatives. We propose a minimal, high level ontology for the support of conceptual mapping between different documents and show in the first instance how this can help map out the overlaps and gaps between and among SC 42 standards currently under development.

ALTAI Tool for Assessing AI-Based Technologies: Lessons Learned and Recommendations from SHAPES Pilots

“…The ethics of an AI means that the AI application complies with all known ethical principles, societal values, and technical requirements. There are several parallel global efforts towards achieving international standards (e.g., ISO/IEC 42001 AI Management System) [28], regulations (e.g., AI Act) [29], and individual-organizational policies relevant to AI and its trustworthiness features [30].…”

“…All the questions can be consulted in the tool guideline: "ETHICS GUIDELINES FOR TRUSTWORTHY AI: High-Level Expert Group on Artificial Intelligence" [31] (pp. [26][27][28][29][30][31]. Each question has a glossary and examples from the Ethics Guidelines for Trustworthy AI.…”

Trustworthy AI: From Principles to Practices

“…With various ethics guidelines released globally as soft recommendations for AI development and converges to consensus principles, standards to concretize these principles in specific AI application areas serve as the further means to regulate the trustworthiness of AI products [448]. Multiple standardization activities are ongoing parallelly, including IEEE P7000 and ISO/IEC SC 42 [234], which specifically focuses on AI trustworthiness. There are multiple different aspects addressed in the current standardization.…”

“…There are multiple different aspects addressed in the current standardization. For example, at the holistic level, ISO/IEC SC 42 Working Group 3 (WG3) works on establishing trustworthiness for general ML systems through transparency, robustness, explainability, etc [234]. Besides generic AI-centered standards, certain applications of AI should also seek compliance with domain-specific standards to build trustworthy AI products, e.g., ISO 26262 for the functional safety of intelligent vehicles and ISO 14155 for clinical practice on medical devices.…”

Comparison and Analysis of 3 Key AI Documents: EU’s Proposed AI Act, Assessment List for Trustworthy AI (ALTAI), and ISO/IEC 42001 AI Management System

“…Amongst these three, we utilise ISO/IEC 42001 as the primary source of requirements given its distinct role as a certifiable standard, and compare the others with it to indicate adherence towards guidelines (ALTAI) and regulations (AI Act). More specifically, we investigate the following questions: We address the aforementioned questions by proposing a methodology to compare AI documents using an upper-level trustworthy AI ontology [3], which enables modelling and linking concepts within AI documents (see Sect. 2).…”