Three Pathways for Standardisation and Ethical Disclosure by Default under the European Union Artificial Intelligence...

Area of Science:

• AI Ethics and Governance
• Technological Standardization
• European Union Law

Background:

• The EU's Artificial Intelligence Act (AIA) aims for harmonized standards on abstract concepts like transparency, fairness, and accountability.
• Applying these concepts necessitates addressing complex normative questions within AI development.

Purpose of the Study:

• To explore three potential pathways for AI standardization under the AIA.
• To analyze the challenges and implications of each pathway for AI ethics and governance.

Main Methods:

• Analysis of the standard-setting history of a major European standard-setting organization (SSO).
• Conceptual examination of democratic legitimacy, consensus tracking, and ethical disclosure models.

Main Results:

• Pathway 1 (SSOs deciding) raises democratic legitimacy concerns due to exclusion of non-experts.
• Pathway 2 (consensus tracking) offers a false sense of safety and defers ethical problem-solving.
• Pathway 3 (ethical disclosure by default) shifts decision-making to legitimate local stakeholders.

Conclusions:

• The EU's AIA must avoid deferring AI ethics decisions.
• "Ethical disclosure by default" standards are proposed to ensure context-sensitive, legitimate AI ethics decision-making.
• This approach enhances transparency and accountability by providing necessary information to local stakeholders.