[Paper Review] Connecting the Dots in Trustworthy Artificial Intelligence: From AI Principles, Ethics, and Key Requirements to Responsible AI Systems and Regulation

Natalia Díaz-Rodríguez, Javier Del Ser+4 more|arXiv (Cornell University)|May 2, 2023

Ethics and Social Impacts of AI34 citations

TL;DR

The paper presents a holistic framework linking ethical principles, philosophical ethics, risk-based regulation, and technical requirements to define responsible AI systems, facilitated by auditing and regulatory sandboxes.

ABSTRACT

Trustworthy Artificial Intelligence (AI) is based on seven technical requirements sustained over three main pillars that should be met throughout the system's entire life cycle: it should be (1) lawful, (2) ethical, and (3) robust, both from a technical and a social perspective. However, attaining truly trustworthy AI concerns a wider vision that comprises the trustworthiness of all processes and actors that are part of the system's life cycle, and considers previous aspects from different lenses. A more holistic vision contemplates four essential axes: the global principles for ethical use and development of AI-based systems, a philosophical take on AI ethics, a risk-based approach to AI regulation, and the mentioned pillars and requirements. The seven requirements (human agency and oversight; robustness and safety; privacy and data governance; transparency; diversity, non-discrimination and fairness; societal and environmental wellbeing; and accountability) are analyzed from a triple perspective: What each requirement for trustworthy AI is, Why it is needed, and How each requirement can be implemented in practice. On the other hand, a practical approach to implement trustworthy AI systems allows defining the concept of responsibility of AI-based systems facing the law, through a given auditing process. Therefore, a responsible AI system is the resulting notion we introduce in this work, and a concept of utmost necessity that can be realized through auditing processes, subject to the challenges posed by the use of regulatory sandboxes. Our multidisciplinary vision of trustworthy AI culminates in a debate on the diverging views published lately about the future of AI. Our reflections in this matter conclude that regulation is a key for reaching a consensus among these views, and that trustworthy and responsible AI systems will be crucial for the present and future of our society.

Motivation & Objective

• Synthesize a multi-axis view of trustworthy AI spanning principles, ethics, regulation, and technical requirements.
• Translate AI principles into practical requirements and auditing processes for responsible AI systems.
• Analyze the EU AI Act risk-based regulation as a pragmatic governance model for trustworthy AI.
• Propose responsible AI systems as an outcome that integrates regulation with technical and organizational practices.

Proposed method

• Analyze UNESCO, Telefónica, and EU ethical principles to distill core trustworthy AI concepts.
• Decompose trustworthy AI into four axes: ethical principles, philosophical ethics, risk-based regulation, and technical requirements.
• Examine the seven technical and governance requirements (human agency & oversight; robustness & safety; privacy & data governance; transparency; diversity & fairness; societal & environmental wellbeing; accountability) through What, Why, and How lenses.
• Discuss regulatory sandboxes and auditing as mechanisms to operationalize responsibility in AI systems.
• Provide a case-study-oriented view for AI healthcare to illustrate auditing and compliance.
• Discuss future directions including general-purpose AI and dynamic regulation.

Experimental results

Research questions

• RQ1 How can trustworthy AI be operationalized across ethical principles, philosophy, regulation, and technical requirements?
• RQ2 What are the concrete implementations and auditing approaches for the seven trustworthiness requirements?
• RQ3 How does a risk-based EU AI Act framework translate into practical obligations for high-risk AI systems?
• RQ4 What role do regulatory sandboxes play in achieving responsible AI systems?
• RQ5 What debates and regulatory gaps exist regarding emerging AI systems such as GPAIS and neuroscience technologies?

Key findings

• Trustworthy AI rests on four axes that connect principles, ethics, regulation, and technology.
• Seven trustworthiness requirements can be analyzed by what they are, why they are needed, and how to implement them in practice.
• Regulatory sandboxes and auditing are proposed as practical tools to realize responsible AI systems.
• The EU AI Act adopts a risk-based approach with four risk levels and specific obligations for high-risk systems.
• High-risk AI categories include areas such as law enforcement, education access, hiring, and critical infrastructure, informing conformity assessments and governance.