How to Design an AI Ethics Board

Summary

This research paper by Jonas Schuett tackles one of the most pressing organizational challenges in AI governance: how to actually build an ethics board that works. Rather than offering abstract principles, Schuett dives into the nitty-gritty structural decisions that determine whether your ethics board becomes a meaningful governance mechanism or an expensive checkbox exercise. The paper systematically examines the trade-offs between internal and external board configurations, explores different operational models, and provides actionable guidance for organizations looking to implement effective AI oversight.

Key structural decisions you'll need to make

Internal vs. External Board Composition

• Scope and Authority
• Accountability Mechanisms

What makes this research different

Unlike much academic work on AI ethics that stays at the philosophical level, this paper operates in the messy reality of organizational implementation. Schuett acknowledges that ethics boards must navigate competing business pressures, regulatory requirements, and stakeholder expectations while maintaining their ethical mandate.

The paper also addresses timing - recognizing that the design choices that work for a startup implementing its first AI system differ dramatically from those needed by a tech giant with hundreds of AI products in market.

Who this resource is for

• Chief Ethics Officers and Compliance Leaders who need to build governance structures from scratch or redesign existing ones that aren't working effectively.
• Executive Leadership Teams wrestling with how to provide meaningful oversight of AI initiatives without slowing down innovation or creating bureaucratic bottlenecks.
• Board Members and Investors who need to understand what effective AI governance looks like at the organizational level and how to evaluate whether companies have adequate oversight mechanisms.
• Policy Researchers and Regulators studying how organizational governance structures can complement regulatory frameworks and industry standards.

Implementation roadmap

• Assessment Phase: The paper suggests starting with a clear-eyed evaluation of your organization's current AI governance gaps, regulatory requirements, and stakeholder expectations. This assessment drives all subsequent design decisions.
• Design Phase: Work through the structural choices systematically - composition, scope, authority, and accountability mechanisms. Schuett emphasizes that these decisions are interconnected and must be made holistically rather than in isolation.
• Pilot Phase: Rather than launching a fully-formed board, the research suggests starting with a limited scope pilot that allows you to test your design assumptions and refine the model based on real-world experience.
• Evolution Phase: Plan for iteration. The paper acknowledges that effective ethics boards must evolve as AI technology, regulatory landscape, and organizational needs change.

Watch out for

The research identifies several common failure modes: ethics boards that lack clear authority and become purely performative; boards that are too removed from technical realities to provide meaningful oversight; and boards that create compliance theater without addressing underlying ethical risks. Schuett's structural approach is designed to avoid these pitfalls, but implementation still requires careful attention to organizational culture and change management.