Ethical AI: Towards Defining a Collective Evaluation Framework

Ethical AI requires a structured, modular approach: The paper proposes using "ontological blocks"—discrete, interpretable units encoding ethical principles like fairness, accountability, and ownership—as a foundation for evaluating AI systems across sectors like healthcare, finance, and autonomous systems.

Current AI systems lack sufficient transparency and accountability: The paper identifies opaque decision-making, misleading outputs, and unfair treatment in high-stakes domains as urgent problems that demonstrate why transparent and accountable AI systems are essential, particularly regarding data ownership and privacy concerns.

A unified framework can support legal compliance and scalability: By integrating ethical principles with FAIR (Findable, Accessible, Interoperable, Reusable) standards, the proposed framework enables scalable evaluations that align with regulations like the EU AI Act, making ethical assessment more systematic and legally defensible.

Real-world application shows practical promise: The paper demonstrates the framework's utility through an AI-powered investor profiling use case, showing how it enables "dynamic, behavior-informed risk classification" and suggests the approach can work in actual business contexts.

Significant implementation challenges remain: While ontological blocks offer a promising path toward explainable AI ethics, the paper acknowledges that obstacles persist in automation and probabilistic reasoning, indicating this is an evolving solution rather than a complete answer.